CINXE.COM
<!doctype html><html lang="en"><head><title data-rh="true">Exploring the latest innovations in Computer Vision | by Shashank Vasisht | Building Fynd</title><meta data-rh="true" charset="utf-8"/><meta data-rh="true" name="viewport" content="width=device-width,minimum-scale=1,initial-scale=1,maximum-scale=1"/><meta data-rh="true" name="theme-color" content="#000000"/><meta data-rh="true" name="twitter:app:name:iphone" content="Medium"/><meta data-rh="true" name="twitter:app:id:iphone" content="828256236"/><meta data-rh="true" property="al:ios:app_name" content="Medium"/><meta data-rh="true" property="al:ios:app_store_id" content="828256236"/><meta data-rh="true" property="al:android:package" content="com.medium.reader"/><meta data-rh="true" property="fb:app_id" content="542599432471018"/><meta data-rh="true" property="og:site_name" content="Medium"/><meta data-rh="true" property="og:type" content="article"/><meta data-rh="true" property="article:published_time" content="2023-03-28T06:24:58.883Z"/><meta data-rh="true" name="title" content="Exploring the latest innovations in Computer Vision | by Shashank Vasisht | Building Fynd"/><meta data-rh="true" property="og:title" content="Exploring the latest innovations in Computer Vision"/><meta data-rh="true" property="al:android:url" content="medium://p/bb3fb4c41bd8"/><meta data-rh="true" property="al:ios:url" content="medium://p/bb3fb4c41bd8"/><meta data-rh="true" property="al:android:app_name" content="Medium"/><meta data-rh="true" name="description" content="The Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) is the premier conference in computer vision, graphics, image processing, and related fields. The ICVGIP 2022 took…"/><meta data-rh="true" property="og:description" content="Insights from Fynd’s visit to The Indian Conference on Computer Vision, Graphics & Image Processing 2022"/><meta data-rh="true" property="og:url" content="https://blog.gofynd.com/exploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8"/><meta data-rh="true" property="al:web:url" content="https://blog.gofynd.com/exploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8"/><meta data-rh="true" property="og:image" content="https://miro.medium.com/v2/da:true/resize:fit:1200/0*rJjyYQz-gKIc45dN"/><meta data-rh="true" property="article:author" content="https://medium.com/@shashankvasisht_8994"/><meta data-rh="true" name="author" content="Shashank Vasisht"/><meta data-rh="true" name="robots" content="index,noarchive,follow,max-image-preview:large"/><meta data-rh="true" name="referrer" content="unsafe-url"/><meta data-rh="true" property="twitter:title" content="Exploring the latest innovations in Computer Vision"/><meta data-rh="true" name="twitter:site" content="@lifeatfynd"/><meta data-rh="true" name="twitter:app:url:iphone" content="medium://p/bb3fb4c41bd8"/><meta data-rh="true" property="twitter:description" content="Insights from Fynd’s visit to The Indian Conference on Computer Vision, Graphics & Image Processing 2022"/><meta data-rh="true" name="twitter:image:src" content="https://miro.medium.com/v2/da:true/resize:fit:1200/0*rJjyYQz-gKIc45dN"/><meta data-rh="true" name="twitter:card" content="summary_large_image"/><meta data-rh="true" name="twitter:label1" content="Reading time"/><meta data-rh="true" name="twitter:data1" content="15 min read"/><link data-rh="true" rel="icon" href="https://miro.medium.com/v2/resize:fill:256:256/1*Q7qNEfm08Fj5NVUQFFIbjQ.png"/><link data-rh="true" rel="search" type="application/opensearchdescription+xml" title="Medium" href="/osd.xml"/><link data-rh="true" rel="apple-touch-icon" sizes="152x152" href="https://miro.medium.com/v2/resize:fill:304:304/10fd5c419ac61637245384e7099e131627900034828f4f386bdaa47a74eae156"/><link data-rh="true" rel="apple-touch-icon" sizes="120x120" href="https://miro.medium.com/v2/resize:fill:240:240/10fd5c419ac61637245384e7099e131627900034828f4f386bdaa47a74eae156"/><link data-rh="true" rel="apple-touch-icon" sizes="76x76" href="https://miro.medium.com/v2/resize:fill:152:152/10fd5c419ac61637245384e7099e131627900034828f4f386bdaa47a74eae156"/><link data-rh="true" rel="apple-touch-icon" sizes="60x60" href="https://miro.medium.com/v2/resize:fill:120:120/10fd5c419ac61637245384e7099e131627900034828f4f386bdaa47a74eae156"/><link data-rh="true" rel="mask-icon" href="https://miro.medium.com/v2/resize:fill:1000:1000/7*GAOKVe--MXbEJmV9230oOQ.png" color="#171717"/><link data-rh="true" id="glyph_preload_link" rel="preload" as="style" type="text/css" href="https://glyph.medium.com/css/unbound.css"/><link data-rh="true" id="glyph_link" rel="stylesheet" type="text/css" href="https://glyph.medium.com/css/unbound.css"/><link data-rh="true" rel="author" href="https://medium.com/@shashankvasisht_8994"/><link data-rh="true" rel="canonical" href="https://blog.gofynd.com/exploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8"/><link data-rh="true" rel="alternate" href="android-app://com.medium.reader/https/medium.com/p/bb3fb4c41bd8"/><script data-rh="true" type="application/ld+json">{"@context":"http:\u002F\u002Fschema.org","@type":"NewsArticle","image":["https:\u002F\u002Fmiro.medium.com\u002Fv2\u002Fda:true\u002Fresize:fit:1200\u002F0*rJjyYQz-gKIc45dN"],"url":"https:\u002F\u002Fblog.gofynd.com\u002Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8","dateCreated":"2023-03-28T06:24:58.883Z","datePublished":"2023-03-28T06:24:58.883Z","dateModified":"2023-04-13T20:05:28.715Z","headline":"Exploring the latest innovations in Computer Vision","name":"Exploring the latest innovations in Computer Vision","description":"The Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) is the premier conference in computer vision, graphics, image processing, and related fields. The ICVGIP 2022 took…","identifier":"bb3fb4c41bd8","author":{"@type":"Person","name":"Shashank Vasisht","url":"https:\u002F\u002Fblog.gofynd.com\u002F@shashankvasisht_8994"},"creator":["Shashank Vasisht"],"publisher":{"@type":"Organization","name":"Building Fynd","url":"blog.gofynd.com","logo":{"@type":"ImageObject","width":119,"height":60,"url":"https:\u002F\u002Fmiro.medium.com\u002Fv2\u002Fresize:fit:238\u002F1*qOeeVf-aNCFCtn2_qeSUFw.png"}},"mainEntityOfPage":"https:\u002F\u002Fblog.gofynd.com\u002Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8"}</script><style type="text/css" data-fela-rehydration="588" data-fela-type="STATIC">html{box-sizing:border-box;-webkit-text-size-adjust:100%}*, *:before, *:after{box-sizing:inherit}body{margin:0;padding:0;text-rendering:optimizeLegibility;-webkit-font-smoothing:antialiased;color:rgba(0,0,0,0.8);position:relative;min-height:100vh}h1, h2, h3, h4, h5, h6, dl, dd, ol, ul, menu, figure, blockquote, p, pre, form{margin:0}menu, ol, ul{padding:0;list-style:none;list-style-image:none}main{display:block}a{color:inherit;text-decoration:none}a, button, input{-webkit-tap-highlight-color:transparent}img, svg{vertical-align:middle}button{background:transparent;overflow:visible}button, input, optgroup, select, textarea{margin:0}:root{--reach-tabs:1;--reach-menu-button:1}#speechify-root{font-family:Sohne, sans-serif}div[data-popper-reference-hidden="true"]{visibility:hidden;pointer-events:none}.grecaptcha-badge{visibility:hidden} /*XCode style (c) Angel Garcia <angelgarcia.mail@gmail.com>*/.hljs {background: #fff;color: black; }/* Gray DOCTYPE selectors like WebKit */ .xml .hljs-meta {color: #c0c0c0; }.hljs-comment, .hljs-quote {color: #007400; }.hljs-tag, .hljs-attribute, .hljs-keyword, .hljs-selector-tag, .hljs-literal, .hljs-name {color: #aa0d91; }.hljs-variable, .hljs-template-variable {color: #3F6E74; }.hljs-code, .hljs-string, .hljs-meta .hljs-string {color: #c41a16; }.hljs-regexp, .hljs-link {color: #0E0EFF; }.hljs-title, .hljs-symbol, .hljs-bullet, .hljs-number {color: #1c00cf; }.hljs-section, .hljs-meta {color: #643820; }.hljs-title.class_, .hljs-class .hljs-title, .hljs-type, .hljs-built_in, .hljs-params {color: #5c2699; }.hljs-attr {color: #836C28; }.hljs-subst {color: #000; }.hljs-formula {background-color: #eee;font-style: italic; }.hljs-addition {background-color: #baeeba; }.hljs-deletion {background-color: #ffc8bd; }.hljs-selector-id, .hljs-selector-class {color: #9b703f; }.hljs-doctag, .hljs-strong {font-weight: bold; }.hljs-emphasis {font-style: italic; } </style><style type="text/css" data-fela-rehydration="588" data-fela-type="KEYFRAME">@-webkit-keyframes k1{0%{opacity:0.8}50%{opacity:0.5}100%{opacity:0.8}}@-moz-keyframes k1{0%{opacity:0.8}50%{opacity:0.5}100%{opacity:0.8}}@keyframes k1{0%{opacity:0.8}50%{opacity:0.5}100%{opacity:0.8}}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE">.a{font-family:medium-content-sans-serif-font, -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Oxygen, Ubuntu, Cantarell, "Open Sans", "Helvetica Neue", sans-serif}.b{font-weight:400}.c{background-color:rgba(255, 255, 255, 1)}.l{display:block}.m{position:sticky}.n{top:0}.o{z-index:500}.p{padding:0 24px}.q{align-items:center}.r{border-bottom:solid 1px #F2F2F2}.y{height:41px}.z{line-height:20px}.ab{display:flex}.ac{height:57px}.ae{flex:1 0 auto}.af{color:inherit}.ag{fill:inherit}.ah{font-size:inherit}.ai{border:inherit}.aj{font-family:inherit}.ak{letter-spacing:inherit}.al{font-weight:inherit}.am{padding:0}.an{margin:0}.ao{cursor:pointer}.ap:disabled{cursor:not-allowed}.aq:disabled{color:#6B6B6B}.ar:disabled{fill:#6B6B6B}.au{width:auto}.av path{fill:#242424}.aw{height:25px}.ax{margin-left:16px}.ay{border:none}.az{border-radius:20px}.ba{width:240px}.bb{background:#F9F9F9}.bc path{fill:#6B6B6B}.be{outline:none}.bf{font-family:sohne, "Helvetica Neue", Helvetica, Arial, sans-serif}.bg{font-size:14px}.bh{width:100%}.bi{padding:10px 20px 10px 0}.bj{background-color:transparent}.bk{color:#242424}.bl::placeholder{color:#6B6B6B}.bm{display:inline-block}.bn{margin-left:12px}.bo{margin-right:12px}.bp{border-radius:4px}.bq{margin-left:24px}.br{height:24px}.bx{background-color:#F9F9F9}.by{border-radius:50%}.bz{height:32px}.ca{width:32px}.cb{justify-content:center}.ch{max-width:680px}.ci{min-width:0}.cj{animation:k1 1.2s ease-in-out infinite}.ck{height:100vh}.cl{margin-bottom:16px}.cm{margin-top:48px}.cn{align-items:flex-start}.co{flex-direction:column}.cp{justify-content:space-between}.cq{margin-bottom:24px}.cw{width:80%}.cx{background-color:#F2F2F2}.dd{height:44px}.de{width:44px}.df{margin:auto 0}.dg{margin-bottom:4px}.dh{height:16px}.di{width:120px}.dj{width:80px}.dp{margin-bottom:8px}.dq{width:96%}.dr{width:98%}.ds{width:81%}.dt{margin-left:8px}.du{color:#6B6B6B}.dv{font-size:13px}.dw{height:100%}.ep{color:#FFFFFF}.eq{fill:#FFFFFF}.er{background:rgba(105, 128, 229, 1)}.es{border-color:rgba(105, 128, 229, 1)}.ew:disabled{cursor:inherit !important}.ex:disabled{opacity:0.3}.ey:disabled:hover{background:rgba(105, 128, 229, 1)}.ez:disabled:hover{border-color:rgba(105, 128, 229, 1)}.fa{border-radius:99em}.fb{border-width:1px}.fc{border-style:solid}.fd{box-sizing:border-box}.fe{text-decoration:none}.ff{text-align:center}.fi{margin-right:32px}.fj{position:relative}.fk{fill:#6B6B6B}.fn{background:transparent}.fo svg{margin-left:4px}.fp svg{fill:#6B6B6B}.fr{box-shadow:inset 0 0 0 1px rgba(0, 0, 0, 0.05)}.fs{position:absolute}.fz{margin:0 24px}.gd{background:rgba(255, 255, 255, 1)}.ge{border:1px solid #F2F2F2}.gf{box-shadow:0 1px 4px #F2F2F2}.gg{max-height:100vh}.gh{overflow-y:auto}.gi{left:0}.gj{top:calc(100vh + 100px)}.gk{bottom:calc(100vh + 100px)}.gl{width:10px}.gm{pointer-events:none}.gn{word-break:break-word}.go{word-wrap:break-word}.gp:after{display:block}.gq:after{content:""}.gr:after{clear:both}.gs{line-height:18px}.gt{letter-spacing:0.077em}.gu{font-style:normal}.ha{margin-bottom:-0.31em}.hb{text-transform:uppercase}.hc{line-height:1.23}.hd{letter-spacing:0}.he{font-weight:700}.hu{margin-top:12px}.hv{margin-bottom:-0.27em}.hw{line-height:1.394}.ir{align-items:baseline}.is{width:48px}.it{height:48px}.iu{border:2px solid rgba(255, 255, 255, 1)}.iv{z-index:0}.iw{box-shadow:none}.ix{border:1px solid rgba(0, 0, 0, 0.05)}.iy{margin-left:-12px}.iz{width:28px}.ja{height:28px}.jb{z-index:1}.jc{width:24px}.jd{margin-bottom:2px}.je{flex-wrap:nowrap}.jf{font-size:16px}.jg{line-height:24px}.ji{margin:0 8px}.jj{display:inline}.jk{color:rgba(105, 128, 229, 1)}.jl{fill:rgba(105, 128, 229, 1)}.jo{flex:0 0 auto}.jr{flex-wrap:wrap}.ju{white-space:pre-wrap}.jv{margin-right:4px}.jw{overflow:hidden}.jx{max-height:20px}.jy{text-overflow:ellipsis}.jz{display:-webkit-box}.ka{-webkit-line-clamp:1}.kb{-webkit-box-orient:vertical}.kc{word-break:break-all}.ke{padding-left:8px}.kf{padding-right:8px}.lg> *{flex-shrink:0}.lh{overflow-x:scroll}.li::-webkit-scrollbar{display:none}.lj{scrollbar-width:none}.lk{-ms-overflow-style:none}.ll{width:74px}.lm{flex-direction:row}.ln{z-index:2}.lq{-webkit-user-select:none}.lr{border:0}.ls{fill:rgba(117, 117, 117, 1)}.lv{outline:0}.lw{user-select:none}.lx> svg{pointer-events:none}.mg{cursor:progress}.mh{margin-left:4px}.mi{margin-top:0px}.mj{opacity:1}.mk{padding:4px 0}.mn{width:16px}.mp{display:inline-flex}.mv{max-width:100%}.mw{padding:8px 2px}.mx svg{color:#6B6B6B}.no{margin-left:auto}.np{margin-right:auto}.nq{max-width:1600px}.nw{clear:both}.ny{cursor:zoom-in}.nz{z-index:auto}.ob{height:auto}.oc{line-height:1.58}.od{letter-spacing:-0.004em}.oe{font-family:source-serif-pro, Georgia, Cambria, "Times New Roman", Times, serif}.ox{margin-bottom:-0.46em}.oy{line-height:1.18}.oz{letter-spacing:-0.022em}.pa{font-weight:600}.pv{text-decoration:underline}.pw{margin-top:10px}.px{max-width:728px}.qa{font-style:inherit}.qb{max-width:688px}.qc{max-width:1422px}.qd{max-width:1478px}.qe{max-width:700px}.qf{list-style-type:disc}.qg{margin-left:30px}.qh{padding-left:0px}.qn{max-width:1280px}.qo{max-width:600px}.qp{max-width:1120px}.qq{max-width:949px}.qr{max-width:1109px}.qs{max-width:1127px}.qt{max-width:981px}.qu{max-width:1059px}.qv{max-width:1142px}.qw{max-width:1128px}.qx{max-width:1011px}.qy{max-width:1152px}.qz{max-width:1257px}.ra{max-width:1162px}.rb{max-width:1024px}.rc{max-width:1344px}.rd{max-width:652px}.re{margin-bottom:26px}.rf{margin-top:6px}.rg{margin-top:8px}.rh{margin-right:8px}.ri{padding:8px 16px}.rj{border-radius:100px}.rk{transition:background 300ms ease}.rm{white-space:nowrap}.rn{border-top:none}.ro{margin-bottom:14px}.rp{height:52px}.rq{max-height:52px}.rr{box-sizing:content-box}.rs{position:static}.ru{max-width:155px}.sa{margin-right:20px}.sg{height:0px}.sh{margin-bottom:40px}.si{margin-bottom:48px}.sw{border-radius:2px}.sy{height:64px}.sz{width:64px}.ta{align-self:flex-end}.tb{color:rgba(255, 255, 255, 1)}.tc{fill:rgba(255, 255, 255, 1)}.td{background:rgba(25, 25, 25, 1)}.te{border-color:rgba(25, 25, 25, 1)}.th:disabled{opacity:0.1}.ti:disabled:hover{background:rgba(25, 25, 25, 1)}.tj:disabled:hover{border-color:rgba(25, 25, 25, 1)}.tk{flex:1 1 auto}.tq{padding-right:4px}.tr{font-weight:500}.ty{margin-top:16px}.us{gap:18px}.ut{fill:rgba(61, 61, 61, 1)}.uv{margin-top:32px}.uw{fill:#242424}.ux{background:0}.uy{border-color:#242424}.uz:disabled:hover{color:#242424}.va:disabled:hover{fill:#242424}.vb:disabled:hover{border-color:#242424}.vm{border-bottom:solid 1px #E5E5E5}.vn{margin-top:72px}.vo{padding:24px 0}.vp{margin-bottom:0px}.vq{margin-right:16px}.as:hover:not(:disabled){color:rgba(25, 25, 25, 1)}.at:hover:not(:disabled){fill:rgba(25, 25, 25, 1)}.et:hover{background:rgba(92, 110, 191, 1)}.eu:hover{border-color:rgba(92, 110, 191, 1)}.ev:hover{cursor:pointer}.fl:hover{color:#242424}.fm:hover{fill:#242424}.fq:hover svg{fill:#242424}.ft:hover{background-color:rgba(0, 0, 0, 0.1)}.jh:hover{text-decoration:underline}.jm:hover:not(:disabled){color:rgba(92, 110, 191, 1)}.jn:hover:not(:disabled){fill:rgba(92, 110, 191, 1)}.lu:hover{fill:rgba(8, 8, 8, 1)}.ml:hover{fill:#000000}.mm:hover p{color:#000000}.mo:hover{color:#000000}.my:hover svg{color:#000000}.rl:hover{background-color:#F2F2F2}.sx:hover{background-color:none}.tf:hover{background:#000000}.tg:hover{border-color:#242424}.uu:hover{fill:rgba(25, 25, 25, 1)}.bd:focus-within path{fill:#242424}.lt:focus{fill:rgba(8, 8, 8, 1)}.mz:focus svg{color:#000000}.oa:focus{transform:scale(1.01)}.ly:active{border-style:none}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (min-width: 1080px)">.d{display:none}.bw{width:64px}.cg{margin:0 64px}.cv{height:48px}.dc{margin-bottom:52px}.do{margin-bottom:48px}.ef{font-size:14px}.eg{line-height:20px}.em{font-size:13px}.eo{padding:5px 12px}.fh{display:flex}.fy{margin-bottom:68px}.gc{max-width:680px}.gz{margin-top:3.88em}.hr{font-size:42px}.hs{line-height:52px}.ht{letter-spacing:-0.011em}.ij{font-size:22px}.ik{margin-top:0.92em}.il{line-height:28px}.iq{align-items:center}.ks{border-top:solid 1px #F2F2F2}.kt{border-bottom:solid 1px #F2F2F2}.ku{margin:32px 0 0}.kv{padding:3px 8px}.le> *{margin-right:24px}.lf> :last-child{margin-right:0}.mf{margin-top:0px}.mu{margin:0}.nv{margin-top:56px}.ot{font-size:20px}.ou{margin-top:2.14em}.ov{line-height:32px}.ow{letter-spacing:-0.003em}.pn{margin-top:1.72em}.po{line-height:24px}.pp{letter-spacing:0}.pu{margin-top:0.94em}.qm{margin-top:1.14em}.rz{display:inline-block}.sf{margin-bottom:104px}.sj{flex-direction:row}.sm{margin-bottom:0}.sn{margin-right:20px}.tl{max-width:500px}.ud{margin-bottom:88px}.ug{margin-bottom:72px}.up{font-size:24px}.uq{line-height:30px}.ur{letter-spacing:-0.016em}.vg{width:min-width}.vl{padding-top:72px}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (max-width: 1079.98px)">.e{display:none}.me{margin-top:0px}.py{margin-left:auto}.pz{text-align:center}.ry{display:inline-block}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (max-width: 903.98px)">.f{display:none}.md{margin-top:0px}.rx{display:inline-block}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (max-width: 727.98px)">.g{display:none}.mb{margin-top:0px}.mc{margin-right:0px}.rw{display:inline-block}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (max-width: 551.98px)">.h{display:none}.s{display:flex}.t{justify-content:space-between}.bs{width:24px}.cc{margin:0 24px}.cr{height:40px}.cy{margin-bottom:44px}.dk{margin-bottom:32px}.dx{font-size:13px}.dy{line-height:20px}.eh{padding:0px 8px 1px}.fu{margin-bottom:4px}.gv{margin-top:2.64em}.hf{font-size:32px}.hg{line-height:38px}.hh{letter-spacing:-0.014em}.hx{font-size:18px}.hy{margin-top:0.79em}.hz{line-height:24px}.im{align-items:flex-start}.jp{flex-direction:column}.js{margin-bottom:2px}.kg{margin:24px -24px 0}.kh{padding:0}.kw> *{margin-right:8px}.kx> :last-child{margin-right:24px}.lo{margin-left:0px}.lz{margin-top:0px}.ma{margin-right:0px}.mq{margin:0}.na{border:1px solid #F2F2F2}.nb{border-radius:99em}.nc{padding:0px 16px 0px 12px}.nd{height:38px}.ne{align-items:center}.ng svg{margin-right:8px}.nr{margin-top:40px}.of{margin-top:1.56em}.og{line-height:28px}.oh{letter-spacing:-0.003em}.pb{font-size:16px}.pc{margin-top:1.23em}.pd{letter-spacing:0}.pq{margin-top:0.67em}.qi{margin-top:1.34em}.rv{display:inline-block}.sb{margin-bottom:96px}.su{margin-bottom:20px}.sv{margin-right:0}.tp{max-width:100%}.ts{font-size:24px}.tt{line-height:30px}.tu{letter-spacing:-0.016em}.tz{margin-bottom:64px}.uh{font-size:20px}.vc{width:100%}.vh{padding-top:48px}.nf:hover{border-color:#E5E5E5}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (min-width: 904px) and (max-width: 1079.98px)">.i{display:none}.bv{width:64px}.cf{margin:0 64px}.cu{height:48px}.db{margin-bottom:52px}.dn{margin-bottom:48px}.ed{font-size:14px}.ee{line-height:20px}.ek{font-size:13px}.el{padding:5px 12px}.fg{display:flex}.fx{margin-bottom:68px}.gb{max-width:680px}.gy{margin-top:3.88em}.ho{font-size:42px}.hp{line-height:52px}.hq{letter-spacing:-0.011em}.ig{font-size:22px}.ih{margin-top:0.92em}.ii{line-height:28px}.ip{align-items:center}.ko{border-top:solid 1px #F2F2F2}.kp{border-bottom:solid 1px #F2F2F2}.kq{margin:32px 0 0}.kr{padding:3px 8px}.lc> *{margin-right:24px}.ld> :last-child{margin-right:0}.mt{margin:0}.nu{margin-top:56px}.op{font-size:20px}.oq{margin-top:2.14em}.or{line-height:32px}.os{letter-spacing:-0.003em}.pk{margin-top:1.72em}.pl{line-height:24px}.pm{letter-spacing:0}.pt{margin-top:0.94em}.ql{margin-top:1.14em}.se{margin-bottom:104px}.sk{flex-direction:row}.so{margin-bottom:0}.sp{margin-right:20px}.tm{max-width:500px}.uc{margin-bottom:88px}.uf{margin-bottom:72px}.um{font-size:24px}.un{line-height:30px}.uo{letter-spacing:-0.016em}.vf{width:min-width}.vk{padding-top:72px}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (min-width: 728px) and (max-width: 903.98px)">.j{display:none}.w{display:flex}.x{justify-content:space-between}.bu{width:64px}.ce{margin:0 48px}.ct{height:48px}.da{margin-bottom:52px}.dm{margin-bottom:48px}.eb{font-size:13px}.ec{line-height:20px}.ej{padding:0px 8px 1px}.fw{margin-bottom:68px}.ga{max-width:680px}.gx{margin-top:3.88em}.hl{font-size:42px}.hm{line-height:52px}.hn{letter-spacing:-0.011em}.id{font-size:22px}.ie{margin-top:0.92em}.if{line-height:28px}.io{align-items:center}.kk{border-top:solid 1px #F2F2F2}.kl{border-bottom:solid 1px #F2F2F2}.km{margin:32px 0 0}.kn{padding:3px 8px}.la> *{margin-right:24px}.lb> :last-child{margin-right:0}.ms{margin:0}.nt{margin-top:56px}.ol{font-size:20px}.om{margin-top:2.14em}.on{line-height:32px}.oo{letter-spacing:-0.003em}.ph{margin-top:1.72em}.pi{line-height:24px}.pj{letter-spacing:0}.ps{margin-top:0.94em}.qk{margin-top:1.14em}.sd{margin-bottom:104px}.sl{flex-direction:row}.sq{margin-bottom:0}.sr{margin-right:20px}.tn{max-width:500px}.ub{margin-bottom:88px}.ue{margin-bottom:72px}.uj{font-size:24px}.uk{line-height:30px}.ul{letter-spacing:-0.016em}.ve{width:min-width}.vj{padding-top:72px}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="all and (min-width: 552px) and (max-width: 727.98px)">.k{display:none}.u{display:flex}.v{justify-content:space-between}.bt{width:24px}.cd{margin:0 24px}.cs{height:40px}.cz{margin-bottom:44px}.dl{margin-bottom:32px}.dz{font-size:13px}.ea{line-height:20px}.ei{padding:0px 8px 1px}.fv{margin-bottom:4px}.gw{margin-top:2.64em}.hi{font-size:32px}.hj{line-height:38px}.hk{letter-spacing:-0.014em}.ia{font-size:18px}.ib{margin-top:0.79em}.ic{line-height:24px}.in{align-items:flex-start}.jq{flex-direction:column}.jt{margin-bottom:2px}.ki{margin:24px 0 0}.kj{padding:0}.ky> *{margin-right:8px}.kz> :last-child{margin-right:8px}.lp{margin-left:0px}.mr{margin:0}.nh{border:1px solid #F2F2F2}.ni{border-radius:99em}.nj{padding:0px 16px 0px 12px}.nk{height:38px}.nl{align-items:center}.nn svg{margin-right:8px}.ns{margin-top:40px}.oi{margin-top:1.56em}.oj{line-height:28px}.ok{letter-spacing:-0.003em}.pe{font-size:16px}.pf{margin-top:1.23em}.pg{letter-spacing:0}.pr{margin-top:0.67em}.qj{margin-top:1.34em}.sc{margin-bottom:96px}.ss{margin-bottom:20px}.st{margin-right:0}.to{max-width:100%}.tv{font-size:24px}.tw{line-height:30px}.tx{letter-spacing:-0.016em}.ua{margin-bottom:64px}.ui{font-size:20px}.vd{width:100%}.vi{padding-top:48px}.nm:hover{border-color:#E5E5E5}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="print">.rt{display:none}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="(orientation: landscape) and (max-width: 903.98px)">.kd{max-height:none}</style><style type="text/css" data-fela-rehydration="588" data-fela-type="RULE" media="(prefers-reduced-motion: no-preference)">.nx{transition:transform 300ms cubic-bezier(0.2, 0, 0.2, 1)}</style></head><body><div id="root"><div class="a b c"><div class="d e f g h i j k"></div><script>document.domain = document.domain;</script><div class="l c"><div class="l m n o c"><div class="p q r s t u v w x i d y z"><a class="du ag dv bf ak b am an ao ap aq ar as at s u w i d q dw z" href="https://rsci.app.link/?%24canonical_url=https%3A%2F%2Fmedium.com%2Fp%2Fbb3fb4c41bd8&%7Efeature=LoOpenInAppButton&%7Echannel=ShowPostUnderCollection&source=---top_nav_layout_nav----------------------------------" rel="noopener follow">Open in app<svg xmlns="http://www.w3.org/2000/svg" width="10" height="10" fill="none" viewBox="0 0 10 10" class="dt"><path fill="currentColor" d="M.985 8.485a.375.375 0 1 0 .53.53zM8.75 1.25h.375A.375.375 0 0 0 8.75.875zM8.375 6.5a.375.375 0 1 0 .75 0zM3.5.875a.375.375 0 1 0 0 .75zm-1.985 8.14 7.5-7.5-.53-.53-7.5 7.5zm6.86-7.765V6.5h.75V1.25zM3.5 1.625h5.25v-.75H3.5z"></path></svg></a><div class="ab q"><p class="bf b dx dy dz ea eb ec ed ee ef eg du"><span><a class="bf b dx dy eh dz ea ei eb ec ej ek ee el em eg eo ep eq er es et eu ev ew ex ey ez fa fb fc fd bm fe ff" data-testid="headerSignUpButton" href="https://medium.com/m/signin?operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=post_page---top_nav_layout_nav-----------------------global_nav-----------" rel="noopener follow">Sign up</a></span></p><div class="ax l"><p class="bf b dx dy dz ea eb ec ed ee ef eg du"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerSignInButton" href="https://medium.com/m/signin?operation=login&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=post_page---top_nav_layout_nav-----------------------global_nav-----------" rel="noopener follow">Sign in</a></span></p></div></div></div><div class="p q r ab ac"><div class="ab q ae"><a class="af ag ah ai aj ak al am an ao ap aq ar as at ab" aria-label="Homepage" data-testid="headerMediumLogo" href="https://medium.com/?source=---top_nav_layout_nav----------------------------------" rel="noopener follow"><svg xmlns="http://www.w3.org/2000/svg" width="719" height="160" fill="none" viewBox="0 0 719 160" class="au av aw"><path fill="#242424" d="m174.104 9.734.215-.047V8.02H130.39L89.6 103.89 48.81 8.021H1.472v1.666l.212.047c8.018 1.81 12.09 4.509 12.09 14.242V137.93c0 9.734-4.087 12.433-12.106 14.243l-.212.047v1.671h32.118v-1.665l-.213-.048c-8.018-1.809-12.089-4.509-12.089-14.242V30.586l52.399 123.305h2.972l53.925-126.743V140.75c-.687 7.688-4.721 10.062-11.982 11.701l-.215.05v1.652h55.948v-1.652l-.215-.05c-7.269-1.639-11.4-4.013-12.087-11.701l-.037-116.774h.037c0-9.733 4.071-12.432 12.087-14.242m25.555 75.488c.915-20.474 8.268-35.252 20.606-35.507 3.806.063 6.998 1.312 9.479 3.714 5.272 5.118 7.751 15.812 7.368 31.793zm-.553 5.77h65.573v-.275c-.186-15.656-4.721-27.834-13.466-36.196-7.559-7.227-18.751-11.203-30.507-11.203h-.263c-6.101 0-13.584 1.48-18.909 4.16-6.061 2.807-11.407 7.003-15.855 12.511-7.161 8.874-11.499 20.866-12.554 34.343q-.05.606-.092 1.212a50 50 0 0 0-.065 1.151 85.807 85.807 0 0 0-.094 5.689c.71 30.524 17.198 54.917 46.483 54.917 25.705 0 40.675-18.791 44.407-44.013l-1.886-.664c-6.557 13.556-18.334 21.771-31.738 20.769-18.297-1.369-32.314-19.922-31.042-42.395m139.722 41.359c-2.151 5.101-6.639 7.908-12.653 7.908s-11.513-4.129-15.418-11.63c-4.197-8.053-6.405-19.436-6.405-32.92 0-28.067 8.729-46.22 22.24-46.22 5.657 0 10.111 2.807 12.236 7.704zm43.499 20.008c-8.019-1.897-12.089-4.722-12.089-14.951V1.309l-48.716 14.353v1.757l.299-.024c6.72-.543 11.278.386 13.925 2.83 2.072 1.915 3.082 4.853 3.082 8.987v18.66c-4.803-3.067-10.516-4.56-17.448-4.56-14.059 0-26.909 5.92-36.176 16.672-9.66 11.205-14.767 26.518-14.767 44.278-.003 31.72 15.612 53.039 38.851 53.039 13.595 0 24.533-7.449 29.54-20.013v16.865h43.711v-1.746zM424.1 19.819c0-9.904-7.468-17.374-17.375-17.374-9.859 0-17.573 7.632-17.573 17.374s7.721 17.374 17.573 17.374c9.907 0 17.375-7.47 17.375-17.374m11.499 132.546c-8.019-1.897-12.089-4.722-12.089-14.951h-.035V43.635l-43.714 12.551v1.705l.263.024c9.458.842 12.047 4.1 12.047 15.152v81.086h43.751v-1.746zm112.013 0c-8.018-1.897-12.089-4.722-12.089-14.951V43.635l-41.621 12.137v1.71l.246.026c7.733.813 9.967 4.257 9.967 15.36v59.279c-2.578 5.102-7.415 8.131-13.274 8.336-9.503 0-14.736-6.419-14.736-18.073V43.638l-43.714 12.55v1.703l.262.024c9.459.84 12.05 4.097 12.05 15.152v50.17a56.3 56.3 0 0 0 .91 10.444l.787 3.423c3.701 13.262 13.398 20.197 28.59 20.197 12.868 0 24.147-7.966 29.115-20.43v17.311h43.714v-1.747zm169.818 1.788v-1.749l-.213-.05c-8.7-2.006-12.089-5.789-12.089-13.49v-63.79c0-19.89-11.171-31.761-29.883-31.761-13.64 0-25.141 7.882-29.569 20.16-3.517-13.01-13.639-20.16-28.606-20.16-13.146 0-23.449 6.938-27.869 18.657V43.643L545.487 55.68v1.715l.263.024c9.345.829 12.047 4.181 12.047 14.95v81.784h40.787v-1.746l-.215-.053c-6.941-1.631-9.181-4.606-9.181-12.239V66.998c1.836-4.289 5.537-9.37 12.853-9.37 9.086 0 13.692 6.296 13.692 18.697v77.828h40.797v-1.746l-.215-.053c-6.94-1.631-9.18-4.606-9.18-12.239V75.066a42 42 0 0 0-.578-7.26c1.947-4.661 5.86-10.177 13.475-10.177 9.214 0 13.691 6.114 13.691 18.696v77.828z"></path></svg></a><div class="ax h"><div class="ab ay az ba bb q bc bd"><div class="bm" aria-hidden="false" aria-describedby="searchResults" aria-labelledby="searchResults"></div><div class="bn bo ab"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24"><path fill="currentColor" fill-rule="evenodd" d="M4.092 11.06a6.95 6.95 0 1 1 13.9 0 6.95 6.95 0 0 1-13.9 0m6.95-8.05a8.05 8.05 0 1 0 5.13 14.26l3.75 3.75a.56.56 0 1 0 .79-.79l-3.73-3.73A8.05 8.05 0 0 0 11.042 3z" clip-rule="evenodd"></path></svg></div><input role="combobox" aria-controls="searchResults" aria-expanded="false" aria-label="search" data-testid="headerSearchInput" tabindex="0" class="ay be bf bg z bh bi bj bk bl" placeholder="Search" value=""/></div></div></div><div class="h k w fg fh"><div class="fi ab"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerWriteButton" href="https://medium.com/m/signin?operation=register&redirect=https%3A%2F%2Fmedium.com%2Fnew-story&source=---top_nav_layout_nav-----------------------new_post_topnav-----------" rel="noopener follow"><div class="bf b bg z du fj fk ab q fl fm"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24" aria-label="Write"><path fill="currentColor" d="M14 4a.5.5 0 0 0 0-1zm7 6a.5.5 0 0 0-1 0zm-7-7H4v1h10zM3 4v16h1V4zm1 17h16v-1H4zm17-1V10h-1v10zm-1 1a1 1 0 0 0 1-1h-1zM3 20a1 1 0 0 0 1 1v-1zM4 3a1 1 0 0 0-1 1h1z"></path><path stroke="currentColor" d="m17.5 4.5-8.458 8.458a.25.25 0 0 0-.06.098l-.824 2.47a.25.25 0 0 0 .316.316l2.47-.823a.25.25 0 0 0 .098-.06L19.5 6.5m-2-2 2.323-2.323a.25.25 0 0 1 .354 0l1.646 1.646a.25.25 0 0 1 0 .354L19.5 6.5m-2-2 2 2"></path></svg><div class="dt l">Write</div></div></a></span></div></div><div class="k j i d"><div class="fi ab"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerSearchButton" href="https://medium.com/search?source=---top_nav_layout_nav----------------------------------" rel="noopener follow"><div class="bf b bg z du fj fk ab q fl fm"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24" aria-label="Search"><path fill="currentColor" fill-rule="evenodd" d="M4.092 11.06a6.95 6.95 0 1 1 13.9 0 6.95 6.95 0 0 1-13.9 0m6.95-8.05a8.05 8.05 0 1 0 5.13 14.26l3.75 3.75a.56.56 0 1 0 .79-.79l-3.73-3.73A8.05 8.05 0 0 0 11.042 3z" clip-rule="evenodd"></path></svg></div></a></div></div><div class="fi h k j"><div class="ab q"><p class="bf b dx dy dz ea eb ec ed ee ef eg du"><span><a class="bf b dx dy eh dz ea ei eb ec ej ek ee el em eg eo ep eq er es et eu ev ew ex ey ez fa fb fc fd bm fe ff" data-testid="headerSignUpButton" href="https://medium.com/m/signin?operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=post_page---top_nav_layout_nav-----------------------global_nav-----------" rel="noopener follow">Sign up</a></span></p><div class="ax l"><p class="bf b dx dy dz ea eb ec ed ee ef eg du"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerSignInButton" href="https://medium.com/m/signin?operation=login&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=post_page---top_nav_layout_nav-----------------------global_nav-----------" rel="noopener follow">Sign in</a></span></p></div></div></div><div class="l" aria-hidden="false"><button class="ay fn am ab q ao fo fp fq" aria-label="user options menu" data-testid="headerUserIcon"><div class="l fj"><img alt="" class="l fd by bz ca cx" src="https://miro.medium.com/v2/resize:fill:64:64/1*dmbNkD5D-u45r44go_cf0g.png" width="32" height="32" loading="lazy" role="presentation"/><div class="fr by l bz ca fs n ay ft"></div></div></button></div></div></div><div class="l"><div class="fu fv fw fx fy l"><div class="ab cb"><div class="ci bh fz ga gb gc"></div></div><article><div class="l"><div class="l"><span class="l"></span><section><div><div class="fs gi gj gk gl gm"></div><div class="gn go gp gq gr"><div class="ab cb"><div class="ci bh fz ga gb gc"><h2 id="2ea3" class="gs gt gu bf b dv gv gw gx gy gz ha du hb" aria-label="kicker paragraph">Machine Learning</h2><div><h1 id="64eb" class="pw-post-title hc hd gu bf he hf hg hh hi hj hk hl hm hn ho hp hq hr hs ht hu hv bk" data-testid="storyTitle"><strong class="al">Exploring the latest innovations in Computer Vision</strong></h1></div><div><h2 id="fb83" class="pw-subtitle-paragraph hw hd gu bf b hx hy hz ia ib ic id ie if ig ih ii ij ik il cq du">Insights from Fynd’s visit to The Indian Conference on Computer Vision, Graphics & Image Processing 2022</h2><div><div class="speechify-ignore ab cp"><div class="speechify-ignore bh l"><div class="im in io ip iq ab"><div><div class="ab ir"><div><div class="bm" aria-hidden="false"><a href="https://medium.com/@shashankvasisht_8994?source=post_page---byline--bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="l is it by iu iv"><div class="l fj"><img alt="Shashank Vasisht" class="l fd by dd de cx" src="https://miro.medium.com/v2/resize:fill:88:88/1*CI03_gV0KvR2a-WttD4uiw.jpeg" width="44" height="44" loading="lazy" data-testid="authorPhoto"/><div class="iw by l dd de fs n ix ft"></div></div></div></a></div></div><div class="iy ab fj"><div><div class="bm" aria-hidden="false"><a href="https://blog.gofynd.com/?source=post_page---byline--bb3fb4c41bd8--------------------------------" rel="noopener ugc nofollow"><div class="l iz ja by iu jb"><div class="l fj"><img alt="Building Fynd" class="l fd by br jc cx" src="https://miro.medium.com/v2/resize:fill:48:48/1*Q7qNEfm08Fj5NVUQFFIbjQ.png" width="24" height="24" loading="lazy" data-testid="publicationPhoto"/><div class="iw by l br jc fs n ix ft"></div></div></div></a></div></div></div></div></div><div class="bn bh l"><div class="ab"><div style="flex:1"><span class="bf b bg z bk"><div class="jd ab q"><div class="ab q je"><div class="ab q"><div><div class="bm" aria-hidden="false"><p class="bf b jf jg bk"><a class="af ag ah ai aj ak al am an ao ap aq ar jh" data-testid="authorName" href="https://medium.com/@shashankvasisht_8994?source=post_page---byline--bb3fb4c41bd8--------------------------------" rel="noopener follow">Shashank Vasisht</a></p></div></div></div><span class="ji jj" aria-hidden="true"><span class="bf b bg z du">·</span></span><p class="bf b jf jg du"><span><a class="jk jl ah ai aj ak al am an ao ap aq ar ex jm jn" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fsubscribe%2Fuser%2F6408aa1c1489&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=post_page-6408aa1c1489--byline--bb3fb4c41bd8---------------------post_header-----------" rel="noopener follow">Follow</a></span></p></div></div></span></div></div><div class="l jo"><span class="bf b bg z du"><div class="ab cn jp jq jr"><div class="js jt ab"><div class="bf b bg z du ab ju"><span class="jv l jo">Published in</span><div><div class="l" aria-hidden="false"><a class="af ag ah ai aj ak al am an ao ap aq ar jh ab q" data-testid="publicationName" href="https://blog.gofynd.com/?source=post_page---byline--bb3fb4c41bd8--------------------------------" rel="noopener ugc nofollow"><p class="bf b bg z jw jx jy jz ka kb kc kd bk">Building Fynd</p></a></div></div></div><div class="h k"><span class="ji jj" aria-hidden="true"><span class="bf b bg z du">·</span></span></div></div><span class="bf b bg z du"><div class="ab ae"><span data-testid="storyReadTime">15 min read</span><div class="ke kf l" aria-hidden="true"><span class="l" aria-hidden="true"><span class="bf b bg z du">·</span></span></div><span data-testid="storyPublishDate">Mar 28, 2023</span></div></span></div></span></div></div></div><div class="ab cp kg kh ki kj kk kl km kn ko kp kq kr ks kt ku kv"><div class="h k w fg fh q"><div class="ll l"><div class="ab q lm ln"><div class="pw-multi-vote-icon fj jv lo lp lq"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerClapButton" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fvote%2Ffynd-team%2Fbb3fb4c41bd8&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=---header_actions--bb3fb4c41bd8---------------------clap_footer-----------" rel="noopener follow"><div><div class="bm" aria-hidden="false"><div class="lr ao ls lt lu lv am lw lx ly lq"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" aria-label="clap"><path fill-rule="evenodd" d="M11.37.828 12 3.282l.63-2.454zM13.916 3.953l1.523-2.112-1.184-.39zM8.589 1.84l1.522 2.112-.337-2.501zM18.523 18.92c-.86.86-1.75 1.246-2.62 1.33a6 6 0 0 0 .407-.372c2.388-2.389 2.86-4.951 1.399-7.623l-.912-1.603-.79-1.672c-.26-.56-.194-.98.203-1.288a.7.7 0 0 1 .546-.132c.283.046.546.231.728.5l2.363 4.157c.976 1.624 1.141 4.237-1.324 6.702m-10.999-.438L3.37 14.328a.828.828 0 0 1 .585-1.408.83.83 0 0 1 .585.242l2.158 2.157a.365.365 0 0 0 .516-.516l-2.157-2.158-1.449-1.449a.826.826 0 0 1 1.167-1.17l3.438 3.44a.363.363 0 0 0 .516 0 .364.364 0 0 0 0-.516L5.293 9.513l-.97-.97a.826.826 0 0 1 0-1.166.84.84 0 0 1 1.167 0l.97.968 3.437 3.436a.36.36 0 0 0 .517 0 .366.366 0 0 0 0-.516L6.977 7.83a.82.82 0 0 1-.241-.584.82.82 0 0 1 .824-.826c.219 0 .43.087.584.242l5.787 5.787a.366.366 0 0 0 .587-.415l-1.117-2.363c-.26-.56-.194-.98.204-1.289a.7.7 0 0 1 .546-.132c.283.046.545.232.727.501l2.193 3.86c1.302 2.38.883 4.59-1.277 6.75-1.156 1.156-2.602 1.627-4.19 1.367-1.418-.236-2.866-1.033-4.079-2.246M10.75 5.971l2.12 2.12c-.41.502-.465 1.17-.128 1.89l.22.465-3.523-3.523a.8.8 0 0 1-.097-.368c0-.22.086-.428.241-.584a.847.847 0 0 1 1.167 0m7.355 1.705c-.31-.461-.746-.758-1.23-.837a1.44 1.44 0 0 0-1.11.275c-.312.24-.505.543-.59.881a1.74 1.74 0 0 0-.906-.465 1.47 1.47 0 0 0-.82.106l-2.182-2.182a1.56 1.56 0 0 0-2.2 0 1.54 1.54 0 0 0-.396.701 1.56 1.56 0 0 0-2.21-.01 1.55 1.55 0 0 0-.416.753c-.624-.624-1.649-.624-2.237-.037a1.557 1.557 0 0 0 0 2.2c-.239.1-.501.238-.715.453a1.56 1.56 0 0 0 0 2.2l.516.515a1.556 1.556 0 0 0-.753 2.615L7.01 19c1.32 1.319 2.909 2.189 4.475 2.449q.482.08.971.08c.85 0 1.653-.198 2.393-.579.231.033.46.054.686.054 1.266 0 2.457-.52 3.505-1.567 2.763-2.763 2.552-5.734 1.439-7.586z" clip-rule="evenodd"></path></svg></div></div></div></a></span></div><div class="pw-multi-vote-count l lz ma mb mc md me mf"><p class="bf b dv z du"><span class="mg">--</span></p></div></div></div><div><div class="bm" aria-hidden="false"><button class="ao lr mj mk ab q fk ml mm" aria-label="responses"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" class="mi"><path d="M18.006 16.803c1.533-1.456 2.234-3.325 2.234-5.321C20.24 7.357 16.709 4 12.191 4S4 7.357 4 11.482c0 4.126 3.674 7.482 8.191 7.482.817 0 1.622-.111 2.393-.327.231.2.48.391.744.559 1.06.693 2.203 1.044 3.399 1.044.224-.008.4-.112.486-.287a.49.49 0 0 0-.042-.518c-.495-.67-.845-1.364-1.04-2.057a4 4 0 0 1-.125-.598zm-3.122 1.055-.067-.223-.315.096a8 8 0 0 1-2.311.338c-4.023 0-7.292-2.955-7.292-6.587 0-3.633 3.269-6.588 7.292-6.588 4.014 0 7.112 2.958 7.112 6.593 0 1.794-.608 3.469-2.027 4.72l-.195.168v.255c0 .056 0 .151.016.295.025.231.081.478.154.733.154.558.398 1.117.722 1.659a5.3 5.3 0 0 1-2.165-.845c-.276-.176-.714-.383-.941-.59z"></path></svg><p class="bf b dv z du"><span class="pw-responses-count mh mi">1</span></p></button></div></div></div><div class="ab q kw kx ky kz la lb lc ld le lf lg lh li lj lk"><div class="mn k j i d"></div><div class="h k"><div><div class="bm" aria-hidden="false"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="headerBookmarkButton" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fbookmark%2Fp%2Fbb3fb4c41bd8&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=---header_actions--bb3fb4c41bd8---------------------bookmark_footer-----------" rel="noopener follow"><svg xmlns="http://www.w3.org/2000/svg" width="25" height="25" fill="none" viewBox="0 0 25 25" class="du mo" aria-label="Add to list bookmark button"><path fill="currentColor" d="M18 2.5a.5.5 0 0 1 1 0V5h2.5a.5.5 0 0 1 0 1H19v2.5a.5.5 0 1 1-1 0V6h-2.5a.5.5 0 0 1 0-1H18zM7 7a1 1 0 0 1 1-1h3.5a.5.5 0 0 0 0-1H8a2 2 0 0 0-2 2v14a.5.5 0 0 0 .805.396L12.5 17l5.695 4.396A.5.5 0 0 0 19 21v-8.5a.5.5 0 0 0-1 0v7.485l-5.195-4.012a.5.5 0 0 0-.61 0L7 19.985z"></path></svg></a></span></div></div></div><div class="fd mp cn"><div class="l ae"><div class="ab cb"><div class="mq mr ms mt mu mv ci bh"><div class="ab"><div class="bm bh" aria-hidden="false"><div><div class="bm" aria-hidden="false"><button aria-label="Listen" data-testid="audioPlayButton" class="af fk ah ai aj ak al mw an ao ap ex mx my mm mz na nb nc nd s ne nf ng nh ni nj nk u nl nm nn"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24"><path fill="currentColor" fill-rule="evenodd" d="M3 12a9 9 0 1 1 18 0 9 9 0 0 1-18 0m9-10C6.477 2 2 6.477 2 12s4.477 10 10 10 10-4.477 10-10S17.523 2 12 2m3.376 10.416-4.599 3.066a.5.5 0 0 1-.777-.416V8.934a.5.5 0 0 1 .777-.416l4.599 3.066a.5.5 0 0 1 0 .832" clip-rule="evenodd"></path></svg><div class="j i d"><p class="bf b bg z du">Listen</p></div></button></div></div></div></div></div></div></div></div><div class="bm" aria-hidden="false" aria-describedby="postFooterSocialMenu" aria-labelledby="postFooterSocialMenu"><div><div class="bm" aria-hidden="false"><button aria-controls="postFooterSocialMenu" aria-expanded="false" aria-label="Share Post" data-testid="headerSocialShareButton" class="af fk ah ai aj ak al mw an ao ap ex mx my mm mz na nb nc nd s ne nf ng nh ni nj nk u nl nm nn"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24"><path fill="currentColor" fill-rule="evenodd" d="M15.218 4.931a.4.4 0 0 1-.118.132l.012.006a.45.45 0 0 1-.292.074.5.5 0 0 1-.3-.13l-2.02-2.02v7.07c0 .28-.23.5-.5.5s-.5-.22-.5-.5v-7.04l-2 2a.45.45 0 0 1-.57.04h-.02a.4.4 0 0 1-.16-.3.4.4 0 0 1 .1-.32l2.8-2.8a.5.5 0 0 1 .7 0l2.8 2.79a.42.42 0 0 1 .068.498m-.106.138.008.004v-.01zM16 7.063h1.5a2 2 0 0 1 2 2v10a2 2 0 0 1-2 2h-11c-1.1 0-2-.9-2-2v-10a2 2 0 0 1 2-2H8a.5.5 0 0 1 .35.15.5.5 0 0 1 .15.35.5.5 0 0 1-.15.35.5.5 0 0 1-.35.15H6.4c-.5 0-.9.4-.9.9v10.2a.9.9 0 0 0 .9.9h11.2c.5 0 .9-.4.9-.9v-10.2c0-.5-.4-.9-.9-.9H16a.5.5 0 0 1 0-1" clip-rule="evenodd"></path></svg><div class="j i d"><p class="bf b bg z du">Share</p></div></button></div></div></div></div></div></div></div></div></div><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np nq"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*rJjyYQz-gKIc45dN 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*rJjyYQz-gKIc45dN 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*rJjyYQz-gKIc45dN 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*rJjyYQz-gKIc45dN 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*rJjyYQz-gKIc45dN 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*rJjyYQz-gKIc45dN 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*rJjyYQz-gKIc45dN 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*rJjyYQz-gKIc45dN 640w, https://miro.medium.com/v2/resize:fit:720/0*rJjyYQz-gKIc45dN 720w, https://miro.medium.com/v2/resize:fit:750/0*rJjyYQz-gKIc45dN 750w, https://miro.medium.com/v2/resize:fit:786/0*rJjyYQz-gKIc45dN 786w, https://miro.medium.com/v2/resize:fit:828/0*rJjyYQz-gKIc45dN 828w, https://miro.medium.com/v2/resize:fit:1100/0*rJjyYQz-gKIc45dN 1100w, https://miro.medium.com/v2/resize:fit:1400/0*rJjyYQz-gKIc45dN 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="350" loading="lazy" role="presentation"/></picture></div></div></figure><p id="a09b" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) is the premier conference in computer vision, graphics, image processing, and related fields.</p><p id="4c7b" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The ICVGIP 2022 took place at IIT Gandhinagar. The Computer Vision Research team at Fynd got a chance to attend! The 3-day event included exciting events like tutorials, paper presentations, industry sessions, plenary talks, and Vision India. Each day also featured poster presentations and demo sessions by independent researchers and industry members, offering opportunities for engaging discussions about their work.</p><h2 id="7816" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk"><strong class="al">Learnings from Tutorial Sessions</strong></h2><p id="cab1" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Two tutorial sessions were conducted in parallel on Physics-based rendering in the service of computational imaging and Designing and Optimizing Computational Imaging Systems with End-to-End Learning.</p><p id="1435" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The first was more inclined towards computer graphics and rendering while the other was about incorporating end-to-end deep learning into Imaging systems. Since the latter was closer to our field of interest, we chose to attend that.</p><h2 id="6876" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk"><strong class="al">Designing and Optimising Computational Imaging Systems with End-to-End Learning</strong></h2><p id="88c3" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">The speakers for this session were<a class="af pv" href="https://vivekboominathan.com/" rel="noopener ugc nofollow" target="_blank"> <strong class="oe he">Dr Vivek Boominathan</strong></a><strong class="oe he">,</strong><a class="af pv" href="https://www.eee.hku.hk/~evanpeng/" rel="noopener ugc nofollow" target="_blank"><strong class="oe he"> Dr Evan Y. Peng</strong></a><strong class="oe he"> </strong>and<a class="af pv" href="https://www.cs.umd.edu/~metzler/" rel="noopener ugc nofollow" target="_blank"><strong class="oe he"> Dr Chris Metzler</strong></a><strong class="oe he">. </strong>Computational imaging systems combine optics and algorithms to perform imaging and computer vision tasks more effectively than conventional imaging systems. However, end-to-end learning has emerged as a new system design paradigm where both optics and algorithms are designed automatically using training data and machine learning.</p><p id="539e" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">Advantages of end-to-end learning algorithms</strong></p><p id="51c5" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">This tutorial presents an end-to-end learning method that integrates optical models. Traditional optical lenses function by focusing light to a single point, mimicking human vision, and are commonly used in camera systems to capture visual information. However, this approach may not be optimal for all imaging tasks, such as monocular depth estimation and super-resolution. By using modified lenses or incorporating additional information with standard camera lenses, these computer vision tasks can benefit greatly.</p><p id="3904" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Typically, in camera systems, the optical design is established first, and then the image processing algorithm’s parameters are adjusted to achieve high-quality image reproduction. In contrast to this sequential design approach, the authors consider joint optimising of an optical system (such as the physical shape of the lens) simultaneously with the reconstruction algorithm’s parameters. They developed a fully-differentiable simulation model that optimizes both sets of parameters to minimize the deviation between the true and reconstructed image.</p><p id="272d" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">They published their ideas in a paper called<a class="af pv" href="https://web.stanford.edu/~boyd/papers/pdf/end_to_end_opt_optics.pdf" rel="noopener ugc nofollow" target="_blank"> End-to-end Optimization of Optics and Image Processing for Achromatic Extended Depth of Field and Super-resolution Imaging</a>.</p><p id="8bef" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The broad idea is to use a simulation module, which can simulate point spread functions (PSF) of differently shaped lenses and learn the best-tuned parameters for the optics model (hence the best-shaped lens) for this particular task which was on super-resolution.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np nq"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*qi793U05hDQBPt4U 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*qi793U05hDQBPt4U 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*qi793U05hDQBPt4U 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*qi793U05hDQBPt4U 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*qi793U05hDQBPt4U 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*qi793U05hDQBPt4U 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*qi793U05hDQBPt4U 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*qi793U05hDQBPt4U 640w, https://miro.medium.com/v2/resize:fit:720/0*qi793U05hDQBPt4U 720w, https://miro.medium.com/v2/resize:fit:750/0*qi793U05hDQBPt4U 750w, https://miro.medium.com/v2/resize:fit:786/0*qi793U05hDQBPt4U 786w, https://miro.medium.com/v2/resize:fit:828/0*qi793U05hDQBPt4U 828w, https://miro.medium.com/v2/resize:fit:1100/0*qi793U05hDQBPt4U 1100w, https://miro.medium.com/v2/resize:fit:1400/0*qi793U05hDQBPt4U 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="265" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Differentiable PSF Simulation</em></figcaption></figure><p id="7481" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">Applications of end-to-end learning</strong></p><p id="e5d9" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The same idea can be applied to other applications like monocular depth estimation which has been demonstrated in the paper<a class="af pv" href="http://www.computationalimaging.org/wp-content/uploads/2021/04/DeepDfD_ICCP2021.pdf" rel="noopener ugc nofollow" target="_blank"> Depth from Defocus with Learned Optics for Imaging and Occlusion-aware Depth Estimation</a>. The authors jointly optimize a deep CNN (a U-Net-like model) along with a fully differentiable optics model to produce high-quality depth maps using a single camera.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np nq"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*JE03Ta72Fk-4GyN8 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*JE03Ta72Fk-4GyN8 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*JE03Ta72Fk-4GyN8 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*JE03Ta72Fk-4GyN8 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*JE03Ta72Fk-4GyN8 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*JE03Ta72Fk-4GyN8 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*JE03Ta72Fk-4GyN8 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*JE03Ta72Fk-4GyN8 640w, https://miro.medium.com/v2/resize:fit:720/0*JE03Ta72Fk-4GyN8 720w, https://miro.medium.com/v2/resize:fit:750/0*JE03Ta72Fk-4GyN8 750w, https://miro.medium.com/v2/resize:fit:786/0*JE03Ta72Fk-4GyN8 786w, https://miro.medium.com/v2/resize:fit:828/0*JE03Ta72Fk-4GyN8 828w, https://miro.medium.com/v2/resize:fit:1100/0*JE03Ta72Fk-4GyN8 1100w, https://miro.medium.com/v2/resize:fit:1400/0*JE03Ta72Fk-4GyN8 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="155" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Deep CNN Model</em></figcaption></figure><p id="a725" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Carrying this idea forward, the authors went on to showcase an innovative<a class="af pv" href="https://ieeexplore.ieee.org/ielaam/34/9108332/9076617-aam.pdf" rel="noopener ugc nofollow" target="_blank"> Lens-less camera</a> where they proposed to eliminate the need for lenses in cameras and use a very thin phase mask instead. Phase masks are essentially transparent materials with different heights at different locations.</p><p id="5cd6" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">This causes phase modulation of the incoming wavefront and resultant wave interference produces the PSF at the sensor plane. Their proposed phase-mask framework takes the input of the target PSF and the desired device geometry (which as stated above can be learnt using a fully differentiable simulated optics model) and outputs an optimized phase-mask design.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div class="no np qb"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*LY2S60fQoxZXDFrd 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*LY2S60fQoxZXDFrd 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*LY2S60fQoxZXDFrd 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*LY2S60fQoxZXDFrd 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*LY2S60fQoxZXDFrd 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*LY2S60fQoxZXDFrd 1100w, https://miro.medium.com/v2/resize:fit:1376/format:webp/0*LY2S60fQoxZXDFrd 1376w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 688px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*LY2S60fQoxZXDFrd 640w, https://miro.medium.com/v2/resize:fit:720/0*LY2S60fQoxZXDFrd 720w, https://miro.medium.com/v2/resize:fit:750/0*LY2S60fQoxZXDFrd 750w, https://miro.medium.com/v2/resize:fit:786/0*LY2S60fQoxZXDFrd 786w, https://miro.medium.com/v2/resize:fit:828/0*LY2S60fQoxZXDFrd 828w, https://miro.medium.com/v2/resize:fit:1100/0*LY2S60fQoxZXDFrd 1100w, https://miro.medium.com/v2/resize:fit:1376/0*LY2S60fQoxZXDFrd 1376w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 688px"/><img alt="" class="bh mv ob c" width="688" height="250" loading="lazy" role="presentation"/></picture></div></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div class="no np qb"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*2cwccNQDVRvbwMZ5 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*2cwccNQDVRvbwMZ5 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*2cwccNQDVRvbwMZ5 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*2cwccNQDVRvbwMZ5 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*2cwccNQDVRvbwMZ5 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*2cwccNQDVRvbwMZ5 1100w, https://miro.medium.com/v2/resize:fit:1376/format:webp/0*2cwccNQDVRvbwMZ5 1376w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 688px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*2cwccNQDVRvbwMZ5 640w, https://miro.medium.com/v2/resize:fit:720/0*2cwccNQDVRvbwMZ5 720w, https://miro.medium.com/v2/resize:fit:750/0*2cwccNQDVRvbwMZ5 750w, https://miro.medium.com/v2/resize:fit:786/0*2cwccNQDVRvbwMZ5 786w, https://miro.medium.com/v2/resize:fit:828/0*2cwccNQDVRvbwMZ5 828w, https://miro.medium.com/v2/resize:fit:1100/0*2cwccNQDVRvbwMZ5 1100w, https://miro.medium.com/v2/resize:fit:1376/0*2cwccNQDVRvbwMZ5 1376w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 688px"/><img alt="" class="bh mv ob c" width="688" height="250" loading="lazy" role="presentation"/></picture></div></figure><h2 id="585f" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk"><strong class="al">Using FlatNet to enhance the output quality of phase masks</strong></h2><p id="6fd6" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Most of the phase mask output cannot be interpreted by humans. In another proposed paper called<a class="af pv" href="https://arxiv.org/pdf/2010.15440.pdf" rel="noopener ugc nofollow" target="_blank"> FlatNet: Towards Photorealistic Scene Reconstruction from Lensless Measurements</a>, the authors first train a model to learn to invert the forward operation of the lensless camera model. This allows them to obtain an intermediate representation with local structures intact.</p><p id="7f6f" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Once they obtain the output of the trainable inversion stage, which is of the same dimension as that of the natural image they want to recover, they use a fully convolutional network to map it to the perceptually enhanced image. They choose a U-Net to map the intermediate reconstruction to the final perceptually enhanced image.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qc"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*SJGGMmXq2jg7CS-I 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*SJGGMmXq2jg7CS-I 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*SJGGMmXq2jg7CS-I 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*SJGGMmXq2jg7CS-I 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*SJGGMmXq2jg7CS-I 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*SJGGMmXq2jg7CS-I 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*SJGGMmXq2jg7CS-I 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*SJGGMmXq2jg7CS-I 640w, https://miro.medium.com/v2/resize:fit:720/0*SJGGMmXq2jg7CS-I 720w, https://miro.medium.com/v2/resize:fit:750/0*SJGGMmXq2jg7CS-I 750w, https://miro.medium.com/v2/resize:fit:786/0*SJGGMmXq2jg7CS-I 786w, https://miro.medium.com/v2/resize:fit:828/0*SJGGMmXq2jg7CS-I 828w, https://miro.medium.com/v2/resize:fit:1100/0*SJGGMmXq2jg7CS-I 1100w, https://miro.medium.com/v2/resize:fit:1400/0*SJGGMmXq2jg7CS-I 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="147" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Reconstruction Of Scenes From Lensless Cameras</em></figcaption></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qd"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*0r1l563RyXrBur5f 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*0r1l563RyXrBur5f 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*0r1l563RyXrBur5f 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*0r1l563RyXrBur5f 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*0r1l563RyXrBur5f 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*0r1l563RyXrBur5f 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*0r1l563RyXrBur5f 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*0r1l563RyXrBur5f 640w, https://miro.medium.com/v2/resize:fit:720/0*0r1l563RyXrBur5f 720w, https://miro.medium.com/v2/resize:fit:750/0*0r1l563RyXrBur5f 750w, https://miro.medium.com/v2/resize:fit:786/0*0r1l563RyXrBur5f 786w, https://miro.medium.com/v2/resize:fit:828/0*0r1l563RyXrBur5f 828w, https://miro.medium.com/v2/resize:fit:1100/0*0r1l563RyXrBur5f 1100w, https://miro.medium.com/v2/resize:fit:1400/0*0r1l563RyXrBur5f 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="347" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Architecture of the Flatnet</em></figcaption></figure><h2 id="edba" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Key Takeaways from Plenary Talks</h2><p id="baad" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">The plenary talks were led by top-notch researchers and held during the final two days of the conference. The talks covered a diverse range of topics, including multi-sensory perception, efficient networks for graphics and rendering, transformers etc. offering attendees invaluable insights into the latest trends and advancements in the field.</p><h2 id="7721" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk"><strong class="al">Instant NGP: Neural Networks in High-Performance Graphics</strong></h2><p id="9259" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">This session was hosted by<a class="af pv" href="https://research.nvidia.com/person/thomas-muller" rel="noopener ugc nofollow" target="_blank"> Dr Thomas Müller</a>, the principal research scientist at NVIDIA. The talk was a case study on how the research team at NVIDIA was able to successfully train a<a class="af pv" href="https://www.matthewtancik.com/nerf" rel="noopener ugc nofollow" target="_blank"> Neural Radiance field (NeRF)</a> model in a matter of minutes or even less. They call it the<a class="af pv" href="https://github.com/NVlabs/instant-ngp" rel="noopener ugc nofollow" target="_blank"> Instant NGP (Instant Neural Graphics Primitive)</a>. But let’s back up a bit and understand what NeRF is first.</p><p id="f1c0" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">A<a class="af pv" href="https://arxiv.org/abs/2003.08934" rel="noopener ugc nofollow" target="_blank"> neural radiance field</a> (NeRF) is a fully-connected neural network that can generate novel views of complex 3D scenes, based on a partial set of 2D images. It is trained to use a rendering loss to reproduce input views of a scene. It works by taking input images representing a scene and interpolating between them to render one complete scene. NeRF is a highly effective way to generate images for synthetic data. A NeRF network is trained to map directly from viewing direction and spatial location (5D input) to opacity and colour (4D output), using volume rendering to render new views. You can read more about it in their<a class="af pv" href="https://arxiv.org/pdf/2003.08934.pdf" rel="noopener ugc nofollow" target="_blank"> paper</a>.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qe"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*UPUMf41JiMnBl-Lw 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*UPUMf41JiMnBl-Lw 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*UPUMf41JiMnBl-Lw 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*UPUMf41JiMnBl-Lw 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*UPUMf41JiMnBl-Lw 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*UPUMf41JiMnBl-Lw 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*UPUMf41JiMnBl-Lw 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*UPUMf41JiMnBl-Lw 640w, https://miro.medium.com/v2/resize:fit:720/0*UPUMf41JiMnBl-Lw 720w, https://miro.medium.com/v2/resize:fit:750/0*UPUMf41JiMnBl-Lw 750w, https://miro.medium.com/v2/resize:fit:786/0*UPUMf41JiMnBl-Lw 786w, https://miro.medium.com/v2/resize:fit:828/0*UPUMf41JiMnBl-Lw 828w, https://miro.medium.com/v2/resize:fit:1100/0*UPUMf41JiMnBl-Lw 1100w, https://miro.medium.com/v2/resize:fit:1400/0*UPUMf41JiMnBl-Lw 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="226" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">NeRF forward pass</em></figcaption></figure><p id="7fce" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">NeRF is a computationally-intensive algorithm, and rendering complex scenes can take hours or even days. Despite their reputation for being computationally expensive, neural networks can be trained and run efficiently for high-performance tasks. With the use of the appropriate data structures and algorithms, neural networks can run in the inner loops of real-time renderers and 3D reconstruction, resulting in an “instant NeRF.” Details about this approach can be found in their paper titled <a class="af pv" href="https://nvlabs.github.io/instant-ngp/assets/mueller2022instant.pdf" rel="noopener ugc nofollow" target="_blank">Instant Neural Graphics Primitives with a Multiresolution Hash Encoding.</a></p><p id="2493" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The speaker credits their success to the three pillars of Neural High-Performance Graphics which are</p><ul class=""><li id="5a2a" class="oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox qf qg qh bk">Small Neural Networks</li><li id="f899" class="oc od gu oe b hx qi og oh ia qj oj ok ol qk on oo op ql or os ot qm ov ow ox qf qg qh bk">Hybrid Data Structures</li><li id="b468" class="oc od gu oe b hx qi og oh ia qj oj ok ol qk on oo op ql or os ot qm ov ow ox qf qg qh bk">Task Specific GPU implementations</li></ul><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qn"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*6PqjLsysfFrGj9ty 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*6PqjLsysfFrGj9ty 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*6PqjLsysfFrGj9ty 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*6PqjLsysfFrGj9ty 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*6PqjLsysfFrGj9ty 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*6PqjLsysfFrGj9ty 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*6PqjLsysfFrGj9ty 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*6PqjLsysfFrGj9ty 640w, https://miro.medium.com/v2/resize:fit:720/0*6PqjLsysfFrGj9ty 720w, https://miro.medium.com/v2/resize:fit:750/0*6PqjLsysfFrGj9ty 750w, https://miro.medium.com/v2/resize:fit:786/0*6PqjLsysfFrGj9ty 786w, https://miro.medium.com/v2/resize:fit:828/0*6PqjLsysfFrGj9ty 828w, https://miro.medium.com/v2/resize:fit:1100/0*6PqjLsysfFrGj9ty 1100w, https://miro.medium.com/v2/resize:fit:1400/0*6PqjLsysfFrGj9ty 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="421" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Pillars of Neural High-Performance Graphics</em></figcaption></figure><p id="45e0" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">The importance of special priors</strong></p><p id="09ce" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Smaller and more efficient neural networks can significantly reduce computing time, but this can sometimes compromise the accuracy and quality of the output. To address this issue, the use of special priors, such as positional encodings, can help produce similar results when generating a 3D scene with a smaller NeRF model than the original larger model. The speakers argued that without the input positional encodings, this would not have been possible and thus emphasize the importance of smaller networks aided with priors. Read more about it<a class="af pv" href="https://bmild.github.io/fourfeat/index.html" rel="noopener ugc nofollow" target="_blank"> here</a>.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qn"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*EfDGGWcqHhT3j96m 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*EfDGGWcqHhT3j96m 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*EfDGGWcqHhT3j96m 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*EfDGGWcqHhT3j96m 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*EfDGGWcqHhT3j96m 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*EfDGGWcqHhT3j96m 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*EfDGGWcqHhT3j96m 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*EfDGGWcqHhT3j96m 640w, https://miro.medium.com/v2/resize:fit:720/0*EfDGGWcqHhT3j96m 720w, https://miro.medium.com/v2/resize:fit:750/0*EfDGGWcqHhT3j96m 750w, https://miro.medium.com/v2/resize:fit:786/0*EfDGGWcqHhT3j96m 786w, https://miro.medium.com/v2/resize:fit:828/0*EfDGGWcqHhT3j96m 828w, https://miro.medium.com/v2/resize:fit:1100/0*EfDGGWcqHhT3j96m 1100w, https://miro.medium.com/v2/resize:fit:1400/0*EfDGGWcqHhT3j96m 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="395" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">The Effectiveness of Input Encoding</em></figcaption></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div class="no np qo"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*TQo-ryaCR2ju0gdz 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*TQo-ryaCR2ju0gdz 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*TQo-ryaCR2ju0gdz 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*TQo-ryaCR2ju0gdz 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*TQo-ryaCR2ju0gdz 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*TQo-ryaCR2ju0gdz 1100w, https://miro.medium.com/v2/resize:fit:1200/format:webp/0*TQo-ryaCR2ju0gdz 1200w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 600px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*TQo-ryaCR2ju0gdz 640w, https://miro.medium.com/v2/resize:fit:720/0*TQo-ryaCR2ju0gdz 720w, https://miro.medium.com/v2/resize:fit:750/0*TQo-ryaCR2ju0gdz 750w, https://miro.medium.com/v2/resize:fit:786/0*TQo-ryaCR2ju0gdz 786w, https://miro.medium.com/v2/resize:fit:828/0*TQo-ryaCR2ju0gdz 828w, https://miro.medium.com/v2/resize:fit:1100/0*TQo-ryaCR2ju0gdz 1100w, https://miro.medium.com/v2/resize:fit:1200/0*TQo-ryaCR2ju0gdz 1200w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 600px"/><img alt="" class="bh mv ob c" width="600" height="325" loading="lazy" role="presentation"/></picture></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Fourier Feature Input Results vs. Standard Input Results</em></figcaption></figure><p id="5f0e" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Usually, a lot of time is wasted in I/O read-and-write operations. The authors state that to improve speed it is necessary to modify the existing data structures to something specific to the task. For the 2D to 3D scene projection, they designed Multiresolution Hash Encoding.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qn"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*Fu3-9-725acwQyFs 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*Fu3-9-725acwQyFs 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*Fu3-9-725acwQyFs 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*Fu3-9-725acwQyFs 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*Fu3-9-725acwQyFs 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*Fu3-9-725acwQyFs 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*Fu3-9-725acwQyFs 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*Fu3-9-725acwQyFs 640w, https://miro.medium.com/v2/resize:fit:720/0*Fu3-9-725acwQyFs 720w, https://miro.medium.com/v2/resize:fit:750/0*Fu3-9-725acwQyFs 750w, https://miro.medium.com/v2/resize:fit:786/0*Fu3-9-725acwQyFs 786w, https://miro.medium.com/v2/resize:fit:828/0*Fu3-9-725acwQyFs 828w, https://miro.medium.com/v2/resize:fit:1100/0*Fu3-9-725acwQyFs 1100w, https://miro.medium.com/v2/resize:fit:1400/0*Fu3-9-725acwQyFs 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="381" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Multiresolution Hash Encoding</em></figcaption></figure><p id="1b40" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Finally, they stated that the current deep learning frameworks are not optimal enough to exploit the speed of GPUs to the fullest. They instead proposed a new framework to train your Neural networks called<a class="af pv" href="https://github.com/NVlabs/tiny-cuda-nn" rel="noopener ugc nofollow" target="_blank"> Tiny Cuda</a>. It is a small, self-contained framework for training and querying neural networks. It contains a lightning-fast<a class="af pv" href="https://raw.githubusercontent.com/NVlabs/tiny-cuda-nn/master/data/readme/fully-fused-mlp-diagram.png" rel="noopener ugc nofollow" target="_blank"> “fully fused” multi-layer perceptron</a> (<a class="af pv" href="https://tom94.net/data/publications/mueller21realtime/mueller21realtime.pdf" rel="noopener ugc nofollow" target="_blank">paper</a>), a versatile<a class="af pv" href="https://raw.githubusercontent.com/NVlabs/tiny-cuda-nn/master/data/readme/multiresolution-hash-encoding-diagram.png" rel="noopener ugc nofollow" target="_blank"> multiresolution hash encoding</a> (<a class="af pv" href="https://nvlabs.github.io/instant-ngp/assets/mueller2022instant.pdf" rel="noopener ugc nofollow" target="_blank">paper</a>), as well as support for various other input encodings, losses, and optimizers.</p><h2 id="ae59" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Strong Interpretable Priors Are All We Need</h2><p id="be1d" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">This session was led by<a class="af pv" href="https://www.weizmann.ac.il/math/dekel/home" rel="noopener ugc nofollow" target="_blank"> Dr Tali Dekel</a>, a research scientist at Google. Computer vision has recently made exciting progress, with new architectures and self-supervised learning paradigms rapidly improving. As computing power increases, models scale in size and training data, resulting in “foundation models” — billion-parameter neural networks trained in a self-supervised manner on massive amounts of unlabelled imagery.</p><p id="12e9" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Such models learn extraordinary priors about our visual world, as evident by their breakthrough results in a plethora of visual inference and synthesis tasks. Nevertheless, their knowledge is buried and hidden in the vast space of the network’s weights.</p><p id="fb30" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The speaker presented a series of works that aim to investigate the internal representations learned by large-scale models. By studying their priors and utilizing them in classical and new visual tasks, the research covers co-segmenting two images into coherent object parts and using text to modify the appearance of moving objects in real-world videos.</p><p id="367e" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">There has been a constant evolution in visual descriptors used for Computer Vision tasks. Starting from the early hand-crafted features (SIFT, HOG, SURF, ORB), people eventually moved towards Deep CNN-based features with the rise of the Deep Learning era. However, with the latest developments in the field of Transformers, specifically Vision Transformers, is it time to move towards Deep ViT-based Features?</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qp"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*S2SaV_2DRT753seQ 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*S2SaV_2DRT753seQ 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*S2SaV_2DRT753seQ 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*S2SaV_2DRT753seQ 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*S2SaV_2DRT753seQ 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*S2SaV_2DRT753seQ 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*S2SaV_2DRT753seQ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*S2SaV_2DRT753seQ 640w, https://miro.medium.com/v2/resize:fit:720/0*S2SaV_2DRT753seQ 720w, https://miro.medium.com/v2/resize:fit:750/0*S2SaV_2DRT753seQ 750w, https://miro.medium.com/v2/resize:fit:786/0*S2SaV_2DRT753seQ 786w, https://miro.medium.com/v2/resize:fit:828/0*S2SaV_2DRT753seQ 828w, https://miro.medium.com/v2/resize:fit:1100/0*S2SaV_2DRT753seQ 1100w, https://miro.medium.com/v2/resize:fit:1400/0*S2SaV_2DRT753seQ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="410" loading="lazy" role="presentation"/></picture></div></div></figure><p id="8884" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">Exciting innovations in AI: Self-Supervised Learning & Transformers</strong></p><p id="c2b2" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The speaker mentioned how many of the most exciting new AI breakthroughs have come from two recent innovations:<a class="af pv" href="https://ai.facebook.com/blog/self-supervised-learning-the-dark-matter-of-intelligence/" rel="noopener ugc nofollow" target="_blank"> self-supervised learning</a>, which allows machines to learn from random, unlabelled examples; and<a class="af pv" href="https://ai.facebook.com/blog/roberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems/" rel="noopener ugc nofollow" target="_blank"> Transformers</a>, which enable AI models to selectively focus on certain parts of their input and thus reason more effectively. A recent work called<a class="af pv" href="https://arxiv.org/pdf/2104.14294.pdf" rel="noopener ugc nofollow" target="_blank"> Self-Supervised ViT — DINO</a> is a great example of this. Interestingly, the acronym DINO comes from self-<strong class="oe he">di</strong>stillation with <strong class="oe he">no</strong> labels.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qq"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*wOJJWyg0kIUkynp8 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*wOJJWyg0kIUkynp8 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*wOJJWyg0kIUkynp8 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*wOJJWyg0kIUkynp8 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*wOJJWyg0kIUkynp8 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*wOJJWyg0kIUkynp8 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*wOJJWyg0kIUkynp8 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*wOJJWyg0kIUkynp8 640w, https://miro.medium.com/v2/resize:fit:720/0*wOJJWyg0kIUkynp8 720w, https://miro.medium.com/v2/resize:fit:750/0*wOJJWyg0kIUkynp8 750w, https://miro.medium.com/v2/resize:fit:786/0*wOJJWyg0kIUkynp8 786w, https://miro.medium.com/v2/resize:fit:828/0*wOJJWyg0kIUkynp8 828w, https://miro.medium.com/v2/resize:fit:1100/0*wOJJWyg0kIUkynp8 1100w, https://miro.medium.com/v2/resize:fit:1400/0*wOJJWyg0kIUkynp8 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="394" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Training ViT with the DINO algorithm</em></figcaption></figure><p id="eae4" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">By training ViT with the DINO algorithm, the authors observed that the model automatically learns an interpretable representation and separates the main object from the background clutter. It learns to segment objects without any human-generated annotation or any form of dedicated dense pixel-level loss.</p><p id="ee20" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The core component of Vision Transformers is self-attention layers. In this model, each spatial location builds its representation by “attending” to the other locations. That way, by “looking” at other, potentially distant pieces of the image, the network builds a rich, high-level understanding of the scene. When visualizing the local attention maps in the network, it is apparent that they correspond to coherent semantic regions in the image.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qr"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*UhvWx9FQcNukxcDW 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*UhvWx9FQcNukxcDW 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*UhvWx9FQcNukxcDW 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*UhvWx9FQcNukxcDW 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*UhvWx9FQcNukxcDW 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*UhvWx9FQcNukxcDW 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*UhvWx9FQcNukxcDW 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*UhvWx9FQcNukxcDW 640w, https://miro.medium.com/v2/resize:fit:720/0*UhvWx9FQcNukxcDW 720w, https://miro.medium.com/v2/resize:fit:750/0*UhvWx9FQcNukxcDW 750w, https://miro.medium.com/v2/resize:fit:786/0*UhvWx9FQcNukxcDW 786w, https://miro.medium.com/v2/resize:fit:828/0*UhvWx9FQcNukxcDW 828w, https://miro.medium.com/v2/resize:fit:1100/0*UhvWx9FQcNukxcDW 1100w, https://miro.medium.com/v2/resize:fit:1400/0*UhvWx9FQcNukxcDW 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="404" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">DINO Attention Heads</em></figcaption></figure><p id="525d" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">How does DINO work?</strong></p><p id="71c4" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">DINO works by interpreting self-supervision as a special case of self-distillation, where no labels are used at all. It trains a student network by simply matching the output of a teacher network over different views of the same image.</p><p id="d236" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The authors of this paper identified two components from previous self-supervised approaches that are particularly important for strong performance on ViT, the momentum teacher and multi-crop training, and integrated them into their framework.</p><p id="273d" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">In the image below you can see the difference between the feature map representations of both supervised and self-supervised variants of DINO ViT and ResNet. It is visible that deeper layers of Self-supervised DINO ViT produce more semantically coherent features and can even identify similar objects.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qs"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*NUiAMYgs7eolUiwm 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*NUiAMYgs7eolUiwm 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*NUiAMYgs7eolUiwm 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*NUiAMYgs7eolUiwm 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*NUiAMYgs7eolUiwm 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*NUiAMYgs7eolUiwm 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*NUiAMYgs7eolUiwm 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*NUiAMYgs7eolUiwm 640w, https://miro.medium.com/v2/resize:fit:720/0*NUiAMYgs7eolUiwm 720w, https://miro.medium.com/v2/resize:fit:750/0*NUiAMYgs7eolUiwm 750w, https://miro.medium.com/v2/resize:fit:786/0*NUiAMYgs7eolUiwm 786w, https://miro.medium.com/v2/resize:fit:828/0*NUiAMYgs7eolUiwm 828w, https://miro.medium.com/v2/resize:fit:1100/0*NUiAMYgs7eolUiwm 1100w, https://miro.medium.com/v2/resize:fit:1400/0*NUiAMYgs7eolUiwm 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="417" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Feature maps of Supervised and Self-supervised DINO ViT & ResNet</em></figcaption></figure><p id="681a" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">These DINO ViT features can be used for a plethora of applications such as Zero-shot Co-segmentation and Part-Cosegmentation. In all cases, lightweight methodologies are designed, that leverage the universal knowledge learned by large-scale models through new visual descriptors and perceptual losses. The methods are “zero-shot’’. They require no training data and are self-supervised — requiring no manual labels and thus can be applied across different domains and tasks for which training data is scarce.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qt"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*0Y_osIMD1Q24NxgY 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*0Y_osIMD1Q24NxgY 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*0Y_osIMD1Q24NxgY 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*0Y_osIMD1Q24NxgY 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*0Y_osIMD1Q24NxgY 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*0Y_osIMD1Q24NxgY 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*0Y_osIMD1Q24NxgY 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*0Y_osIMD1Q24NxgY 640w, https://miro.medium.com/v2/resize:fit:720/0*0Y_osIMD1Q24NxgY 720w, https://miro.medium.com/v2/resize:fit:750/0*0Y_osIMD1Q24NxgY 750w, https://miro.medium.com/v2/resize:fit:786/0*0Y_osIMD1Q24NxgY 786w, https://miro.medium.com/v2/resize:fit:828/0*0Y_osIMD1Q24NxgY 828w, https://miro.medium.com/v2/resize:fit:1100/0*0Y_osIMD1Q24NxgY 1100w, https://miro.medium.com/v2/resize:fit:1400/0*0Y_osIMD1Q24NxgY 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="411" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Zero-shot Co-segmentation and Part-Cosegmentation</em></figcaption></figure><h2 id="99eb" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Insights from Paper Presentations:</h2><h2 id="f9d0" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">FLOAT: Factorized Learning of Object Attributes for Improved Multi-object Multi-part Scene Parsing</h2><p id="a633" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Multi-object multi-part scene parsing is a challenging task which requires detecting multiple object classes in a scene and segmenting the semantic parts within each object.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qd"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*1UmLHr1hW83Cqm-v 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*1UmLHr1hW83Cqm-v 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*1UmLHr1hW83Cqm-v 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*1UmLHr1hW83Cqm-v 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*1UmLHr1hW83Cqm-v 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*1UmLHr1hW83Cqm-v 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*1UmLHr1hW83Cqm-v 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*1UmLHr1hW83Cqm-v 640w, https://miro.medium.com/v2/resize:fit:720/0*1UmLHr1hW83Cqm-v 720w, https://miro.medium.com/v2/resize:fit:750/0*1UmLHr1hW83Cqm-v 750w, https://miro.medium.com/v2/resize:fit:786/0*1UmLHr1hW83Cqm-v 786w, https://miro.medium.com/v2/resize:fit:828/0*1UmLHr1hW83Cqm-v 828w, https://miro.medium.com/v2/resize:fit:1100/0*1UmLHr1hW83Cqm-v 1100w, https://miro.medium.com/v2/resize:fit:1400/0*1UmLHr1hW83Cqm-v 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="395" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Multi-object Multi-part Segmentation</em></figcaption></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qd"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*p4KhfHluCqaGO03N 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*p4KhfHluCqaGO03N 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*p4KhfHluCqaGO03N 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*p4KhfHluCqaGO03N 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*p4KhfHluCqaGO03N 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*p4KhfHluCqaGO03N 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*p4KhfHluCqaGO03N 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*p4KhfHluCqaGO03N 640w, https://miro.medium.com/v2/resize:fit:720/0*p4KhfHluCqaGO03N 720w, https://miro.medium.com/v2/resize:fit:750/0*p4KhfHluCqaGO03N 750w, https://miro.medium.com/v2/resize:fit:786/0*p4KhfHluCqaGO03N 786w, https://miro.medium.com/v2/resize:fit:828/0*p4KhfHluCqaGO03N 828w, https://miro.medium.com/v2/resize:fit:1100/0*p4KhfHluCqaGO03N 1100w, https://miro.medium.com/v2/resize:fit:1400/0*p4KhfHluCqaGO03N 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="395" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Ground Truth Label Map Changes</em></figcaption></figure><p id="1f43" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">For this, the authors produce changes in the monolithic object label map structures to introduce more information such as front/back, left/right, and animate/inanimate parts of the object. They use this information to create the Pascal-Part201 dataset. They propose the following model to solve the multi-object multi-part scene parsing challenge. The model consists of encoder-decoder-style architecture with different decoders for object level segmentation, front/back, left/right, and animate/inanimate parts of the object. Finally, the feature maps are merged and an Inference Time Zoom Refinement (IZR module) is used to get the final output.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qu"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*8SElPbryjLaj6w3U 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*8SElPbryjLaj6w3U 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*8SElPbryjLaj6w3U 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*8SElPbryjLaj6w3U 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*8SElPbryjLaj6w3U 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*8SElPbryjLaj6w3U 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*8SElPbryjLaj6w3U 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*8SElPbryjLaj6w3U 640w, https://miro.medium.com/v2/resize:fit:720/0*8SElPbryjLaj6w3U 720w, https://miro.medium.com/v2/resize:fit:750/0*8SElPbryjLaj6w3U 750w, https://miro.medium.com/v2/resize:fit:786/0*8SElPbryjLaj6w3U 786w, https://miro.medium.com/v2/resize:fit:828/0*8SElPbryjLaj6w3U 828w, https://miro.medium.com/v2/resize:fit:1100/0*8SElPbryjLaj6w3U 1100w, https://miro.medium.com/v2/resize:fit:1400/0*8SElPbryjLaj6w3U 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="396" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">The FLOAT model</em></figcaption></figure><h2 id="d8d6" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Can you even tell left from right? Presenting a new challenge for VQA</h2><p id="7f78" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Visual Question Answering (VQA) research aims to create a computer system that can answer questions using both an image and natural language. VQA needs a means of evaluating the strengths and weaknesses of models. One is the evaluation of compositional generalisation, or the ability of a model to answer well on scenes whose scene setups are different from the training set. For this, we need datasets whose train and test sets differ significantly in composition.</p><p id="a994" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">This study introduces quantitative measures of compositional separation and shows that current VQA datasets are inadequate for evaluation. To solve this, they present Uncommon Objects in Unseen Configurations (UOUC), a synthetic dataset for VQA. UOUC is at once fairly complex while also being well-separated, compositionally. UOUC contains 380 object classes from 528 characters in the Dungeons and Dragons game, with 200,000 scenes in the train set and 30,000 in the test set.</p><p id="d44a" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">To study compositional generalisation, simple reasoning, and memorisation, each scene of UOUC is annotated with up to 10 novel questions. These deal with spatial relationships, hypothetical changes to scenes, counting, comparison, memorisation and memory-based reasoning. In total, UOUC presents over 2 million questions. UOUC also finds itself as a strong challenger to well-performing models for VQA. Read the full paper<a class="af pv" href="https://arxiv.org/pdf/2203.07664.pdf" rel="noopener ugc nofollow" target="_blank"> here</a>.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qv"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*qXS8MOGvb3l3CHaP 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*qXS8MOGvb3l3CHaP 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*qXS8MOGvb3l3CHaP 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*qXS8MOGvb3l3CHaP 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*qXS8MOGvb3l3CHaP 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*qXS8MOGvb3l3CHaP 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*qXS8MOGvb3l3CHaP 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*qXS8MOGvb3l3CHaP 640w, https://miro.medium.com/v2/resize:fit:720/0*qXS8MOGvb3l3CHaP 720w, https://miro.medium.com/v2/resize:fit:750/0*qXS8MOGvb3l3CHaP 750w, https://miro.medium.com/v2/resize:fit:786/0*qXS8MOGvb3l3CHaP 786w, https://miro.medium.com/v2/resize:fit:828/0*qXS8MOGvb3l3CHaP 828w, https://miro.medium.com/v2/resize:fit:1100/0*qXS8MOGvb3l3CHaP 1100w, https://miro.medium.com/v2/resize:fit:1400/0*qXS8MOGvb3l3CHaP 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="300" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Visual Question Answering datasets</em></figcaption></figure><p id="f745" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">Learning compositional structures for deep learning: Why routing-by-agreement is necessary</strong></p><p id="2a96" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">A formal description of the compositionality of neural networks is associated directly with the formal grammar structure of the objects it seeks to represent. This formal grammar structure specifies the kind of components that make up an object, and also the configurations they are allowed to be in. In other words, objects can be described as a parse tree of its components — a structure that can be seen as a candidate for building connection patterns among neurons in neural networks. The authors present a formal grammar description of convolutional neural networks and capsule networks that shows how capsule networks can enforce such parse-tree structures, while CNNs do not. Read the full paper<a class="af pv" href="https://arxiv.org/pdf/2010.01488.pdf" rel="noopener ugc nofollow" target="_blank"> here</a>.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qv"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*Ncl84v9pZEiFMhU3 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*Ncl84v9pZEiFMhU3 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*Ncl84v9pZEiFMhU3 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*Ncl84v9pZEiFMhU3 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*Ncl84v9pZEiFMhU3 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*Ncl84v9pZEiFMhU3 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*Ncl84v9pZEiFMhU3 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*Ncl84v9pZEiFMhU3 640w, https://miro.medium.com/v2/resize:fit:720/0*Ncl84v9pZEiFMhU3 720w, https://miro.medium.com/v2/resize:fit:750/0*Ncl84v9pZEiFMhU3 750w, https://miro.medium.com/v2/resize:fit:786/0*Ncl84v9pZEiFMhU3 786w, https://miro.medium.com/v2/resize:fit:828/0*Ncl84v9pZEiFMhU3 828w, https://miro.medium.com/v2/resize:fit:1100/0*Ncl84v9pZEiFMhU3 1100w, https://miro.medium.com/v2/resize:fit:1400/0*Ncl84v9pZEiFMhU3 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="300" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Change in Compositionality</em></figcaption></figure><h2 id="0579" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Learnings from Industry Sessions:</h2><p id="87f0" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">The industry sessions featured interesting research work published by various competitors in the market. They had set up posters and demo booths where people could discuss more about their work and can see real-time demos. The sessions were from companies like Qualcomm, Adobe, Samsung R&D, L&T etc.</p><h2 id="4875" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Samsung R&D</h2><p id="cb5b" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Researchers showcased work on various image editing features that they have incorporated into their latest mobile phones. Some of the notable features are shadow remover, photo remaster, image in-painting, and portrait mode, which are all deployed in their latest smartphones.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qw"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*zmq28qzYWLzLa0gp 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*zmq28qzYWLzLa0gp 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*zmq28qzYWLzLa0gp 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*zmq28qzYWLzLa0gp 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*zmq28qzYWLzLa0gp 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*zmq28qzYWLzLa0gp 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*zmq28qzYWLzLa0gp 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*zmq28qzYWLzLa0gp 640w, https://miro.medium.com/v2/resize:fit:720/0*zmq28qzYWLzLa0gp 720w, https://miro.medium.com/v2/resize:fit:750/0*zmq28qzYWLzLa0gp 750w, https://miro.medium.com/v2/resize:fit:786/0*zmq28qzYWLzLa0gp 786w, https://miro.medium.com/v2/resize:fit:828/0*zmq28qzYWLzLa0gp 828w, https://miro.medium.com/v2/resize:fit:1100/0*zmq28qzYWLzLa0gp 1100w, https://miro.medium.com/v2/resize:fit:1400/0*zmq28qzYWLzLa0gp 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="425" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Samsung’s new image editing features</em></figcaption></figure><p id="336d" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk"><strong class="oe he">Under Display Camera</strong></p><p id="dcf5" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Apart from this, their major contribution was the development of the under-display camera. An Under Display Camera (UDC) is a breakthrough innovation that enables an uninterrupted viewing experience on a mobile device by hiding the camera under the display and dedicating the whole screen to users while applications are running. It not only requires hardware innovation by placing a camera under a display panel but also requires algorithm innovation for restoring image quality — one of the most complex image restoration problems.</p><p id="1ca2" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">As the camera is placed underneath the display, the Under Display Camera can suffer from poor image quality caused by diffraction artefacts, which results in flare, saturation, blur and haze. Therefore, while the Under Display Camera brings a better display experience, it also affects camera image quality and other downstream vision tasks. These complex and diverse distortions make restoring Under Display Camera images extremely challenging.</p><p id="a7bc" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">In this talk, the author discussed some of the challenges with the Under Display Camera system & presented their work on image restoration for Under Display Camera.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np nq"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*fARlGdHBK5cDgZhK 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*fARlGdHBK5cDgZhK 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*fARlGdHBK5cDgZhK 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*fARlGdHBK5cDgZhK 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*fARlGdHBK5cDgZhK 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*fARlGdHBK5cDgZhK 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*fARlGdHBK5cDgZhK 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*fARlGdHBK5cDgZhK 640w, https://miro.medium.com/v2/resize:fit:720/0*fARlGdHBK5cDgZhK 720w, https://miro.medium.com/v2/resize:fit:750/0*fARlGdHBK5cDgZhK 750w, https://miro.medium.com/v2/resize:fit:786/0*fARlGdHBK5cDgZhK 786w, https://miro.medium.com/v2/resize:fit:828/0*fARlGdHBK5cDgZhK 828w, https://miro.medium.com/v2/resize:fit:1100/0*fARlGdHBK5cDgZhK 1100w, https://miro.medium.com/v2/resize:fit:1400/0*fARlGdHBK5cDgZhK 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="394" loading="lazy" role="presentation"/></picture></div></div></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qx"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*wo6CH75G6vF1jlcJ 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*wo6CH75G6vF1jlcJ 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*wo6CH75G6vF1jlcJ 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*wo6CH75G6vF1jlcJ 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*wo6CH75G6vF1jlcJ 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*wo6CH75G6vF1jlcJ 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*wo6CH75G6vF1jlcJ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*wo6CH75G6vF1jlcJ 640w, https://miro.medium.com/v2/resize:fit:720/0*wo6CH75G6vF1jlcJ 720w, https://miro.medium.com/v2/resize:fit:750/0*wo6CH75G6vF1jlcJ 750w, https://miro.medium.com/v2/resize:fit:786/0*wo6CH75G6vF1jlcJ 786w, https://miro.medium.com/v2/resize:fit:828/0*wo6CH75G6vF1jlcJ 828w, https://miro.medium.com/v2/resize:fit:1100/0*wo6CH75G6vF1jlcJ 1100w, https://miro.medium.com/v2/resize:fit:1400/0*wo6CH75G6vF1jlcJ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="323" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Under Display Camera</em></figcaption></figure><h2 id="761f" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Adobe</h2><p id="8013" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">In this talk, the speakers provided an overview of Adobe Research and the key areas that they are working on. They gave us a peek at some of their recent work on video generation, image out-painting and graphic design harmonization. Their recent work on image out-painting and animating still images show how expressing visual data via intermediate representations and manipulating provides better outputs against direct pixel-level manipulations.</p><p id="1734" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The authors propose a method to interactively control the animation of fluid elements in still images to generate cinemagraphs. Specifically, they focus on the animation of fluid elements like water, smoke, and fire, which have the properties of repeating textures and continuous fluid motion. They represent the motion of such fluid elements in the image in the form of a constant 2D optical flow map. The user can provide any number of arrow directions and the associated speed along with a mask of the regions the user wants to animate. The user-provided input arrow directions, their corresponding speed values, and the mask is then converted into a dense flow map representing a constant optical flow map (FD).</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div class="no np qo"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*tnYUW4uPgCwjkroY 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*tnYUW4uPgCwjkroY 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*tnYUW4uPgCwjkroY 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*tnYUW4uPgCwjkroY 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*tnYUW4uPgCwjkroY 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*tnYUW4uPgCwjkroY 1100w, https://miro.medium.com/v2/resize:fit:1200/format:webp/0*tnYUW4uPgCwjkroY 1200w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 600px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*tnYUW4uPgCwjkroY 640w, https://miro.medium.com/v2/resize:fit:720/0*tnYUW4uPgCwjkroY 720w, https://miro.medium.com/v2/resize:fit:750/0*tnYUW4uPgCwjkroY 750w, https://miro.medium.com/v2/resize:fit:786/0*tnYUW4uPgCwjkroY 786w, https://miro.medium.com/v2/resize:fit:828/0*tnYUW4uPgCwjkroY 828w, https://miro.medium.com/v2/resize:fit:1100/0*tnYUW4uPgCwjkroY 1100w, https://miro.medium.com/v2/resize:fit:1200/0*tnYUW4uPgCwjkroY 1200w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 600px"/><img alt="" class="bh mv ob c" width="600" height="131" loading="lazy" role="presentation"/></picture></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Animation of fluid elements</em></figcaption></figure><p id="cf40" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The authors observe that FD, obtained using simple exponential operations can closely approximate the plausible motion of elements in the image. They further refined a computed dense optical flow map FD using a generative-adversarial network (GAN) to obtain a more realistic flow map. A novel U-Net-based architecture was proposed to auto-regressively generate future frames using the refined optical flow map by forward-warping the input image features at different resolutions.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qy"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*U73Z6WEgRM6TAsVA 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*U73Z6WEgRM6TAsVA 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*U73Z6WEgRM6TAsVA 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*U73Z6WEgRM6TAsVA 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*U73Z6WEgRM6TAsVA 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*U73Z6WEgRM6TAsVA 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*U73Z6WEgRM6TAsVA 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*U73Z6WEgRM6TAsVA 640w, https://miro.medium.com/v2/resize:fit:720/0*U73Z6WEgRM6TAsVA 720w, https://miro.medium.com/v2/resize:fit:750/0*U73Z6WEgRM6TAsVA 750w, https://miro.medium.com/v2/resize:fit:786/0*U73Z6WEgRM6TAsVA 786w, https://miro.medium.com/v2/resize:fit:828/0*U73Z6WEgRM6TAsVA 828w, https://miro.medium.com/v2/resize:fit:1100/0*U73Z6WEgRM6TAsVA 1100w, https://miro.medium.com/v2/resize:fit:1400/0*U73Z6WEgRM6TAsVA 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="419" loading="lazy" role="presentation"/></picture></div></div></figure><p id="014a" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Some other research showcased were:</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qz"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*d_KvKweXnnmGAktu 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*d_KvKweXnnmGAktu 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*d_KvKweXnnmGAktu 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*d_KvKweXnnmGAktu 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*d_KvKweXnnmGAktu 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*d_KvKweXnnmGAktu 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*d_KvKweXnnmGAktu 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*d_KvKweXnnmGAktu 640w, https://miro.medium.com/v2/resize:fit:720/0*d_KvKweXnnmGAktu 720w, https://miro.medium.com/v2/resize:fit:750/0*d_KvKweXnnmGAktu 750w, https://miro.medium.com/v2/resize:fit:786/0*d_KvKweXnnmGAktu 786w, https://miro.medium.com/v2/resize:fit:828/0*d_KvKweXnnmGAktu 828w, https://miro.medium.com/v2/resize:fit:1100/0*d_KvKweXnnmGAktu 1100w, https://miro.medium.com/v2/resize:fit:1400/0*d_KvKweXnnmGAktu 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="396" loading="lazy" role="presentation"/></picture></div></div></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np ra"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*W3A5AeXuVbURV-H_ 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*W3A5AeXuVbURV-H_ 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*W3A5AeXuVbURV-H_ 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*W3A5AeXuVbURV-H_ 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*W3A5AeXuVbURV-H_ 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*W3A5AeXuVbURV-H_ 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*W3A5AeXuVbURV-H_ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*W3A5AeXuVbURV-H_ 640w, https://miro.medium.com/v2/resize:fit:720/0*W3A5AeXuVbURV-H_ 720w, https://miro.medium.com/v2/resize:fit:750/0*W3A5AeXuVbURV-H_ 750w, https://miro.medium.com/v2/resize:fit:786/0*W3A5AeXuVbURV-H_ 786w, https://miro.medium.com/v2/resize:fit:828/0*W3A5AeXuVbURV-H_ 828w, https://miro.medium.com/v2/resize:fit:1100/0*W3A5AeXuVbURV-H_ 1100w, https://miro.medium.com/v2/resize:fit:1400/0*W3A5AeXuVbURV-H_ 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="402" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">Design Understand & Generation (top), Image Understanding & Generation (bottom)</em></figcaption></figure><h2 id="19e6" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">TCS Research</h2><p id="4bee" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Draping a 3D human mesh has garnered broad interest due to its wide applicability in virtual try-on, animations, etc. The 3D garment produced by existing methods are often inconsistent with the body shape, pose, and measurements. This paper proposes a single unified learning-based framework (DeepDraper) to predict garment deformation as a function of body shape, pose, measurements, and garment styles. The authors train the DeepDraper with coupled geometric and multi-view perceptual losses.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np qn"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*HgLCrZosH91anc4I 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*HgLCrZosH91anc4I 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*HgLCrZosH91anc4I 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*HgLCrZosH91anc4I 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*HgLCrZosH91anc4I 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*HgLCrZosH91anc4I 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*HgLCrZosH91anc4I 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*HgLCrZosH91anc4I 640w, https://miro.medium.com/v2/resize:fit:720/0*HgLCrZosH91anc4I 720w, https://miro.medium.com/v2/resize:fit:750/0*HgLCrZosH91anc4I 750w, https://miro.medium.com/v2/resize:fit:786/0*HgLCrZosH91anc4I 786w, https://miro.medium.com/v2/resize:fit:828/0*HgLCrZosH91anc4I 828w, https://miro.medium.com/v2/resize:fit:1100/0*HgLCrZosH91anc4I 1100w, https://miro.medium.com/v2/resize:fit:1400/0*HgLCrZosH91anc4I 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="394" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">GarSim</em></figcaption></figure><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np rb"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*dsr6atoqxkUl5pIM 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*dsr6atoqxkUl5pIM 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*dsr6atoqxkUl5pIM 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*dsr6atoqxkUl5pIM 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*dsr6atoqxkUl5pIM 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*dsr6atoqxkUl5pIM 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*dsr6atoqxkUl5pIM 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*dsr6atoqxkUl5pIM 640w, https://miro.medium.com/v2/resize:fit:720/0*dsr6atoqxkUl5pIM 720w, https://miro.medium.com/v2/resize:fit:750/0*dsr6atoqxkUl5pIM 750w, https://miro.medium.com/v2/resize:fit:786/0*dsr6atoqxkUl5pIM 786w, https://miro.medium.com/v2/resize:fit:828/0*dsr6atoqxkUl5pIM 828w, https://miro.medium.com/v2/resize:fit:1100/0*dsr6atoqxkUl5pIM 1100w, https://miro.medium.com/v2/resize:fit:1400/0*dsr6atoqxkUl5pIM 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="558" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">DeepDraper</em></figcaption></figure><p id="c440" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">Unlike existing methods, they additionally model garment deformations as a function of standard body measurements, which generally a buyer or a designer uses to buy or design perfect-fit clothes. In addition to that, the authors claim that DeepDraper is 10 times smaller in size and 23 times faster than the closest state-of-the-art method (TailorNet), which favours its use in real-time applications with less computational power.</p><figure class="nr ns nt nu nv nw no np paragraph-image"><div role="button" tabindex="0" class="nx ny fj nz bh oa"><div class="no np rc"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*IZ7oyMJ94lePxVg0 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*IZ7oyMJ94lePxVg0 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*IZ7oyMJ94lePxVg0 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*IZ7oyMJ94lePxVg0 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*IZ7oyMJ94lePxVg0 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*IZ7oyMJ94lePxVg0 1100w, https://miro.medium.com/v2/resize:fit:1400/format:webp/0*IZ7oyMJ94lePxVg0 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*IZ7oyMJ94lePxVg0 640w, https://miro.medium.com/v2/resize:fit:720/0*IZ7oyMJ94lePxVg0 720w, https://miro.medium.com/v2/resize:fit:750/0*IZ7oyMJ94lePxVg0 750w, https://miro.medium.com/v2/resize:fit:786/0*IZ7oyMJ94lePxVg0 786w, https://miro.medium.com/v2/resize:fit:828/0*IZ7oyMJ94lePxVg0 828w, https://miro.medium.com/v2/resize:fit:1100/0*IZ7oyMJ94lePxVg0 1100w, https://miro.medium.com/v2/resize:fit:1400/0*IZ7oyMJ94lePxVg0 1400w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 700px"/><img alt="" class="bh mv ob c" width="700" height="380" loading="lazy" role="presentation"/></picture></div></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">DeepDraper training & Inference Pipeline</em></figcaption></figure><h2 id="738c" class="oy oz gu bf pa pb pc dy pd pe pf ea pg ol ph pi pj op pk pl pm ot pn po pp ha bk">Final thoughts and Key Takeaways:</h2><p id="f87c" class="pw-post-body-paragraph oc od gu oe b hx pq og oh ia pr oj ok ol ps on oo op pt or os ot pu ov ow ox gn bk">Our team had an amazing three-day experience full of learning opportunities! Witnessing the growth of the computer vision community in India was nothing short of exhilarating. We were thrilled to see the groundbreaking research being done by these talented individuals, and we couldn’t wait to learn more.</p><p id="7592" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">The key points that I’d like to take away with me are:</p><ul class=""><li id="fa8c" class="oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox qf qg qh bk">It is better to use problem-specific priors to aid learning in your deep networks compared to blind input-to-output mapping.</li><li id="85af" class="oc od gu oe b hx qi og oh ia qj oj ok ol qk on oo op ql or os ot qm ov ow ox qf qg qh bk">Having in-depth knowledge of the latest developments in different problem statements can help in translating ideas across different domains of AI.</li><li id="6d78" class="oc od gu oe b hx qi og oh ia qj oj ok ol qk on oo op ql or os ot qm ov ow ox qf qg qh bk">Full stack expertise of tech often pays well in designing a powerful product.</li></ul><figure class="nr ns nt nu nv nw no np paragraph-image"><div class="no np rd"><picture><source srcSet="https://miro.medium.com/v2/resize:fit:640/format:webp/0*C4absgD21VLDN5jk 640w, https://miro.medium.com/v2/resize:fit:720/format:webp/0*C4absgD21VLDN5jk 720w, https://miro.medium.com/v2/resize:fit:750/format:webp/0*C4absgD21VLDN5jk 750w, https://miro.medium.com/v2/resize:fit:786/format:webp/0*C4absgD21VLDN5jk 786w, https://miro.medium.com/v2/resize:fit:828/format:webp/0*C4absgD21VLDN5jk 828w, https://miro.medium.com/v2/resize:fit:1100/format:webp/0*C4absgD21VLDN5jk 1100w, https://miro.medium.com/v2/resize:fit:1304/format:webp/0*C4absgD21VLDN5jk 1304w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 652px" type="image/webp"/><source data-testid="og" srcSet="https://miro.medium.com/v2/resize:fit:640/0*C4absgD21VLDN5jk 640w, https://miro.medium.com/v2/resize:fit:720/0*C4absgD21VLDN5jk 720w, https://miro.medium.com/v2/resize:fit:750/0*C4absgD21VLDN5jk 750w, https://miro.medium.com/v2/resize:fit:786/0*C4absgD21VLDN5jk 786w, https://miro.medium.com/v2/resize:fit:828/0*C4absgD21VLDN5jk 828w, https://miro.medium.com/v2/resize:fit:1100/0*C4absgD21VLDN5jk 1100w, https://miro.medium.com/v2/resize:fit:1304/0*C4absgD21VLDN5jk 1304w" sizes="(min-resolution: 4dppx) and (max-width: 700px) 50vw, (-webkit-min-device-pixel-ratio: 4) and (max-width: 700px) 50vw, (min-resolution: 3dppx) and (max-width: 700px) 67vw, (-webkit-min-device-pixel-ratio: 3) and (max-width: 700px) 65vw, (min-resolution: 2.5dppx) and (max-width: 700px) 80vw, (-webkit-min-device-pixel-ratio: 2.5) and (max-width: 700px) 80vw, (min-resolution: 2dppx) and (max-width: 700px) 100vw, (-webkit-min-device-pixel-ratio: 2) and (max-width: 700px) 100vw, 652px"/><img alt="" class="bh mv ob c" width="652" height="544" loading="lazy" role="presentation"/></picture></div><figcaption class="pw ff px no np py pz bf b bg z du"><em class="qa">The Fynd Research Team at IIT Gandhinagar (L-R): Bipin Gaikwad, Arnab Mishra, Prasanna Kumar, Shashank Vasisht, Vignesh Prajapati</em></figcaption></figure><p id="1e98" class="pw-post-body-paragraph oc od gu oe b hx of og oh ia oi oj ok ol om on oo op oq or os ot ou ov ow ox gn bk">It was incredibly inspiring to see the strides being made in this field, and we feel grateful to have been a part of it. We can’t wait to attend more conferences like this and hopefully even present the groundbreaking work we’re doing here at Fynd too!</p></div></div></div></div></section></div></div></article></div><div class="ab cb"><div class="ci bh fz ga gb gc"><div class="re rf ab jr"><div class="rg ab"><a class="rh ay am ao" href="https://medium.com/tag/icvgip-2022?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="ri fj cx rj ge rk rl bf b bg z bk rm">Icvgip 2022</div></a></div><div class="rg ab"><a class="rh ay am ao" href="https://medium.com/tag/computer-vision?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="ri fj cx rj ge rk rl bf b bg z bk rm">Computer Vision</div></a></div><div class="rg ab"><a class="rh ay am ao" href="https://medium.com/tag/image-processing?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="ri fj cx rj ge rk rl bf b bg z bk rm">Image Processing</div></a></div><div class="rg ab"><a class="rh ay am ao" href="https://medium.com/tag/computer-graphics?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="ri fj cx rj ge rk rl bf b bg z bk rm">Computer Graphics</div></a></div><div class="rg ab"><a class="rh ay am ao" href="https://medium.com/tag/machine-learning?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="ri fj cx rj ge rk rl bf b bg z bk rm">Machine Learning</div></a></div></div></div></div><div class="l"></div><footer class="rn ro rp rq rr ab q rs jb c"><div class="l ae"><div class="ab cb"><div class="ci bh fz ga gb gc"><div class="ab cp rt"><div class="ab q lm"><div class="ru l"><span class="l rv rw rx e d"><div class="ab q lm ln"><div class="pw-multi-vote-icon fj jv lo lp lq"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="footerClapButton" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fvote%2Ffynd-team%2Fbb3fb4c41bd8&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=---footer_actions--bb3fb4c41bd8---------------------clap_footer-----------" rel="noopener follow"><div><div class="bm" aria-hidden="false"><div class="lr ao ls lt lu lv am lw lx ly lq"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" aria-label="clap"><path fill-rule="evenodd" d="M11.37.828 12 3.282l.63-2.454zM13.916 3.953l1.523-2.112-1.184-.39zM8.589 1.84l1.522 2.112-.337-2.501zM18.523 18.92c-.86.86-1.75 1.246-2.62 1.33a6 6 0 0 0 .407-.372c2.388-2.389 2.86-4.951 1.399-7.623l-.912-1.603-.79-1.672c-.26-.56-.194-.98.203-1.288a.7.7 0 0 1 .546-.132c.283.046.546.231.728.5l2.363 4.157c.976 1.624 1.141 4.237-1.324 6.702m-10.999-.438L3.37 14.328a.828.828 0 0 1 .585-1.408.83.83 0 0 1 .585.242l2.158 2.157a.365.365 0 0 0 .516-.516l-2.157-2.158-1.449-1.449a.826.826 0 0 1 1.167-1.17l3.438 3.44a.363.363 0 0 0 .516 0 .364.364 0 0 0 0-.516L5.293 9.513l-.97-.97a.826.826 0 0 1 0-1.166.84.84 0 0 1 1.167 0l.97.968 3.437 3.436a.36.36 0 0 0 .517 0 .366.366 0 0 0 0-.516L6.977 7.83a.82.82 0 0 1-.241-.584.82.82 0 0 1 .824-.826c.219 0 .43.087.584.242l5.787 5.787a.366.366 0 0 0 .587-.415l-1.117-2.363c-.26-.56-.194-.98.204-1.289a.7.7 0 0 1 .546-.132c.283.046.545.232.727.501l2.193 3.86c1.302 2.38.883 4.59-1.277 6.75-1.156 1.156-2.602 1.627-4.19 1.367-1.418-.236-2.866-1.033-4.079-2.246M10.75 5.971l2.12 2.12c-.41.502-.465 1.17-.128 1.89l.22.465-3.523-3.523a.8.8 0 0 1-.097-.368c0-.22.086-.428.241-.584a.847.847 0 0 1 1.167 0m7.355 1.705c-.31-.461-.746-.758-1.23-.837a1.44 1.44 0 0 0-1.11.275c-.312.24-.505.543-.59.881a1.74 1.74 0 0 0-.906-.465 1.47 1.47 0 0 0-.82.106l-2.182-2.182a1.56 1.56 0 0 0-2.2 0 1.54 1.54 0 0 0-.396.701 1.56 1.56 0 0 0-2.21-.01 1.55 1.55 0 0 0-.416.753c-.624-.624-1.649-.624-2.237-.037a1.557 1.557 0 0 0 0 2.2c-.239.1-.501.238-.715.453a1.56 1.56 0 0 0 0 2.2l.516.515a1.556 1.556 0 0 0-.753 2.615L7.01 19c1.32 1.319 2.909 2.189 4.475 2.449q.482.08.971.08c.85 0 1.653-.198 2.393-.579.231.033.46.054.686.054 1.266 0 2.457-.52 3.505-1.567 2.763-2.763 2.552-5.734 1.439-7.586z" clip-rule="evenodd"></path></svg></div></div></div></a></span></div><div class="pw-multi-vote-count l lz ma mb mc md me mf"><p class="bf b dv z du"><span class="mg">--</span></p></div></div></span><span class="l h g f ry rz"><div class="ab q lm ln"><div class="pw-multi-vote-icon fj jv lo lp lq"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="footerClapButton" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fvote%2Ffynd-team%2Fbb3fb4c41bd8&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=---footer_actions--bb3fb4c41bd8---------------------clap_footer-----------" rel="noopener follow"><div><div class="bm" aria-hidden="false"><div class="lr ao ls lt lu lv am lw lx ly lq"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" aria-label="clap"><path fill-rule="evenodd" d="M11.37.828 12 3.282l.63-2.454zM13.916 3.953l1.523-2.112-1.184-.39zM8.589 1.84l1.522 2.112-.337-2.501zM18.523 18.92c-.86.86-1.75 1.246-2.62 1.33a6 6 0 0 0 .407-.372c2.388-2.389 2.86-4.951 1.399-7.623l-.912-1.603-.79-1.672c-.26-.56-.194-.98.203-1.288a.7.7 0 0 1 .546-.132c.283.046.546.231.728.5l2.363 4.157c.976 1.624 1.141 4.237-1.324 6.702m-10.999-.438L3.37 14.328a.828.828 0 0 1 .585-1.408.83.83 0 0 1 .585.242l2.158 2.157a.365.365 0 0 0 .516-.516l-2.157-2.158-1.449-1.449a.826.826 0 0 1 1.167-1.17l3.438 3.44a.363.363 0 0 0 .516 0 .364.364 0 0 0 0-.516L5.293 9.513l-.97-.97a.826.826 0 0 1 0-1.166.84.84 0 0 1 1.167 0l.97.968 3.437 3.436a.36.36 0 0 0 .517 0 .366.366 0 0 0 0-.516L6.977 7.83a.82.82 0 0 1-.241-.584.82.82 0 0 1 .824-.826c.219 0 .43.087.584.242l5.787 5.787a.366.366 0 0 0 .587-.415l-1.117-2.363c-.26-.56-.194-.98.204-1.289a.7.7 0 0 1 .546-.132c.283.046.545.232.727.501l2.193 3.86c1.302 2.38.883 4.59-1.277 6.75-1.156 1.156-2.602 1.627-4.19 1.367-1.418-.236-2.866-1.033-4.079-2.246M10.75 5.971l2.12 2.12c-.41.502-.465 1.17-.128 1.89l.22.465-3.523-3.523a.8.8 0 0 1-.097-.368c0-.22.086-.428.241-.584a.847.847 0 0 1 1.167 0m7.355 1.705c-.31-.461-.746-.758-1.23-.837a1.44 1.44 0 0 0-1.11.275c-.312.24-.505.543-.59.881a1.74 1.74 0 0 0-.906-.465 1.47 1.47 0 0 0-.82.106l-2.182-2.182a1.56 1.56 0 0 0-2.2 0 1.54 1.54 0 0 0-.396.701 1.56 1.56 0 0 0-2.21-.01 1.55 1.55 0 0 0-.416.753c-.624-.624-1.649-.624-2.237-.037a1.557 1.557 0 0 0 0 2.2c-.239.1-.501.238-.715.453a1.56 1.56 0 0 0 0 2.2l.516.515a1.556 1.556 0 0 0-.753 2.615L7.01 19c1.32 1.319 2.909 2.189 4.475 2.449q.482.08.971.08c.85 0 1.653-.198 2.393-.579.231.033.46.054.686.054 1.266 0 2.457-.52 3.505-1.567 2.763-2.763 2.552-5.734 1.439-7.586z" clip-rule="evenodd"></path></svg></div></div></div></a></span></div><div class="pw-multi-vote-count l lz ma mb mc md me mf"><p class="bf b dv z du"><span class="mg">--</span></p></div></div></span></div><div class="bq ab"><div><div class="bm" aria-hidden="false"><button class="ao lr mj mk ab q fk ml mm" aria-label="responses"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" viewBox="0 0 24 24" class="mi"><path d="M18.006 16.803c1.533-1.456 2.234-3.325 2.234-5.321C20.24 7.357 16.709 4 12.191 4S4 7.357 4 11.482c0 4.126 3.674 7.482 8.191 7.482.817 0 1.622-.111 2.393-.327.231.2.48.391.744.559 1.06.693 2.203 1.044 3.399 1.044.224-.008.4-.112.486-.287a.49.49 0 0 0-.042-.518c-.495-.67-.845-1.364-1.04-2.057a4 4 0 0 1-.125-.598zm-3.122 1.055-.067-.223-.315.096a8 8 0 0 1-2.311.338c-4.023 0-7.292-2.955-7.292-6.587 0-3.633 3.269-6.588 7.292-6.588 4.014 0 7.112 2.958 7.112 6.593 0 1.794-.608 3.469-2.027 4.72l-.195.168v.255c0 .056 0 .151.016.295.025.231.081.478.154.733.154.558.398 1.117.722 1.659a5.3 5.3 0 0 1-2.165-.845c-.276-.176-.714-.383-.941-.59z"></path></svg><p class="bf b bg z du"><span class="pw-responses-count mh mi">1</span></p></button></div></div></div></div><div class="ab q"><div class="sa l jo"><div><div class="bm" aria-hidden="false"><span><a class="af ag ah ai aj ak al am an ao ap aq ar as at" data-testid="footerBookmarkButton" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fbookmark%2Fp%2Fbb3fb4c41bd8&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&source=---footer_actions--bb3fb4c41bd8---------------------bookmark_footer-----------" rel="noopener follow"><svg xmlns="http://www.w3.org/2000/svg" width="25" height="25" fill="none" viewBox="0 0 25 25" class="du mo" aria-label="Add to list bookmark button"><path fill="currentColor" d="M18 2.5a.5.5 0 0 1 1 0V5h2.5a.5.5 0 0 1 0 1H19v2.5a.5.5 0 1 1-1 0V6h-2.5a.5.5 0 0 1 0-1H18zM7 7a1 1 0 0 1 1-1h3.5a.5.5 0 0 0 0-1H8a2 2 0 0 0-2 2v14a.5.5 0 0 0 .805.396L12.5 17l5.695 4.396A.5.5 0 0 0 19 21v-8.5a.5.5 0 0 0-1 0v7.485l-5.195-4.012a.5.5 0 0 0-.61 0L7 19.985z"></path></svg></a></span></div></div></div><div class="sa l jo"><div class="bm" aria-hidden="false" aria-describedby="postFooterSocialMenu" aria-labelledby="postFooterSocialMenu"><div><div class="bm" aria-hidden="false"><button aria-controls="postFooterSocialMenu" aria-expanded="false" aria-label="Share Post" data-testid="footerSocialShareButton" class="af fk ah ai aj ak al mw an ao ap ex mx my mm mz"><svg xmlns="http://www.w3.org/2000/svg" width="24" height="24" fill="none" viewBox="0 0 24 24"><path fill="currentColor" fill-rule="evenodd" d="M15.218 4.931a.4.4 0 0 1-.118.132l.012.006a.45.45 0 0 1-.292.074.5.5 0 0 1-.3-.13l-2.02-2.02v7.07c0 .28-.23.5-.5.5s-.5-.22-.5-.5v-7.04l-2 2a.45.45 0 0 1-.57.04h-.02a.4.4 0 0 1-.16-.3.4.4 0 0 1 .1-.32l2.8-2.8a.5.5 0 0 1 .7 0l2.8 2.79a.42.42 0 0 1 .068.498m-.106.138.008.004v-.01zM16 7.063h1.5a2 2 0 0 1 2 2v10a2 2 0 0 1-2 2h-11c-1.1 0-2-.9-2-2v-10a2 2 0 0 1 2-2H8a.5.5 0 0 1 .35.15.5.5 0 0 1 .15.35.5.5 0 0 1-.15.35.5.5 0 0 1-.35.15H6.4c-.5 0-.9.4-.9.9v10.2a.9.9 0 0 0 .9.9h11.2c.5 0 .9-.4.9-.9v-10.2c0-.5-.4-.9-.9-.9H16a.5.5 0 0 1 0-1" clip-rule="evenodd"></path></svg></button></div></div></div></div></div></div></div></div></div></footer><div class="sb sc sd se sf l"><div class="ab cb"><div class="ci bh fz ga gb gc"><div class="sg bh r sh"></div><div class="si l"><div class="ab sj sk sl jq jp"><div class="sm sn so sp sq sr ss st su sv ab cp"><div class="h k"><a href="https://blog.gofynd.com/?source=post_page---post_publication_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="fj ab"><img alt="Building Fynd" class="sw is it cx" src="https://miro.medium.com/v2/resize:fill:96:96/1*Q7qNEfm08Fj5NVUQFFIbjQ.png" width="48" height="48" loading="lazy"/><div class="sw l it is fs n fr sx"></div></div></a></div><div class="j i d"><a href="https://blog.gofynd.com/?source=post_page---post_publication_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="fj ab"><img alt="Building Fynd" class="sw sz sy cx" src="https://miro.medium.com/v2/resize:fill:128:128/1*Q7qNEfm08Fj5NVUQFFIbjQ.png" width="64" height="64" loading="lazy"/><div class="sw l sy sz fs n fr sx"></div></div></a></div><div class="j i d ta jo"><div class="ab"><span><a class="bf b bg z tb ri tc td te tf tg ev ew th ti tj fa fb fc fd bm fe ff" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fsubscribe%2Fcollection%2Ffynd-team&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&collection=Building+Fynd&collectionId=91d0019cb1ab&source=post_page---post_publication_info--bb3fb4c41bd8---------------------follow_profile-----------" rel="noopener follow">Follow</a></span></div></div></div><div class="ab co tk"><div class="tl tm tn to tp l"><a class="af ag ah aj ak al am an ao ap aq ar as at ab q" href="https://blog.gofynd.com/?source=post_page---post_publication_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><h2 class="pw-author-name bf tr ts tt tu tv tw tx ol pi pj op pl pm ot po pp bk"><span class="gn tq">Published in <!-- -->Building Fynd</span></h2></a><div class="rg ab ir"><div class="l jo"><span class="pw-follower-count bf b bg z du"><a class="af ag ah ai aj ak al am an ao ap aq ar jh" rel="noopener follow" href="/followers?source=post_page---post_publication_info--bb3fb4c41bd8--------------------------------">1.2K Followers</a></span></div><div class="bf b bg z du ab ju"><span class="ji l" aria-hidden="true"><span class="bf b bg z du">·</span></span><a class="af ag ah ai aj ak al am an ao ap aq ar jh" rel="noopener follow" href="/improving-erase-bg-with-synthetic-data-388471a24281?source=post_page---post_publication_info--bb3fb4c41bd8--------------------------------">Last published <span>Mar 8, 2024</span></a></div></div><div class="ty l"><p class="bf b bg z bk"><span class="gn">Latest from our product and engineering teams</span></p></div></div></div><div class="h k"><div class="ab"><span><a class="bf b bg z tb ri tc td te tf tg ev ew th ti tj fa fb fc fd bm fe ff" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fsubscribe%2Fcollection%2Ffynd-team&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&collection=Building+Fynd&collectionId=91d0019cb1ab&source=post_page---post_publication_info--bb3fb4c41bd8---------------------follow_profile-----------" rel="noopener follow">Follow</a></span></div></div></div></div><div class="ab sj sk sl jq jp"><div class="sm sn so sp sq sr ss st su sv ab cp"><div class="h k"><a tabindex="0" href="https://medium.com/@shashankvasisht_8994?source=post_page---post_author_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="l fj"><img alt="Shashank Vasisht" class="l fd by it is cx" src="https://miro.medium.com/v2/resize:fill:96:96/1*CI03_gV0KvR2a-WttD4uiw.jpeg" width="48" height="48" loading="lazy"/><div class="fr by l it is fs n ay sx"></div></div></a></div><div class="j i d"><a tabindex="0" href="https://medium.com/@shashankvasisht_8994?source=post_page---post_author_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><div class="l fj"><img alt="Shashank Vasisht" class="l fd by sy sz cx" src="https://miro.medium.com/v2/resize:fill:128:128/1*CI03_gV0KvR2a-WttD4uiw.jpeg" width="64" height="64" loading="lazy"/><div class="fr by l sy sz fs n ay sx"></div></div></a></div><div class="j i d ta jo"><div class="ab"><span><a class="bf b bg z tb ri tc td te tf tg ev ew th ti tj fa fb fc fd bm fe ff" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fsubscribe%2Fuser%2F6408aa1c1489&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=post_page-6408aa1c1489--post_author_info--bb3fb4c41bd8---------------------follow_profile-----------" rel="noopener follow">Follow</a></span></div></div></div><div class="ab co tk"><div class="tl tm tn to tp l"><a class="af ag ah aj ak al am an ao ap aq ar as at ab q" href="https://medium.com/@shashankvasisht_8994?source=post_page---post_author_info--bb3fb4c41bd8--------------------------------" rel="noopener follow"><h2 class="pw-author-name bf tr ts tt tu tv tw tx ol pi pj op pl pm ot po pp bk"><span class="gn tq">Written by <!-- -->Shashank Vasisht</span></h2></a><div class="rg ab ir"><div class="l jo"><span class="pw-follower-count bf b bg z du"><a class="af ag ah ai aj ak al am an ao ap aq ar jh" href="https://medium.com/@shashankvasisht_8994/followers?source=post_page---post_author_info--bb3fb4c41bd8--------------------------------" rel="noopener follow">14 Followers</a></span></div><div class="bf b bg z du ab ju"><span class="ji l" aria-hidden="true"><span class="bf b bg z du">·</span></span><a class="af ag ah ai aj ak al am an ao ap aq ar jh" href="https://medium.com/@shashankvasisht_8994/following?source=post_page---post_author_info--bb3fb4c41bd8--------------------------------" rel="noopener follow">2 Following</a></div></div><div class="ty l"><p class="bf b bg z bk"><span class="gn">Computer Vision Researcher at Fynd || Deep Learning || Machine Learning || ADAS || Medical Image Processing || AI solutions for Retail</span></p></div></div></div><div class="h k"><div class="ab"><span><a class="bf b bg z tb ri tc td te tf tg ev ew th ti tj fa fb fc fd bm fe ff" href="https://medium.com/m/signin?actionUrl=https%3A%2F%2Fmedium.com%2F_%2Fsubscribe%2Fuser%2F6408aa1c1489&operation=register&redirect=https%3A%2F%2Fblog.gofynd.com%2Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8&user=Shashank+Vasisht&userId=6408aa1c1489&source=post_page-6408aa1c1489--post_author_info--bb3fb4c41bd8---------------------follow_profile-----------" rel="noopener follow">Follow</a></span></div></div></div></div></div></div><div class="tz ua ub uc ud l"><div class="sg bh r tz ua ue uf ug"></div><div class="ab cb"><div class="ci bh fz ga gb gc"><div class="ab q cp"><h2 class="bf tr uh hz pd ui ic pg uj uk ul um un uo up uq ur bk">Responses (<!-- -->1<!-- -->)</h2><div class="ab us"><div><div class="bm" aria-hidden="false"><a class="ut uu" href="https://policy.medium.com/medium-rules-30e5502c4eb4?source=post_page---post_responses--bb3fb4c41bd8--------------------------------" rel="noopener follow" target="_blank"><svg xmlns="http://www.w3.org/2000/svg" width="25" height="25" viewBox="0 0 25 25"><path fill-rule="evenodd" d="M11.987 5.036a.754.754 0 0 1 .914-.01c.972.721 1.767 1.218 2.6 1.543.828.322 1.719.485 2.887.505a.755.755 0 0 1 .741.757c-.018 3.623-.43 6.256-1.449 8.21-1.034 1.984-2.662 3.209-4.966 4.083a.75.75 0 0 1-.537-.003c-2.243-.874-3.858-2.095-4.897-4.074-1.024-1.951-1.457-4.583-1.476-8.216a.755.755 0 0 1 .741-.757c1.195-.02 2.1-.182 2.923-.503.827-.322 1.6-.815 2.519-1.535m.468.903c-.897.69-1.717 1.21-2.623 1.564-.898.35-1.856.527-3.026.565.037 3.45.469 5.817 1.36 7.515.884 1.684 2.25 2.762 4.284 3.571 2.092-.81 3.465-1.89 4.344-3.575.886-1.698 1.299-4.065 1.334-7.512-1.149-.039-2.091-.217-2.99-.567-.906-.353-1.745-.873-2.683-1.561m-.009 9.155a2.672 2.672 0 1 0 0-5.344 2.672 2.672 0 0 0 0 5.344m0 1a3.672 3.672 0 1 0 0-7.344 3.672 3.672 0 0 0 0 7.344m-1.813-3.777.525-.526.916.917 1.623-1.625.526.526-2.149 2.152z" clip-rule="evenodd"></path></svg></a></div></div></div></div><div class="uv l"><button class="bf b bg z bk ri uw ux uy mo ml tg ev ew ex uz va vb fa vc vd ve vf vg fb fc fd bm fe ff">See all responses</button></div></div></div></div><div class="vh vi vj vk vl l bx"><div class="h k j"><div class="sg bh vm vn"></div><div class="ab cb"><div class="ci bh fz ga gb gc"><div class="vo ab lm jr"><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://help.medium.com/hc/en-us?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Help</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://medium.statuspage.io/?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Status</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://medium.com/about?autoplay=1&source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">About</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://medium.com/jobs-at-medium/work-at-medium-959d1a85284e?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Careers</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="pressinquiries@medium.com?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Press</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://blog.medium.com/?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Blog</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://policy.medium.com/medium-privacy-policy-f03bf92035c9?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Privacy</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://policy.medium.com/medium-terms-of-service-9db0094a1e0f?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Terms</p></a></div><div class="vp vq l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://speechify.com/medium?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Text to speech</p></a></div><div class="vp l"><a class="af ag ah ai aj ak al am an ao ap aq ar as at" href="https://medium.com/business?source=post_page-----bb3fb4c41bd8--------------------------------" rel="noopener follow"><p class="bf b dv z du">Teams</p></a></div></div></div></div></div></div></div></div></div></div><script>window.__BUILD_ID__="main-20241129-135346-5cf0f044cd"</script><script>window.__GRAPHQL_URI__ = "https://blog.gofynd.com/_/graphql"</script><script>window.__PRELOADED_STATE__ = {"algolia":{"queries":{}},"cache":{"experimentGroupSet":true,"reason":"This request is not using the cache middleware worker","group":"disabled","tags":["group-edgeCachePosts","post-bb3fb4c41bd8","user-6408aa1c1489","collection-91d0019cb1ab"],"serverVariantState":"","middlewareEnabled":false,"cacheStatus":"DYNAMIC","shouldUseCache":false,"vary":[],"lohpSummerUpsellEnabled":false,"publicationHierarchyEnabledWeb":false,"postBottomResponsesEnabled":false},"client":{"hydrated":false,"isUs":false,"isNativeMedium":false,"isSafariMobile":false,"isSafari":false,"isFirefox":false,"routingEntity":{"type":"COLLECTION","id":"91d0019cb1ab","explicit":true},"viewerIsBot":false},"debug":{"requestId":"6ecf3fec-e267-42f6-874c-8972a20ea08a","hybridDevServices":[],"originalSpanCarrier":{"traceparent":"00-a44085cb342c1c2897bfa4cb3c1de7dd-bddc90d2c62d095c-01"}},"multiVote":{"clapsPerPost":{}},"navigation":{"branch":{"show":null,"hasRendered":null,"blockedByCTA":false},"hideGoogleOneTap":false,"hasRenderedAlternateUserBanner":null,"currentLocation":"https:\u002F\u002Fblog.gofynd.com\u002Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8","host":"blog.gofynd.com","hostname":"blog.gofynd.com","referrer":"","hasSetReferrer":false,"susiModal":{"step":null,"operation":"register"},"postRead":false,"partnerProgram":{"selectedCountryCode":null},"queryString":"?source=collection_home---4------15-----------------------"},"config":{"nodeEnv":"production","version":"main-20241129-135346-5cf0f044cd","target":"production","productName":"Medium","publicUrl":"https:\u002F\u002Fcdn-client.medium.com\u002Flite","authDomain":"medium.com","authGoogleClientId":"216296035834-k1k6qe060s2tp2a2jam4ljdcms00sttg.apps.googleusercontent.com","favicon":"production","glyphUrl":"https:\u002F\u002Fglyph.medium.com","branchKey":"key_live_ofxXr2qTrrU9NqURK8ZwEhknBxiI6KBm","algolia":{"appId":"MQ57UUUQZ2","apiKeySearch":"394474ced050e3911ae2249ecc774921","indexPrefix":"medium_","host":"-dsn.algolia.net"},"recaptchaKey":"6Lfc37IUAAAAAKGGtC6rLS13R1Hrw_BqADfS1LRk","recaptcha3Key":"6Lf8R9wUAAAAABMI_85Wb8melS7Zj6ziuf99Yot5","recaptchaEnterpriseKeyId":"6Le-uGgpAAAAAPprRaokM8AKthQ9KNGdoxaGUvVp","datadog":{"applicationId":"6702d87d-a7e0-42fe-bbcb-95b469547ea0","clientToken":"pub853ea8d17ad6821d9f8f11861d23dfed","rumToken":"pubf9cc52896502b9413b68ba36fc0c7162","context":{"deployment":{"target":"production","tag":"main-20241129-135346-5cf0f044cd","commit":"5cf0f044cde04a296c7f1e11fd4877d75fdab011"}},"datacenter":"us"},"googleAnalyticsCode":"G-7JY7T788PK","googlePay":{"apiVersion":"2","apiVersionMinor":"0","merchantId":"BCR2DN6TV7EMTGBM","merchantName":"Medium","instanceMerchantId":"13685562959212738550"},"applePay":{"version":3},"signInWallCustomDomainCollectionIds":["3a8144eabfe3","336d898217ee","61061eb0c96b","138adf9c44c","819cc2aaeee0"],"mediumMastodonDomainName":"me.dm","mediumOwnedAndOperatedCollectionIds":["8a9336e5bb4","b7e45b22fec3","193b68bd4fba","8d6b8a439e32","54c98c43354d","3f6ecf56618","d944778ce714","92d2092dc598","ae2a65f35510","1285ba81cada","544c7006046e","fc8964313712","40187e704f1c","88d9857e584e","7b6769f2748b","bcc38c8f6edf","cef6983b292","cb8577c9149e","444d13b52878","713d7dbc99b0","ef8e90590e66","191186aaafa0","55760f21cdc5","9dc80918cc93","bdc4052bbdba","8ccfed20cbb2"],"tierOneDomains":["medium.com","thebolditalic.com","arcdigital.media","towardsdatascience.com","uxdesign.cc","codeburst.io","psiloveyou.xyz","writingcooperative.com","entrepreneurshandbook.co","prototypr.io","betterhumans.coach.me","theascent.pub"],"topicsToFollow":["d61cf867d93f","8a146bc21b28","1eca0103fff3","4d562ee63426","aef1078a3ef5","e15e46793f8d","6158eb913466","55f1c20aba7a","3d18b94f6858","4861fee224fd","63c6f1f93ee","1d98b3a9a871","decb52b64abf","ae5d4995e225","830cded25262"],"topicToTagMappings":{"accessibility":"accessibility","addiction":"addiction","android-development":"android-development","art":"art","artificial-intelligence":"artificial-intelligence","astrology":"astrology","basic-income":"basic-income","beauty":"beauty","biotech":"biotech","blockchain":"blockchain","books":"books","business":"business","cannabis":"cannabis","cities":"cities","climate-change":"climate-change","comics":"comics","coronavirus":"coronavirus","creativity":"creativity","cryptocurrency":"cryptocurrency","culture":"culture","cybersecurity":"cybersecurity","data-science":"data-science","design":"design","digital-life":"digital-life","disability":"disability","economy":"economy","education":"education","equality":"equality","family":"family","feminism":"feminism","fiction":"fiction","film":"film","fitness":"fitness","food":"food","freelancing":"freelancing","future":"future","gadgets":"gadgets","gaming":"gaming","gun-control":"gun-control","health":"health","history":"history","humor":"humor","immigration":"immigration","ios-development":"ios-development","javascript":"javascript","justice":"justice","language":"language","leadership":"leadership","lgbtqia":"lgbtqia","lifestyle":"lifestyle","machine-learning":"machine-learning","makers":"makers","marketing":"marketing","math":"math","media":"media","mental-health":"mental-health","mindfulness":"mindfulness","money":"money","music":"music","neuroscience":"neuroscience","nonfiction":"nonfiction","outdoors":"outdoors","parenting":"parenting","pets":"pets","philosophy":"philosophy","photography":"photography","podcasts":"podcast","poetry":"poetry","politics":"politics","privacy":"privacy","product-management":"product-management","productivity":"productivity","programming":"programming","psychedelics":"psychedelics","psychology":"psychology","race":"race","relationships":"relationships","religion":"religion","remote-work":"remote-work","san-francisco":"san-francisco","science":"science","self":"self","self-driving-cars":"self-driving-cars","sexuality":"sexuality","social-media":"social-media","society":"society","software-engineering":"software-engineering","space":"space","spirituality":"spirituality","sports":"sports","startups":"startup","style":"style","technology":"technology","transportation":"transportation","travel":"travel","true-crime":"true-crime","tv":"tv","ux":"ux","venture-capital":"venture-capital","visual-design":"visual-design","work":"work","world":"world","writing":"writing"},"defaultImages":{"avatar":{"imageId":"1*dmbNkD5D-u45r44go_cf0g.png","height":150,"width":150},"orgLogo":{"imageId":"7*V1_7XP4snlmqrc_0Njontw.png","height":110,"width":500},"postLogo":{"imageId":"bd978bb536350a710e8efb012513429cabdc4c28700604261aeda246d0f980b7","height":810,"width":1440},"postPreviewImage":{"imageId":"1*hn4v1tCaJy7cWMyb0bpNpQ.png","height":386,"width":579}},"collectionStructuredData":{"8d6b8a439e32":{"name":"Elemental","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fcdn-images-1.medium.com\u002Fmax\u002F980\u002F1*9ygdqoKprhwuTVKUM0DLPA@2x.png","width":980,"height":159}}},"3f6ecf56618":{"name":"Forge","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fcdn-images-1.medium.com\u002Fmax\u002F596\u002F1*uULpIlImcO5TDuBZ6lm7Lg@2x.png","width":596,"height":183}}},"ae2a65f35510":{"name":"GEN","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fmiro.medium.com\u002Fmax\u002F264\u002F1*RdVZMdvfV3YiZTw6mX7yWA.png","width":264,"height":140}}},"88d9857e584e":{"name":"LEVEL","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fmiro.medium.com\u002Fmax\u002F540\u002F1*JqYMhNX6KNNb2UlqGqO2WQ.png","width":540,"height":108}}},"7b6769f2748b":{"name":"Marker","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fcdn-images-1.medium.com\u002Fmax\u002F383\u002F1*haCUs0wF6TgOOvfoY-jEoQ@2x.png","width":383,"height":92}}},"444d13b52878":{"name":"OneZero","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fmiro.medium.com\u002Fmax\u002F540\u002F1*cw32fIqCbRWzwJaoQw6BUg.png","width":540,"height":123}}},"8ccfed20cbb2":{"name":"Zora","data":{"@type":"NewsMediaOrganization","ethicsPolicy":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Farticles\u002F360043290473","logo":{"@type":"ImageObject","url":"https:\u002F\u002Fmiro.medium.com\u002Fmax\u002F540\u002F1*tZUQqRcCCZDXjjiZ4bDvgQ.png","width":540,"height":106}}}},"embeddedPostIds":{"coronavirus":"cd3010f9d81f"},"sharedCdcMessaging":{"COVID_APPLICABLE_TAG_SLUGS":[],"COVID_APPLICABLE_TOPIC_NAMES":[],"COVID_APPLICABLE_TOPIC_NAMES_FOR_TOPIC_PAGE":[],"COVID_MESSAGES":{"tierA":{"text":"For more information on the novel coronavirus and Covid-19, visit cdc.gov.","markups":[{"start":66,"end":73,"href":"https:\u002F\u002Fwww.cdc.gov\u002Fcoronavirus\u002F2019-nCoV"}]},"tierB":{"text":"Anyone can publish on Medium per our Policies, but we don’t fact-check every story. For more info about the coronavirus, see cdc.gov.","markups":[{"start":37,"end":45,"href":"https:\u002F\u002Fhelp.medium.com\u002Fhc\u002Fen-us\u002Fcategories\u002F201931128-Policies-Safety"},{"start":125,"end":132,"href":"https:\u002F\u002Fwww.cdc.gov\u002Fcoronavirus\u002F2019-nCoV"}]},"paywall":{"text":"This article has been made free for everyone, thanks to Medium Members. For more information on the novel coronavirus and Covid-19, visit cdc.gov.","markups":[{"start":56,"end":70,"href":"https:\u002F\u002Fmedium.com\u002Fmembership"},{"start":138,"end":145,"href":"https:\u002F\u002Fwww.cdc.gov\u002Fcoronavirus\u002F2019-nCoV"}]},"unbound":{"text":"This article is free for everyone, thanks to Medium Members. For more information on the novel coronavirus and Covid-19, visit cdc.gov.","markups":[{"start":45,"end":59,"href":"https:\u002F\u002Fmedium.com\u002Fmembership"},{"start":127,"end":134,"href":"https:\u002F\u002Fwww.cdc.gov\u002Fcoronavirus\u002F2019-nCoV"}]}},"COVID_BANNER_POST_ID_OVERRIDE_WHITELIST":["3b31a67bff4a"]},"sharedVoteMessaging":{"TAGS":["politics","election-2020","government","us-politics","election","2020-presidential-race","trump","donald-trump","democrats","republicans","congress","republican-party","democratic-party","biden","joe-biden","maga"],"TOPICS":["politics","election"],"MESSAGE":{"text":"Find out more about the U.S. election results here.","markups":[{"start":46,"end":50,"href":"https:\u002F\u002Fcookpolitical.com\u002F2020-national-popular-vote-tracker"}]},"EXCLUDE_POSTS":["397ef29e3ca5"]},"embedPostRules":[],"recircOptions":{"v1":{"limit":3},"v2":{"limit":8}},"braintreeClientKey":"production_zjkj96jm_m56f8fqpf7ngnrd4","braintree":{"enabled":true,"merchantId":"m56f8fqpf7ngnrd4","merchantAccountId":{"usd":"AMediumCorporation_instant","eur":"amediumcorporation_EUR","cad":"amediumcorporation_CAD"},"publicKey":"ds2nn34bg2z7j5gd","braintreeEnvironment":"production","dashboardUrl":"https:\u002F\u002Fwww.braintreegateway.com\u002Fmerchants","gracePeriodDurationInDays":14,"mediumMembershipPlanId":{"monthly":"ce105f8c57a3","monthlyV2":"e8a5e126-792b-4ee6-8fba-d574c1b02fc5","monthlyWithTrial":"d5ee3dbe3db8","monthlyPremium":"fa741a9b47a2","yearly":"a40ad4a43185","yearlyV2":"3815d7d6-b8ca-4224-9b8c-182f9047866e","yearlyStaff":"d74fb811198a","yearlyWithTrial":"b3bc7350e5c7","yearlyPremium":"e21bd2c12166","monthlyOneYearFree":"e6c0637a-2bad-4171-ab4f-3c268633d83c","monthly25PercentOffFirstYear":"235ecc62-0cdb-49ae-9378-726cd21c504b","monthly20PercentOffFirstYear":"ba518864-9c13-4a99-91ca-411bf0cac756","monthly15PercentOffFirstYear":"594c029b-9f89-43d5-88f8-8173af4e070e","monthly10PercentOffFirstYear":"c6c7bc9a-40f2-4b51-8126-e28511d5bdb0","monthlyForStudents":"629ebe51-da7d-41fd-8293-34cd2f2030a8","yearlyOneYearFree":"78ba7be9-0d9f-4ece-aa3e-b54b826f2bf1","yearly25PercentOffFirstYear":"2dbb010d-bb8f-4eeb-ad5c-a08509f42d34","yearly20PercentOffFirstYear":"47565488-435b-47f8-bf93-40d5fbe0ebc8","yearly15PercentOffFirstYear":"8259809b-0881-47d9-acf7-6c001c7f720f","yearly10PercentOffFirstYear":"9dd694fb-96e1-472c-8d9e-3c868d5c1506","yearlyForStudents":"e29345ef-ab1c-4234-95c5-70e50fe6bc23","monthlyCad":"p52orjkaceei","yearlyCad":"h4q9g2up9ktt"},"braintreeDiscountId":{"oneMonthFree":"MONTHS_FREE_01","threeMonthsFree":"MONTHS_FREE_03","sixMonthsFree":"MONTHS_FREE_06","fiftyPercentOffOneYear":"FIFTY_PERCENT_OFF_ONE_YEAR"},"3DSecureVersion":"2","defaultCurrency":"usd","providerPlanIdCurrency":{"4ycw":"usd","rz3b":"usd","3kqm":"usd","jzw6":"usd","c2q2":"usd","nnsw":"usd","q8qw":"usd","d9y6":"usd","fx7w":"cad","nwf2":"cad"}},"paypalClientId":"AXj1G4fotC2GE8KzWX9mSxCH1wmPE3nJglf4Z2ig_amnhvlMVX87otaq58niAg9iuLktVNF_1WCMnN7v","paypal":{"host":"https:\u002F\u002Fapi.paypal.com:443","clientMode":"production","serverMode":"live","webhookId":"4G466076A0294510S","monthlyPlan":{"planId":"P-9WR0658853113943TMU5FDQA","name":"Medium Membership (Monthly) with setup fee","description":"Unlimited access to the best and brightest stories on Medium. Membership billed monthly."},"yearlyPlan":{"planId":"P-7N8963881P8875835MU5JOPQ","name":"Medium Membership (Annual) with setup fee","description":"Unlimited access to the best and brightest stories on Medium. Membership billed annually."},"oneYearGift":{"name":"Medium Membership (1 Year, Digital Gift Code)","description":"Unlimited access to the best and brightest stories on Medium. Gift codes can be redeemed at medium.com\u002Fredeem.","price":"50.00","currency":"USD","sku":"membership-gift-1-yr"},"oldMonthlyPlan":{"planId":"P-96U02458LM656772MJZUVH2Y","name":"Medium Membership (Monthly)","description":"Unlimited access to the best and brightest stories on Medium. Membership billed monthly."},"oldYearlyPlan":{"planId":"P-59P80963JF186412JJZU3SMI","name":"Medium Membership (Annual)","description":"Unlimited access to the best and brightest stories on Medium. Membership billed annually."},"monthlyPlanWithTrial":{"planId":"P-66C21969LR178604GJPVKUKY","name":"Medium Membership (Monthly) with setup fee","description":"Unlimited access to the best and brightest stories on Medium. Membership billed monthly."},"yearlyPlanWithTrial":{"planId":"P-6XW32684EX226940VKCT2MFA","name":"Medium Membership (Annual) with setup fee","description":"Unlimited access to the best and brightest stories on Medium. Membership billed annually."},"oldMonthlyPlanNoSetupFee":{"planId":"P-4N046520HR188054PCJC7LJI","name":"Medium Membership (Monthly)","description":"Unlimited access to the best and brightest stories on Medium. Membership billed monthly."},"oldYearlyPlanNoSetupFee":{"planId":"P-7A4913502Y5181304CJEJMXQ","name":"Medium Membership (Annual)","description":"Unlimited access to the best and brightest stories on Medium. Membership billed annually."},"sdkUrl":"https:\u002F\u002Fwww.paypal.com\u002Fsdk\u002Fjs"},"stripePublishableKey":"pk_live_7FReX44VnNIInZwrIIx6ghjl","log":{"json":true,"level":"info"},"imageUploadMaxSizeMb":25,"staffPicks":{"title":"Staff Picks","catalogId":"c7bc6e1ee00f"}},"session":{"xsrf":""}}</script><script>window.__APOLLO_STATE__ = {"ROOT_QUERY":{"__typename":"Query","variantFlags":[{"__typename":"VariantFlag","name":"enable_mastodon_avatar_upload","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_enable_lock_responses","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_syntax_highlight","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_branch_io","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_google_webhook","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"allow_test_auth","valueType":{"__typename":"VariantFlagString","value":"disallow"}},{"__typename":"VariantFlag","name":"enable_speechify_ios","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ios_dynamic_paywall_programming","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_recaptcha_enterprise","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"allow_access","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_sprig","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_enable_friend_links_postpage_banners","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lite_continue_this_thread","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_premium_tier_badge","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_friend_links_postpage_banners","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_group_gifting","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_billing_frequency_on_step2","valueType":{"__typename":"VariantFlagString","value":"control"}},{"__typename":"VariantFlag","name":"mobile_custom_app_icon","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_iceland_forced_android","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_new_stripe_customers","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_tipping_v0_android","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_rex_new_push_notification_endpoint","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_new_manage_membership_flow","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lite_archive_page","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_mastodon_for_members_username_selection","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_newsletter_lo_flow_custom_domains","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_pp_country_expansion","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ios_dynamic_paywall_aspiriational","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_tipping_v0_ios","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_update_explore_wtf","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_enable_friend_links_creation","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_paypal","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_webhook","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_topic_portals","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"coronavirus_topic_recirc","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_susi_redesign_ios","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"reengagement_notification_duration","valueType":{"__typename":"VariantFlagNumber","value":3}},{"__typename":"VariantFlag","name":"enable_tag_recs","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_editor_new_publishing_flow","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_remove_twitter_onboarding_step","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"signin_services","valueType":{"__typename":"VariantFlagString","value":"twitter,facebook,google,email,google-fastidv,google-one-tap,apple"}},{"__typename":"VariantFlag","name":"enable_starspace","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"can_send_tips_v0","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lite_response_markup","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_enable_verified_book_author","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_moc_load_processor_first_story","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_pp_v4","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_recommended_publishers_query","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_sharer_validate_post_share_key","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ios_easy_resubscribe","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_legacy_feed_in_iceland","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_speechify_widget","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_switch_plan_premium_tier","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"reader_fair_distribution_non_qp","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_google_one_tap","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_updated_pub_recs_ui","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_conversion_model_v2","valueType":{"__typename":"VariantFlagString","value":"group_2"}},{"__typename":"VariantFlag","name":"enable_pre_pp_v4","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_configure_pronouns","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lo_homepage","valueType":{"__typename":"VariantFlagString","value":"control"}},{"__typename":"VariantFlag","name":"allow_signup","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_android_dynamic_programming_paywall","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_conversion_ranker_v2","valueType":{"__typename":"VariantFlagString","value":"control"}},{"__typename":"VariantFlag","name":"enable_rex_aggregator_v2","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_seamless_social_sharing","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"signup_services","valueType":{"__typename":"VariantFlagString","value":"twitter,facebook,google,email,google-fastidv,google-one-tap,apple"}},{"__typename":"VariantFlag","name":"enable_apple_webhook","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_post_bottom_responses","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"available_annual_premium_plan","valueType":{"__typename":"VariantFlagString","value":"4a442ace1476"}},{"__typename":"VariantFlag","name":"available_monthly_plan","valueType":{"__typename":"VariantFlagString","value":"60e220181034"}},{"__typename":"VariantFlag","name":"enable_braintree_client","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_update_topic_portals_wtf","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_in_app_free_trial","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_android_verified_author","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"redefined_top_posts","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_friend_links_creation","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_automod","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_lists_v2","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"onboarding_tags_from_top_views","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_publication_hierarchy_web","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_marketing_emails","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_see_pronouns","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_intrinsic_automatic_actions","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_trial_membership","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"browsable_stream_config_bucket","valueType":{"__typename":"VariantFlagString","value":"curated-topics"}},{"__typename":"VariantFlag","name":"enable_aurora_pub_follower_page","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_google_pay","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"goliath_externalsearch_enable_comment_deindexation","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_display_paywall_after_onboarding","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lite_homepage","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_premium_tier","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_android_miro_v2","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_import","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"limit_user_follows","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ios_autorefresh","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_moc_load_processor_c","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ranker_v10","valueType":{"__typename":"VariantFlagString","value":"control"}},{"__typename":"VariantFlag","name":"glyph_font_set","valueType":{"__typename":"VariantFlagString","value":"m2-unbound-source-serif-pro"}},{"__typename":"VariantFlag","name":"available_annual_plan","valueType":{"__typename":"VariantFlagString","value":"2c754bcc2995"}},{"__typename":"VariantFlag","name":"enable_bayesian_average_pub_search","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_sharer_create_post_share_key","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_enable_home_post_menu","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"limit_post_referrers","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_author_cards","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"skip_fs_cache_user_vals","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_apple_pay","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_diversification_rex","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_iceland_nux","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_two_hour_refresh","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_susi_redesign_android","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"ios_social_share_sheet","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"num_post_bottom_responses_to_show","valueType":{"__typename":"VariantFlagString","value":"3"}},{"__typename":"VariantFlag","name":"enable_maim_the_meter","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_tick_landing_page","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_tribute_landing_page","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"price_smoke_test_yearly","valueType":{"__typename":"VariantFlagString","value":""}},{"__typename":"VariantFlag","name":"enable_footer_app_buttons","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_moc_load_processor_all_recs_surfaces","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_pill_based_home_feed","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_auto_follow_on_subscribe","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_boost_nia_v01","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ml_rank_rex_anno","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"available_monthly_premium_plan","valueType":{"__typename":"VariantFlagString","value":"12a660186432"}},{"__typename":"VariantFlag","name":"enable_abandoned_cart_promotion_email","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_apple_sign_in","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_rito_upstream_deadlines","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_verifications_service","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_bg_post_post","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_simplified_digest_v2_b","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_mastodon_for_members","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_cache_less_following_feed","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"textshots_userid","valueType":{"__typename":"VariantFlagString","value":""}},{"__typename":"VariantFlag","name":"enable_medium2_kbfd","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_creator_welcome_email","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_explicit_signals_updated_post_previews","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_recirc_model","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"price_smoke_test_monthly","valueType":{"__typename":"VariantFlagString","value":""}},{"__typename":"VariantFlag","name":"enable_post_bottom_responses_input","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_deprecate_legacy_providers_v3","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_ios_offline_reading","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_rating_prompt_stories_read_threshold","valueType":{"__typename":"VariantFlagNumber","value":2}},{"__typename":"VariantFlag","name":"enable_android_dynamic_aspirational_paywall","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_author_cards_byline","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_braintree_integration","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_eventstats_event_processing","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"disable_partner_program_enrollment","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_android_offline_reading","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"can_receive_tips_v0","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_rex_reading_history","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"rex_generator_max_candidates","valueType":{"__typename":"VariantFlagNumber","value":1000}},{"__typename":"VariantFlag","name":"enable_app_flirty_thirty","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_lite_server_upstream_deadlines","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_explicit_signals","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"android_enable_image_sharer","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_entities_to_follow_v2","valueType":{"__typename":"VariantFlagBoolean","value":true}},{"__typename":"VariantFlag","name":"enable_members_only_audio","valueType":{"__typename":"VariantFlagBoolean","value":true}}],"viewer":null,"collectionByDomainOrSlug({\"domainOrSlug\":\"blog.gofynd.com\"})":{"__ref":"Collection:91d0019cb1ab"},"postResult({\"id\":\"bb3fb4c41bd8\"})":{"__ref":"Post:bb3fb4c41bd8"}},"ImageMetadata:1*Q7qNEfm08Fj5NVUQFFIbjQ.png":{"__typename":"ImageMetadata","id":"1*Q7qNEfm08Fj5NVUQFFIbjQ.png"},"Collection:91d0019cb1ab":{"__typename":"Collection","id":"91d0019cb1ab","favicon":{"__ref":"ImageMetadata:1*Q7qNEfm08Fj5NVUQFFIbjQ.png"},"customStyleSheet":null,"colorPalette":{"__typename":"ColorPalette","highlightSpectrum":{"__typename":"ColorSpectrum","backgroundColor":"#FFFFFFFF","colorPoints":[{"__typename":"ColorPoint","color":"#FFEDF2FF","point":0},{"__typename":"ColorPoint","color":"#FFE9F0FF","point":0.1},{"__typename":"ColorPoint","color":"#FFE6EEFF","point":0.2},{"__typename":"ColorPoint","color":"#FFE2ECFF","point":0.3},{"__typename":"ColorPoint","color":"#FFDFEAFF","point":0.4},{"__typename":"ColorPoint","color":"#FFDBE8FF","point":0.5},{"__typename":"ColorPoint","color":"#FFD7E6FF","point":0.6},{"__typename":"ColorPoint","color":"#FFD4E4FF","point":0.7},{"__typename":"ColorPoint","color":"#FFD0E2FF","point":0.8},{"__typename":"ColorPoint","color":"#FFCCE0FF","point":0.9},{"__typename":"ColorPoint","color":"#FFC9DEFF","point":1}]},"defaultBackgroundSpectrum":{"__typename":"ColorSpectrum","backgroundColor":"#FFFFFFFF","colorPoints":[{"__typename":"ColorPoint","color":"#FF6980E5","point":0},{"__typename":"ColorPoint","color":"#FF6277D2","point":0.1},{"__typename":"ColorPoint","color":"#FF5C6EBF","point":0.2},{"__typename":"ColorPoint","color":"#FF5565AB","point":0.3},{"__typename":"ColorPoint","color":"#FF4D5C99","point":0.4},{"__typename":"ColorPoint","color":"#FF465286","point":0.5},{"__typename":"ColorPoint","color":"#FF3D4873","point":0.6},{"__typename":"ColorPoint","color":"#FF353E61","point":0.7},{"__typename":"ColorPoint","color":"#FF2B334E","point":0.8},{"__typename":"ColorPoint","color":"#FF21273B","point":0.9},{"__typename":"ColorPoint","color":"#FF161A28","point":1}]},"tintBackgroundSpectrum":{"__typename":"ColorSpectrum","backgroundColor":"#FF1D1E8D","colorPoints":[{"__typename":"ColorPoint","color":"#FF1D1E8D","point":0},{"__typename":"ColorPoint","color":"#FF33429C","point":0.1},{"__typename":"ColorPoint","color":"#FF4C5DAD","point":0.2},{"__typename":"ColorPoint","color":"#FF6475BC","point":0.3},{"__typename":"ColorPoint","color":"#FF7B8BCB","point":0.4},{"__typename":"ColorPoint","color":"#FF91A0D8","point":0.5},{"__typename":"ColorPoint","color":"#FFA7B4E5","point":0.6},{"__typename":"ColorPoint","color":"#FFBDC8F1","point":0.7},{"__typename":"ColorPoint","color":"#FFD2DAFD","point":0.8},{"__typename":"ColorPoint","color":"#FFE6EDFF","point":0.9},{"__typename":"ColorPoint","color":"#FFFBFFFF","point":1}]}},"domain":"blog.gofynd.com","slug":"fynd-team","googleAnalyticsId":null,"editors":[{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:28e1453bac1a"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:72c4d2ac98a3"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:6661e265c01b"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:d3d39995d033"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:c27aea54ce7c"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:ad8cce662919"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:3f65b855fbc1"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:e98874ee1d31"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:5f73658d78e7"}},{"__typename":"CollectionMastheadUserItem","user":{"__ref":"User:c788bca90362"}}],"name":"Building Fynd","avatar":{"__ref":"ImageMetadata:1*Q7qNEfm08Fj5NVUQFFIbjQ.png"},"description":"Latest from our product and engineering teams","subscriberCount":1229,"latestPostsConnection({\"paging\":{\"limit\":1}})":{"__typename":"PostConnection","posts":[{"__ref":"Post:388471a24281"}]},"viewerEdge":{"__ref":"CollectionViewerEdge:collectionId:91d0019cb1ab-viewerId:lo_7432eda3b414"},"twitterUsername":"lifeatfynd","facebookPageId":null,"logo":{"__ref":"ImageMetadata:1*qOeeVf-aNCFCtn2_qeSUFw.png"}},"User:28e1453bac1a":{"__typename":"User","id":"28e1453bac1a"},"User:72c4d2ac98a3":{"__typename":"User","id":"72c4d2ac98a3"},"User:6661e265c01b":{"__typename":"User","id":"6661e265c01b"},"User:d3d39995d033":{"__typename":"User","id":"d3d39995d033"},"User:c27aea54ce7c":{"__typename":"User","id":"c27aea54ce7c"},"User:ad8cce662919":{"__typename":"User","id":"ad8cce662919"},"User:3f65b855fbc1":{"__typename":"User","id":"3f65b855fbc1"},"User:e98874ee1d31":{"__typename":"User","id":"e98874ee1d31"},"User:5f73658d78e7":{"__typename":"User","id":"5f73658d78e7"},"User:c788bca90362":{"__typename":"User","id":"c788bca90362"},"User:5103184cd757":{"__typename":"User","id":"5103184cd757","customDomainState":null,"hasSubdomain":false,"username":"this.is.hemant.singh"},"Post:388471a24281":{"__typename":"Post","id":"388471a24281","firstPublishedAt":1709867087981,"creator":{"__ref":"User:5103184cd757"},"collection":{"__ref":"Collection:91d0019cb1ab"},"isSeries":false,"mediumUrl":"https:\u002F\u002Fblog.gofynd.com\u002Fimproving-erase-bg-with-synthetic-data-388471a24281","sequence":null,"uniqueSlug":"improving-erase-bg-with-synthetic-data-388471a24281"},"LinkedAccounts:6408aa1c1489":{"__typename":"LinkedAccounts","mastodon":null,"id":"6408aa1c1489"},"UserViewerEdge:userId:6408aa1c1489-viewerId:lo_7432eda3b414":{"__typename":"UserViewerEdge","id":"userId:6408aa1c1489-viewerId:lo_7432eda3b414","isFollowing":false,"isUser":false,"isMuting":false},"NewsletterV3:c8399db3990":{"__typename":"NewsletterV3","id":"c8399db3990","type":"NEWSLETTER_TYPE_AUTHOR","slug":"6408aa1c1489","name":"6408aa1c1489","collection":null,"user":{"__ref":"User:6408aa1c1489"}},"User:6408aa1c1489":{"__typename":"User","id":"6408aa1c1489","name":"Shashank Vasisht","username":"shashankvasisht_8994","newsletterV3":{"__ref":"NewsletterV3:c8399db3990"},"linkedAccounts":{"__ref":"LinkedAccounts:6408aa1c1489"},"isSuspended":false,"imageId":"1*CI03_gV0KvR2a-WttD4uiw.jpeg","mediumMemberAt":0,"verifications":{"__typename":"VerifiedInfo","isBookAuthor":false},"socialStats":{"__typename":"SocialStats","followerCount":14,"followingCount":2,"collectionFollowingCount":0},"customDomainState":null,"hasSubdomain":false,"bio":"Computer Vision Researcher at Fynd || Deep Learning || Machine Learning || ADAS || Medical Image Processing || AI solutions for Retail","isPartnerProgramEnrolled":false,"viewerEdge":{"__ref":"UserViewerEdge:userId:6408aa1c1489-viewerId:lo_7432eda3b414"},"viewerIsUser":false,"postSubscribeMembershipUpsellShownAt":0,"membership":null,"allowNotes":true,"twitterScreenName":""},"Topic:1eca0103fff3":{"__typename":"Topic","slug":"machine-learning","id":"1eca0103fff3","name":"Machine Learning"},"Paragraph:69cf10d652cf_0":{"__typename":"Paragraph","id":"69cf10d652cf_0","name":"2ea3","type":"H4","href":null,"layout":null,"metadata":null,"text":"Machine Learning","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_1":{"__typename":"Paragraph","id":"69cf10d652cf_1","name":"64eb","type":"H3","href":null,"layout":null,"metadata":null,"text":"Exploring the latest innovations in Computer Vision","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":51,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_2":{"__typename":"Paragraph","id":"69cf10d652cf_2","name":"fb83","type":"H4","href":null,"layout":null,"metadata":null,"text":"Insights from Fynd’s visit to The Indian Conference on Computer Vision, Graphics & Image Processing 2022","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*rJjyYQz-gKIc45dN":{"__typename":"ImageMetadata","id":"0*rJjyYQz-gKIc45dN","originalHeight":800,"originalWidth":1600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_3":{"__typename":"Paragraph","id":"69cf10d652cf_3","name":"5de2","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*rJjyYQz-gKIc45dN"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_4":{"__typename":"Paragraph","id":"69cf10d652cf_4","name":"a09b","type":"P","href":null,"layout":null,"metadata":null,"text":"The Indian Conference on Computer Vision, Graphics and Image Processing (ICVGIP) is the premier conference in computer vision, graphics, image processing, and related fields.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_5":{"__typename":"Paragraph","id":"69cf10d652cf_5","name":"4c7b","type":"P","href":null,"layout":null,"metadata":null,"text":"The ICVGIP 2022 took place at IIT Gandhinagar. The Computer Vision Research team at Fynd got a chance to attend! The 3-day event included exciting events like tutorials, paper presentations, industry sessions, plenary talks, and Vision India. Each day also featured poster presentations and demo sessions by independent researchers and industry members, offering opportunities for engaging discussions about their work.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_6":{"__typename":"Paragraph","id":"69cf10d652cf_6","name":"7816","type":"H4","href":null,"layout":null,"metadata":null,"text":"Learnings from Tutorial Sessions","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":32,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_7":{"__typename":"Paragraph","id":"69cf10d652cf_7","name":"cab1","type":"P","href":null,"layout":null,"metadata":null,"text":"Two tutorial sessions were conducted in parallel on Physics-based rendering in the service of computational imaging and Designing and Optimizing Computational Imaging Systems with End-to-End Learning.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_8":{"__typename":"Paragraph","id":"69cf10d652cf_8","name":"1435","type":"P","href":null,"layout":null,"metadata":null,"text":"The first was more inclined towards computer graphics and rendering while the other was about incorporating end-to-end deep learning into Imaging systems. Since the latter was closer to our field of interest, we chose to attend that.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_9":{"__typename":"Paragraph","id":"69cf10d652cf_9","name":"6876","type":"H4","href":null,"layout":null,"metadata":null,"text":"Designing and Optimising Computational Imaging Systems with End-to-End Learning","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":79,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_10":{"__typename":"Paragraph","id":"69cf10d652cf_10","name":"88c3","type":"P","href":null,"layout":null,"metadata":null,"text":"The speakers for this session were Dr Vivek Boominathan, Dr Evan Y. Peng and Dr Chris Metzler. Computational imaging systems combine optics and algorithms to perform imaging and computer vision tasks more effectively than conventional imaging systems. However, end-to-end learning has emerged as a new system design paradigm where both optics and algorithms are designed automatically using training data and machine learning.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":34,"end":55,"href":"https:\u002F\u002Fvivekboominathan.com\u002F","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":56,"end":72,"href":"https:\u002F\u002Fwww.eee.hku.hk\u002F~evanpeng\u002F","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":76,"end":93,"href":"https:\u002F\u002Fwww.cs.umd.edu\u002F~metzler\u002F","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"STRONG","start":35,"end":73,"href":null,"anchorType":null,"userId":null,"linkMetadata":null},{"__typename":"Markup","type":"STRONG","start":76,"end":95,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_11":{"__typename":"Paragraph","id":"69cf10d652cf_11","name":"539e","type":"P","href":null,"layout":null,"metadata":null,"text":"Advantages of end-to-end learning algorithms","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":44,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_12":{"__typename":"Paragraph","id":"69cf10d652cf_12","name":"51c5","type":"P","href":null,"layout":null,"metadata":null,"text":"This tutorial presents an end-to-end learning method that integrates optical models. Traditional optical lenses function by focusing light to a single point, mimicking human vision, and are commonly used in camera systems to capture visual information. However, this approach may not be optimal for all imaging tasks, such as monocular depth estimation and super-resolution. By using modified lenses or incorporating additional information with standard camera lenses, these computer vision tasks can benefit greatly.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_13":{"__typename":"Paragraph","id":"69cf10d652cf_13","name":"3904","type":"P","href":null,"layout":null,"metadata":null,"text":"Typically, in camera systems, the optical design is established first, and then the image processing algorithm’s parameters are adjusted to achieve high-quality image reproduction. In contrast to this sequential design approach, the authors consider joint optimising of an optical system (such as the physical shape of the lens) simultaneously with the reconstruction algorithm’s parameters. They developed a fully-differentiable simulation model that optimizes both sets of parameters to minimize the deviation between the true and reconstructed image.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_14":{"__typename":"Paragraph","id":"69cf10d652cf_14","name":"272d","type":"P","href":null,"layout":null,"metadata":null,"text":"They published their ideas in a paper called End-to-end Optimization of Optics and Image Processing for Achromatic Extended Depth of Field and Super-resolution Imaging.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":44,"end":167,"href":"https:\u002F\u002Fweb.stanford.edu\u002F~boyd\u002Fpapers\u002Fpdf\u002Fend_to_end_opt_optics.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_15":{"__typename":"Paragraph","id":"69cf10d652cf_15","name":"8bef","type":"P","href":null,"layout":null,"metadata":null,"text":"The broad idea is to use a simulation module, which can simulate point spread functions (PSF) of differently shaped lenses and learn the best-tuned parameters for the optics model (hence the best-shaped lens) for this particular task which was on super-resolution.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*qi793U05hDQBPt4U":{"__typename":"ImageMetadata","id":"0*qi793U05hDQBPt4U","originalHeight":604,"originalWidth":1600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_16":{"__typename":"Paragraph","id":"69cf10d652cf_16","name":"0060","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*qi793U05hDQBPt4U"},"text":"Differentiable PSF Simulation","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":29,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_17":{"__typename":"Paragraph","id":"69cf10d652cf_17","name":"7481","type":"P","href":null,"layout":null,"metadata":null,"text":"Applications of end-to-end learning","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":35,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_18":{"__typename":"Paragraph","id":"69cf10d652cf_18","name":"e5d9","type":"P","href":null,"layout":null,"metadata":null,"text":"The same idea can be applied to other applications like monocular depth estimation which has been demonstrated in the paper Depth from Defocus with Learned Optics for Imaging and Occlusion-aware Depth Estimation. The authors jointly optimize a deep CNN (a U-Net-like model) along with a fully differentiable optics model to produce high-quality depth maps using a single camera.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":123,"end":211,"href":"http:\u002F\u002Fwww.computationalimaging.org\u002Fwp-content\u002Fuploads\u002F2021\u002F04\u002FDeepDfD_ICCP2021.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*JE03Ta72Fk-4GyN8":{"__typename":"ImageMetadata","id":"0*JE03Ta72Fk-4GyN8","originalHeight":354,"originalWidth":1600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_19":{"__typename":"Paragraph","id":"69cf10d652cf_19","name":"ac00","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*JE03Ta72Fk-4GyN8"},"text":"Deep CNN Model","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":14,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_20":{"__typename":"Paragraph","id":"69cf10d652cf_20","name":"a725","type":"P","href":null,"layout":null,"metadata":null,"text":"Carrying this idea forward, the authors went on to showcase an innovative Lens-less camera where they proposed to eliminate the need for lenses in cameras and use a very thin phase mask instead. Phase masks are essentially transparent materials with different heights at different locations.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":73,"end":90,"href":"https:\u002F\u002Fieeexplore.ieee.org\u002Fielaam\u002F34\u002F9108332\u002F9076617-aam.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_21":{"__typename":"Paragraph","id":"69cf10d652cf_21","name":"5cd6","type":"P","href":null,"layout":null,"metadata":null,"text":"This causes phase modulation of the incoming wavefront and resultant wave interference produces the PSF at the sensor plane. Their proposed phase-mask framework takes the input of the target PSF and the desired device geometry (which as stated above can be learnt using a fully differentiable simulated optics model) and outputs an optimized phase-mask design.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*LY2S60fQoxZXDFrd":{"__typename":"ImageMetadata","id":"0*LY2S60fQoxZXDFrd","originalHeight":250,"originalWidth":688,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_22":{"__typename":"Paragraph","id":"69cf10d652cf_22","name":"8749","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*LY2S60fQoxZXDFrd"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*2cwccNQDVRvbwMZ5":{"__typename":"ImageMetadata","id":"0*2cwccNQDVRvbwMZ5","originalHeight":250,"originalWidth":688,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_23":{"__typename":"Paragraph","id":"69cf10d652cf_23","name":"5cb0","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*2cwccNQDVRvbwMZ5"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_24":{"__typename":"Paragraph","id":"69cf10d652cf_24","name":"585f","type":"H4","href":null,"layout":null,"metadata":null,"text":"Using FlatNet to enhance the output quality of phase masks","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":58,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_25":{"__typename":"Paragraph","id":"69cf10d652cf_25","name":"6fd6","type":"P","href":null,"layout":null,"metadata":null,"text":"Most of the phase mask output cannot be interpreted by humans. In another proposed paper called FlatNet: Towards Photorealistic Scene Reconstruction from Lensless Measurements, the authors first train a model to learn to invert the forward operation of the lensless camera model. This allows them to obtain an intermediate representation with local structures intact.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":95,"end":175,"href":"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2010.15440.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_26":{"__typename":"Paragraph","id":"69cf10d652cf_26","name":"7f6f","type":"P","href":null,"layout":null,"metadata":null,"text":"Once they obtain the output of the trainable inversion stage, which is of the same dimension as that of the natural image they want to recover, they use a fully convolutional network to map it to the perceptually enhanced image. They choose a U-Net to map the intermediate reconstruction to the final perceptually enhanced image.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*SJGGMmXq2jg7CS-I":{"__typename":"ImageMetadata","id":"0*SJGGMmXq2jg7CS-I","originalHeight":298,"originalWidth":1422,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_27":{"__typename":"Paragraph","id":"69cf10d652cf_27","name":"69ca","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*SJGGMmXq2jg7CS-I"},"text":"Reconstruction Of Scenes From Lensless Cameras","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":46,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*0r1l563RyXrBur5f":{"__typename":"ImageMetadata","id":"0*0r1l563RyXrBur5f","originalHeight":732,"originalWidth":1478,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_28":{"__typename":"Paragraph","id":"69cf10d652cf_28","name":"f5c8","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*0r1l563RyXrBur5f"},"text":"Architecture of the Flatnet","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":27,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_29":{"__typename":"Paragraph","id":"69cf10d652cf_29","name":"edba","type":"H4","href":null,"layout":null,"metadata":null,"text":"Key Takeaways from Plenary Talks","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_30":{"__typename":"Paragraph","id":"69cf10d652cf_30","name":"baad","type":"P","href":null,"layout":null,"metadata":null,"text":"The plenary talks were led by top-notch researchers and held during the final two days of the conference. The talks covered a diverse range of topics, including multi-sensory perception, efficient networks for graphics and rendering, transformers etc. offering attendees invaluable insights into the latest trends and advancements in the field.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_31":{"__typename":"Paragraph","id":"69cf10d652cf_31","name":"7721","type":"H4","href":null,"layout":null,"metadata":null,"text":"Instant NGP: Neural Networks in High-Performance Graphics","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":57,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_32":{"__typename":"Paragraph","id":"69cf10d652cf_32","name":"9259","type":"P","href":null,"layout":null,"metadata":null,"text":"This session was hosted by Dr Thomas Müller, the principal research scientist at NVIDIA. The talk was a case study on how the research team at NVIDIA was able to successfully train a Neural Radiance field (NeRF) model in a matter of minutes or even less. They call it the Instant NGP (Instant Neural Graphics Primitive). But let’s back up a bit and understand what NeRF is first.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":26,"end":43,"href":"https:\u002F\u002Fresearch.nvidia.com\u002Fperson\u002Fthomas-muller","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":182,"end":211,"href":"https:\u002F\u002Fwww.matthewtancik.com\u002Fnerf","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":271,"end":319,"href":"https:\u002F\u002Fgithub.com\u002FNVlabs\u002Finstant-ngp","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_33":{"__typename":"Paragraph","id":"69cf10d652cf_33","name":"f1c0","type":"P","href":null,"layout":null,"metadata":null,"text":"A neural radiance field (NeRF) is a fully-connected neural network that can generate novel views of complex 3D scenes, based on a partial set of 2D images. It is trained to use a rendering loss to reproduce input views of a scene. It works by taking input images representing a scene and interpolating between them to render one complete scene. NeRF is a highly effective way to generate images for synthetic data. A NeRF network is trained to map directly from viewing direction and spatial location (5D input) to opacity and colour (4D output), using volume rendering to render new views. You can read more about it in their paper.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":1,"end":23,"href":"https:\u002F\u002Farxiv.org\u002Fabs\u002F2003.08934","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":626,"end":632,"href":"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2003.08934.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*UPUMf41JiMnBl-Lw":{"__typename":"ImageMetadata","id":"0*UPUMf41JiMnBl-Lw","originalHeight":225,"originalWidth":700,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_34":{"__typename":"Paragraph","id":"69cf10d652cf_34","name":"cae2","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*UPUMf41JiMnBl-Lw"},"text":"NeRF forward pass","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":17,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_35":{"__typename":"Paragraph","id":"69cf10d652cf_35","name":"7fce","type":"P","href":null,"layout":null,"metadata":null,"text":"NeRF is a computationally-intensive algorithm, and rendering complex scenes can take hours or even days. Despite their reputation for being computationally expensive, neural networks can be trained and run efficiently for high-performance tasks. With the use of the appropriate data structures and algorithms, neural networks can run in the inner loops of real-time renderers and 3D reconstruction, resulting in an “instant NeRF.” Details about this approach can be found in their paper titled Instant Neural Graphics Primitives with a Multiresolution Hash Encoding.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":494,"end":566,"href":"https:\u002F\u002Fnvlabs.github.io\u002Finstant-ngp\u002Fassets\u002Fmueller2022instant.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_36":{"__typename":"Paragraph","id":"69cf10d652cf_36","name":"2493","type":"P","href":null,"layout":null,"metadata":null,"text":"The speaker credits their success to the three pillars of Neural High-Performance Graphics which are","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_37":{"__typename":"Paragraph","id":"69cf10d652cf_37","name":"5a2a","type":"ULI","href":null,"layout":null,"metadata":null,"text":"Small Neural Networks","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_38":{"__typename":"Paragraph","id":"69cf10d652cf_38","name":"f899","type":"ULI","href":null,"layout":null,"metadata":null,"text":"Hybrid Data Structures","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_39":{"__typename":"Paragraph","id":"69cf10d652cf_39","name":"b468","type":"ULI","href":null,"layout":null,"metadata":null,"text":"Task Specific GPU implementations","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*6PqjLsysfFrGj9ty":{"__typename":"ImageMetadata","id":"0*6PqjLsysfFrGj9ty","originalHeight":769,"originalWidth":1280,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_40":{"__typename":"Paragraph","id":"69cf10d652cf_40","name":"ccb9","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*6PqjLsysfFrGj9ty"},"text":"Pillars of Neural High-Performance Graphics","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":43,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_41":{"__typename":"Paragraph","id":"69cf10d652cf_41","name":"45e0","type":"P","href":null,"layout":null,"metadata":null,"text":"The importance of special priors","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":32,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_42":{"__typename":"Paragraph","id":"69cf10d652cf_42","name":"09ce","type":"P","href":null,"layout":null,"metadata":null,"text":"Smaller and more efficient neural networks can significantly reduce computing time, but this can sometimes compromise the accuracy and quality of the output. To address this issue, the use of special priors, such as positional encodings, can help produce similar results when generating a 3D scene with a smaller NeRF model than the original larger model. The speakers argued that without the input positional encodings, this would not have been possible and thus emphasize the importance of smaller networks aided with priors. Read more about it here.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":546,"end":551,"href":"https:\u002F\u002Fbmild.github.io\u002Ffourfeat\u002Findex.html","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*EfDGGWcqHhT3j96m":{"__typename":"ImageMetadata","id":"0*EfDGGWcqHhT3j96m","originalHeight":721,"originalWidth":1280,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_43":{"__typename":"Paragraph","id":"69cf10d652cf_43","name":"86e5","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*EfDGGWcqHhT3j96m"},"text":"The Effectiveness of Input Encoding","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":35,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*TQo-ryaCR2ju0gdz":{"__typename":"ImageMetadata","id":"0*TQo-ryaCR2ju0gdz","originalHeight":325,"originalWidth":600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_44":{"__typename":"Paragraph","id":"69cf10d652cf_44","name":"c458","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*TQo-ryaCR2ju0gdz"},"text":"Fourier Feature Input Results vs. Standard Input Results","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":56,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_45":{"__typename":"Paragraph","id":"69cf10d652cf_45","name":"5f0e","type":"P","href":null,"layout":null,"metadata":null,"text":"Usually, a lot of time is wasted in I\u002FO read-and-write operations. The authors state that to improve speed it is necessary to modify the existing data structures to something specific to the task. For the 2D to 3D scene projection, they designed Multiresolution Hash Encoding.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*Fu3-9-725acwQyFs":{"__typename":"ImageMetadata","id":"0*Fu3-9-725acwQyFs","originalHeight":696,"originalWidth":1280,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_46":{"__typename":"Paragraph","id":"69cf10d652cf_46","name":"be5a","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*Fu3-9-725acwQyFs"},"text":"Multiresolution Hash Encoding","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":29,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_47":{"__typename":"Paragraph","id":"69cf10d652cf_47","name":"1b40","type":"P","href":null,"layout":null,"metadata":null,"text":"Finally, they stated that the current deep learning frameworks are not optimal enough to exploit the speed of GPUs to the fullest. They instead proposed a new framework to train your Neural networks called Tiny Cuda. It is a small, self-contained framework for training and querying neural networks. It contains a lightning-fast “fully fused” multi-layer perceptron (paper), a versatile multiresolution hash encoding (paper), as well as support for various other input encodings, losses, and optimizers.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":205,"end":215,"href":"https:\u002F\u002Fgithub.com\u002FNVlabs\u002Ftiny-cuda-nn","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":328,"end":365,"href":"https:\u002F\u002Fraw.githubusercontent.com\u002FNVlabs\u002Ftiny-cuda-nn\u002Fmaster\u002Fdata\u002Freadme\u002Ffully-fused-mlp-diagram.png","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":367,"end":372,"href":"https:\u002F\u002Ftom94.net\u002Fdata\u002Fpublications\u002Fmueller21realtime\u002Fmueller21realtime.pdf","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":386,"end":416,"href":"https:\u002F\u002Fraw.githubusercontent.com\u002FNVlabs\u002Ftiny-cuda-nn\u002Fmaster\u002Fdata\u002Freadme\u002Fmultiresolution-hash-encoding-diagram.png","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":418,"end":423,"href":"https:\u002F\u002Fnvlabs.github.io\u002Finstant-ngp\u002Fassets\u002Fmueller2022instant.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_48":{"__typename":"Paragraph","id":"69cf10d652cf_48","name":"ae59","type":"H4","href":null,"layout":null,"metadata":null,"text":"Strong Interpretable Priors Are All We Need","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_49":{"__typename":"Paragraph","id":"69cf10d652cf_49","name":"be1d","type":"P","href":null,"layout":null,"metadata":null,"text":"This session was led by Dr Tali Dekel, a research scientist at Google. Computer vision has recently made exciting progress, with new architectures and self-supervised learning paradigms rapidly improving. As computing power increases, models scale in size and training data, resulting in “foundation models” — billion-parameter neural networks trained in a self-supervised manner on massive amounts of unlabelled imagery.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":23,"end":37,"href":"https:\u002F\u002Fwww.weizmann.ac.il\u002Fmath\u002Fdekel\u002Fhome","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_50":{"__typename":"Paragraph","id":"69cf10d652cf_50","name":"12e9","type":"P","href":null,"layout":null,"metadata":null,"text":"Such models learn extraordinary priors about our visual world, as evident by their breakthrough results in a plethora of visual inference and synthesis tasks. Nevertheless, their knowledge is buried and hidden in the vast space of the network’s weights.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_51":{"__typename":"Paragraph","id":"69cf10d652cf_51","name":"fb30","type":"P","href":null,"layout":null,"metadata":null,"text":"The speaker presented a series of works that aim to investigate the internal representations learned by large-scale models. By studying their priors and utilizing them in classical and new visual tasks, the research covers co-segmenting two images into coherent object parts and using text to modify the appearance of moving objects in real-world videos.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_52":{"__typename":"Paragraph","id":"69cf10d652cf_52","name":"367e","type":"P","href":null,"layout":null,"metadata":null,"text":"There has been a constant evolution in visual descriptors used for Computer Vision tasks. Starting from the early hand-crafted features (SIFT, HOG, SURF, ORB), people eventually moved towards Deep CNN-based features with the rise of the Deep Learning era. However, with the latest developments in the field of Transformers, specifically Vision Transformers, is it time to move towards Deep ViT-based Features?","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*S2SaV_2DRT753seQ":{"__typename":"ImageMetadata","id":"0*S2SaV_2DRT753seQ","originalHeight":656,"originalWidth":1120,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_53":{"__typename":"Paragraph","id":"69cf10d652cf_53","name":"83a9","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*S2SaV_2DRT753seQ"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_54":{"__typename":"Paragraph","id":"69cf10d652cf_54","name":"8884","type":"P","href":null,"layout":null,"metadata":null,"text":"Exciting innovations in AI: Self-Supervised Learning & Transformers","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":67,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_55":{"__typename":"Paragraph","id":"69cf10d652cf_55","name":"c2b2","type":"P","href":null,"layout":null,"metadata":null,"text":"The speaker mentioned how many of the most exciting new AI breakthroughs have come from two recent innovations: self-supervised learning, which allows machines to learn from random, unlabelled examples; and Transformers, which enable AI models to selectively focus on certain parts of their input and thus reason more effectively. A recent work called Self-Supervised ViT — DINO is a great example of this. Interestingly, the acronym DINO comes from self-distillation with no labels.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":111,"end":136,"href":"https:\u002F\u002Fai.facebook.com\u002Fblog\u002Fself-supervised-learning-the-dark-matter-of-intelligence\u002F","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":206,"end":219,"href":"https:\u002F\u002Fai.facebook.com\u002Fblog\u002Froberta-an-optimized-method-for-pretraining-self-supervised-nlp-systems\u002F","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"A","start":351,"end":378,"href":"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2104.14294.pdf","anchorType":"LINK","userId":null,"linkMetadata":null},{"__typename":"Markup","type":"STRONG","start":455,"end":457,"href":null,"anchorType":null,"userId":null,"linkMetadata":null},{"__typename":"Markup","type":"STRONG","start":473,"end":475,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*wOJJWyg0kIUkynp8":{"__typename":"ImageMetadata","id":"0*wOJJWyg0kIUkynp8","originalHeight":534,"originalWidth":949,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_56":{"__typename":"Paragraph","id":"69cf10d652cf_56","name":"7269","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*wOJJWyg0kIUkynp8"},"text":"Training ViT with the DINO algorithm","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":36,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_57":{"__typename":"Paragraph","id":"69cf10d652cf_57","name":"eae4","type":"P","href":null,"layout":null,"metadata":null,"text":"By training ViT with the DINO algorithm, the authors observed that the model automatically learns an interpretable representation and separates the main object from the background clutter. It learns to segment objects without any human-generated annotation or any form of dedicated dense pixel-level loss.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_58":{"__typename":"Paragraph","id":"69cf10d652cf_58","name":"ee20","type":"P","href":null,"layout":null,"metadata":null,"text":"The core component of Vision Transformers is self-attention layers. In this model, each spatial location builds its representation by “attending” to the other locations. That way, by “looking” at other, potentially distant pieces of the image, the network builds a rich, high-level understanding of the scene. When visualizing the local attention maps in the network, it is apparent that they correspond to coherent semantic regions in the image.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*UhvWx9FQcNukxcDW":{"__typename":"ImageMetadata","id":"0*UhvWx9FQcNukxcDW","originalHeight":640,"originalWidth":1109,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_59":{"__typename":"Paragraph","id":"69cf10d652cf_59","name":"2253","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*UhvWx9FQcNukxcDW"},"text":"DINO Attention Heads","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":20,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_60":{"__typename":"Paragraph","id":"69cf10d652cf_60","name":"525d","type":"P","href":null,"layout":null,"metadata":null,"text":"How does DINO work?","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":19,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_61":{"__typename":"Paragraph","id":"69cf10d652cf_61","name":"71c4","type":"P","href":null,"layout":null,"metadata":null,"text":"DINO works by interpreting self-supervision as a special case of self-distillation, where no labels are used at all. It trains a student network by simply matching the output of a teacher network over different views of the same image.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_62":{"__typename":"Paragraph","id":"69cf10d652cf_62","name":"d236","type":"P","href":null,"layout":null,"metadata":null,"text":"The authors of this paper identified two components from previous self-supervised approaches that are particularly important for strong performance on ViT, the momentum teacher and multi-crop training, and integrated them into their framework.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_63":{"__typename":"Paragraph","id":"69cf10d652cf_63","name":"273d","type":"P","href":null,"layout":null,"metadata":null,"text":"In the image below you can see the difference between the feature map representations of both supervised and self-supervised variants of DINO ViT and ResNet. It is visible that deeper layers of Self-supervised DINO ViT produce more semantically coherent features and can even identify similar objects.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*NUiAMYgs7eolUiwm":{"__typename":"ImageMetadata","id":"0*NUiAMYgs7eolUiwm","originalHeight":671,"originalWidth":1127,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_64":{"__typename":"Paragraph","id":"69cf10d652cf_64","name":"ad7a","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*NUiAMYgs7eolUiwm"},"text":"Feature maps of Supervised and Self-supervised DINO ViT & ResNet","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":64,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_65":{"__typename":"Paragraph","id":"69cf10d652cf_65","name":"681a","type":"P","href":null,"layout":null,"metadata":null,"text":"These DINO ViT features can be used for a plethora of applications such as Zero-shot Co-segmentation and Part-Cosegmentation. In all cases, lightweight methodologies are designed, that leverage the universal knowledge learned by large-scale models through new visual descriptors and perceptual losses. The methods are “zero-shot’’. They require no training data and are self-supervised — requiring no manual labels and thus can be applied across different domains and tasks for which training data is scarce.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*0Y_osIMD1Q24NxgY":{"__typename":"ImageMetadata","id":"0*0Y_osIMD1Q24NxgY","originalHeight":575,"originalWidth":981,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_66":{"__typename":"Paragraph","id":"69cf10d652cf_66","name":"55db","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*0Y_osIMD1Q24NxgY"},"text":"Zero-shot Co-segmentation and Part-Cosegmentation","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":49,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_67":{"__typename":"Paragraph","id":"69cf10d652cf_67","name":"99eb","type":"H4","href":null,"layout":null,"metadata":null,"text":"Insights from Paper Presentations:","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_68":{"__typename":"Paragraph","id":"69cf10d652cf_68","name":"f9d0","type":"H4","href":null,"layout":null,"metadata":null,"text":"FLOAT: Factorized Learning of Object Attributes for Improved Multi-object Multi-part Scene Parsing","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_69":{"__typename":"Paragraph","id":"69cf10d652cf_69","name":"a633","type":"P","href":null,"layout":null,"metadata":null,"text":"Multi-object multi-part scene parsing is a challenging task which requires detecting multiple object classes in a scene and segmenting the semantic parts within each object.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*1UmLHr1hW83Cqm-v":{"__typename":"ImageMetadata","id":"0*1UmLHr1hW83Cqm-v","originalHeight":832,"originalWidth":1478,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_70":{"__typename":"Paragraph","id":"69cf10d652cf_70","name":"cec9","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*1UmLHr1hW83Cqm-v"},"text":"Multi-object Multi-part Segmentation","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":36,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*p4KhfHluCqaGO03N":{"__typename":"ImageMetadata","id":"0*p4KhfHluCqaGO03N","originalHeight":832,"originalWidth":1478,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_71":{"__typename":"Paragraph","id":"69cf10d652cf_71","name":"4a04","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*p4KhfHluCqaGO03N"},"text":"Ground Truth Label Map Changes","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":30,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_72":{"__typename":"Paragraph","id":"69cf10d652cf_72","name":"1f43","type":"P","href":null,"layout":null,"metadata":null,"text":"For this, the authors produce changes in the monolithic object label map structures to introduce more information such as front\u002Fback, left\u002Fright, and animate\u002Finanimate parts of the object. They use this information to create the Pascal-Part201 dataset. They propose the following model to solve the multi-object multi-part scene parsing challenge. The model consists of encoder-decoder-style architecture with different decoders for object level segmentation, front\u002Fback, left\u002Fright, and animate\u002Finanimate parts of the object. Finally, the feature maps are merged and an Inference Time Zoom Refinement (IZR module) is used to get the final output.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*8SElPbryjLaj6w3U":{"__typename":"ImageMetadata","id":"0*8SElPbryjLaj6w3U","originalHeight":598,"originalWidth":1059,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_73":{"__typename":"Paragraph","id":"69cf10d652cf_73","name":"d12a","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*8SElPbryjLaj6w3U"},"text":"The FLOAT model","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":15,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_74":{"__typename":"Paragraph","id":"69cf10d652cf_74","name":"d8d6","type":"H4","href":null,"layout":null,"metadata":null,"text":"Can you even tell left from right? Presenting a new challenge for VQA","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_75":{"__typename":"Paragraph","id":"69cf10d652cf_75","name":"7f78","type":"P","href":null,"layout":null,"metadata":null,"text":"Visual Question Answering (VQA) research aims to create a computer system that can answer questions using both an image and natural language. VQA needs a means of evaluating the strengths and weaknesses of models. One is the evaluation of compositional generalisation, or the ability of a model to answer well on scenes whose scene setups are different from the training set. For this, we need datasets whose train and test sets differ significantly in composition.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_76":{"__typename":"Paragraph","id":"69cf10d652cf_76","name":"a994","type":"P","href":null,"layout":null,"metadata":null,"text":"This study introduces quantitative measures of compositional separation and shows that current VQA datasets are inadequate for evaluation. To solve this, they present Uncommon Objects in Unseen Configurations (UOUC), a synthetic dataset for VQA. UOUC is at once fairly complex while also being well-separated, compositionally. UOUC contains 380 object classes from 528 characters in the Dungeons and Dragons game, with 200,000 scenes in the train set and 30,000 in the test set.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_77":{"__typename":"Paragraph","id":"69cf10d652cf_77","name":"d44a","type":"P","href":null,"layout":null,"metadata":null,"text":"To study compositional generalisation, simple reasoning, and memorisation, each scene of UOUC is annotated with up to 10 novel questions. These deal with spatial relationships, hypothetical changes to scenes, counting, comparison, memorisation and memory-based reasoning. In total, UOUC presents over 2 million questions. UOUC also finds itself as a strong challenger to well-performing models for VQA. Read the full paper here.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":422,"end":427,"href":"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2203.07664.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*qXS8MOGvb3l3CHaP":{"__typename":"ImageMetadata","id":"0*qXS8MOGvb3l3CHaP","originalHeight":488,"originalWidth":1142,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_78":{"__typename":"Paragraph","id":"69cf10d652cf_78","name":"084c","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*qXS8MOGvb3l3CHaP"},"text":"Visual Question Answering datasets","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":34,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_79":{"__typename":"Paragraph","id":"69cf10d652cf_79","name":"f745","type":"P","href":null,"layout":null,"metadata":null,"text":"Learning compositional structures for deep learning: Why routing-by-agreement is necessary","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":90,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_80":{"__typename":"Paragraph","id":"69cf10d652cf_80","name":"2a96","type":"P","href":null,"layout":null,"metadata":null,"text":"A formal description of the compositionality of neural networks is associated directly with the formal grammar structure of the objects it seeks to represent. This formal grammar structure specifies the kind of components that make up an object, and also the configurations they are allowed to be in. In other words, objects can be described as a parse tree of its components — a structure that can be seen as a candidate for building connection patterns among neurons in neural networks. The authors present a formal grammar description of convolutional neural networks and capsule networks that shows how capsule networks can enforce such parse-tree structures, while CNNs do not. Read the full paper here.","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"A","start":702,"end":707,"href":"https:\u002F\u002Farxiv.org\u002Fpdf\u002F2010.01488.pdf","anchorType":"LINK","userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*Ncl84v9pZEiFMhU3":{"__typename":"ImageMetadata","id":"0*Ncl84v9pZEiFMhU3","originalHeight":488,"originalWidth":1142,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_81":{"__typename":"Paragraph","id":"69cf10d652cf_81","name":"35dc","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*Ncl84v9pZEiFMhU3"},"text":"Change in Compositionality","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":26,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_82":{"__typename":"Paragraph","id":"69cf10d652cf_82","name":"0579","type":"H4","href":null,"layout":null,"metadata":null,"text":"Learnings from Industry Sessions:","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_83":{"__typename":"Paragraph","id":"69cf10d652cf_83","name":"87f0","type":"P","href":null,"layout":null,"metadata":null,"text":"The industry sessions featured interesting research work published by various competitors in the market. They had set up posters and demo booths where people could discuss more about their work and can see real-time demos. The sessions were from companies like Qualcomm, Adobe, Samsung R&D, L&T etc.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_84":{"__typename":"Paragraph","id":"69cf10d652cf_84","name":"4875","type":"H4","href":null,"layout":null,"metadata":null,"text":"Samsung R&D","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_85":{"__typename":"Paragraph","id":"69cf10d652cf_85","name":"cb5b","type":"P","href":null,"layout":null,"metadata":null,"text":"Researchers showcased work on various image editing features that they have incorporated into their latest mobile phones. Some of the notable features are shadow remover, photo remaster, image in-painting, and portrait mode, which are all deployed in their latest smartphones.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*zmq28qzYWLzLa0gp":{"__typename":"ImageMetadata","id":"0*zmq28qzYWLzLa0gp","originalHeight":684,"originalWidth":1128,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_86":{"__typename":"Paragraph","id":"69cf10d652cf_86","name":"e274","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*zmq28qzYWLzLa0gp"},"text":"Samsung’s new image editing features","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":36,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_87":{"__typename":"Paragraph","id":"69cf10d652cf_87","name":"336d","type":"P","href":null,"layout":null,"metadata":null,"text":"Under Display Camera","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"STRONG","start":0,"end":20,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_88":{"__typename":"Paragraph","id":"69cf10d652cf_88","name":"dcf5","type":"P","href":null,"layout":null,"metadata":null,"text":"Apart from this, their major contribution was the development of the under-display camera. An Under Display Camera (UDC) is a breakthrough innovation that enables an uninterrupted viewing experience on a mobile device by hiding the camera under the display and dedicating the whole screen to users while applications are running. It not only requires hardware innovation by placing a camera under a display panel but also requires algorithm innovation for restoring image quality — one of the most complex image restoration problems.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_89":{"__typename":"Paragraph","id":"69cf10d652cf_89","name":"1ca2","type":"P","href":null,"layout":null,"metadata":null,"text":"As the camera is placed underneath the display, the Under Display Camera can suffer from poor image quality caused by diffraction artefacts, which results in flare, saturation, blur and haze. Therefore, while the Under Display Camera brings a better display experience, it also affects camera image quality and other downstream vision tasks. These complex and diverse distortions make restoring Under Display Camera images extremely challenging.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_90":{"__typename":"Paragraph","id":"69cf10d652cf_90","name":"a7bc","type":"P","href":null,"layout":null,"metadata":null,"text":"In this talk, the author discussed some of the challenges with the Under Display Camera system & presented their work on image restoration for Under Display Camera.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*fARlGdHBK5cDgZhK":{"__typename":"ImageMetadata","id":"0*fARlGdHBK5cDgZhK","originalHeight":900,"originalWidth":1600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_91":{"__typename":"Paragraph","id":"69cf10d652cf_91","name":"2125","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*fARlGdHBK5cDgZhK"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*wo6CH75G6vF1jlcJ":{"__typename":"ImageMetadata","id":"0*wo6CH75G6vF1jlcJ","originalHeight":466,"originalWidth":1011,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_92":{"__typename":"Paragraph","id":"69cf10d652cf_92","name":"5a7a","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*wo6CH75G6vF1jlcJ"},"text":"Under Display Camera","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":20,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_93":{"__typename":"Paragraph","id":"69cf10d652cf_93","name":"761f","type":"H4","href":null,"layout":null,"metadata":null,"text":"Adobe","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_94":{"__typename":"Paragraph","id":"69cf10d652cf_94","name":"8013","type":"P","href":null,"layout":null,"metadata":null,"text":"In this talk, the speakers provided an overview of Adobe Research and the key areas that they are working on. They gave us a peek at some of their recent work on video generation, image out-painting and graphic design harmonization. Their recent work on image out-painting and animating still images show how expressing visual data via intermediate representations and manipulating provides better outputs against direct pixel-level manipulations.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_95":{"__typename":"Paragraph","id":"69cf10d652cf_95","name":"1734","type":"P","href":null,"layout":null,"metadata":null,"text":"The authors propose a method to interactively control the animation of fluid elements in still images to generate cinemagraphs. Specifically, they focus on the animation of fluid elements like water, smoke, and fire, which have the properties of repeating textures and continuous fluid motion. They represent the motion of such fluid elements in the image in the form of a constant 2D optical flow map. The user can provide any number of arrow directions and the associated speed along with a mask of the regions the user wants to animate. The user-provided input arrow directions, their corresponding speed values, and the mask is then converted into a dense flow map representing a constant optical flow map (FD).","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*tnYUW4uPgCwjkroY":{"__typename":"ImageMetadata","id":"0*tnYUW4uPgCwjkroY","originalHeight":131,"originalWidth":600,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_96":{"__typename":"Paragraph","id":"69cf10d652cf_96","name":"4e6e","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*tnYUW4uPgCwjkroY"},"text":"Animation of fluid elements","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":27,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_97":{"__typename":"Paragraph","id":"69cf10d652cf_97","name":"cf40","type":"P","href":null,"layout":null,"metadata":null,"text":"The authors observe that FD, obtained using simple exponential operations can closely approximate the plausible motion of elements in the image. They further refined a computed dense optical flow map FD using a generative-adversarial network (GAN) to obtain a more realistic flow map. A novel U-Net-based architecture was proposed to auto-regressively generate future frames using the refined optical flow map by forward-warping the input image features at different resolutions.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*U73Z6WEgRM6TAsVA":{"__typename":"ImageMetadata","id":"0*U73Z6WEgRM6TAsVA","originalHeight":688,"originalWidth":1152,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_98":{"__typename":"Paragraph","id":"69cf10d652cf_98","name":"d8bd","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*U73Z6WEgRM6TAsVA"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_99":{"__typename":"Paragraph","id":"69cf10d652cf_99","name":"014a","type":"P","href":null,"layout":null,"metadata":null,"text":"Some other research showcased were:","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*d_KvKweXnnmGAktu":{"__typename":"ImageMetadata","id":"0*d_KvKweXnnmGAktu","originalHeight":710,"originalWidth":1257,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_100":{"__typename":"Paragraph","id":"69cf10d652cf_100","name":"7f5e","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*d_KvKweXnnmGAktu"},"text":"","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*W3A5AeXuVbURV-H_":{"__typename":"ImageMetadata","id":"0*W3A5AeXuVbURV-H_","originalHeight":667,"originalWidth":1162,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_101":{"__typename":"Paragraph","id":"69cf10d652cf_101","name":"92e3","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*W3A5AeXuVbURV-H_"},"text":"Design Understand & Generation (top), Image Understanding & Generation (bottom)","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":79,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_102":{"__typename":"Paragraph","id":"69cf10d652cf_102","name":"19e6","type":"H4","href":null,"layout":null,"metadata":null,"text":"TCS Research","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_103":{"__typename":"Paragraph","id":"69cf10d652cf_103","name":"4bee","type":"P","href":null,"layout":null,"metadata":null,"text":"Draping a 3D human mesh has garnered broad interest due to its wide applicability in virtual try-on, animations, etc. The 3D garment produced by existing methods are often inconsistent with the body shape, pose, and measurements. This paper proposes a single unified learning-based framework (DeepDraper) to predict garment deformation as a function of body shape, pose, measurements, and garment styles. The authors train the DeepDraper with coupled geometric and multi-view perceptual losses.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*HgLCrZosH91anc4I":{"__typename":"ImageMetadata","id":"0*HgLCrZosH91anc4I","originalHeight":720,"originalWidth":1280,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_104":{"__typename":"Paragraph","id":"69cf10d652cf_104","name":"fb60","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*HgLCrZosH91anc4I"},"text":"GarSim","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":6,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*dsr6atoqxkUl5pIM":{"__typename":"ImageMetadata","id":"0*dsr6atoqxkUl5pIM","originalHeight":816,"originalWidth":1024,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_105":{"__typename":"Paragraph","id":"69cf10d652cf_105","name":"75e8","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*dsr6atoqxkUl5pIM"},"text":"DeepDraper","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":10,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_106":{"__typename":"Paragraph","id":"69cf10d652cf_106","name":"c440","type":"P","href":null,"layout":null,"metadata":null,"text":"Unlike existing methods, they additionally model garment deformations as a function of standard body measurements, which generally a buyer or a designer uses to buy or design perfect-fit clothes. In addition to that, the authors claim that DeepDraper is 10 times smaller in size and 23 times faster than the closest state-of-the-art method (TailorNet), which favours its use in real-time applications with less computational power.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*IZ7oyMJ94lePxVg0":{"__typename":"ImageMetadata","id":"0*IZ7oyMJ94lePxVg0","originalHeight":728,"originalWidth":1344,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_107":{"__typename":"Paragraph","id":"69cf10d652cf_107","name":"a7d0","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*IZ7oyMJ94lePxVg0"},"text":"DeepDraper training & Inference Pipeline","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":40,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_108":{"__typename":"Paragraph","id":"69cf10d652cf_108","name":"738c","type":"H4","href":null,"layout":null,"metadata":null,"text":"Final thoughts and Key Takeaways:","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_109":{"__typename":"Paragraph","id":"69cf10d652cf_109","name":"f87c","type":"P","href":null,"layout":null,"metadata":null,"text":"Our team had an amazing three-day experience full of learning opportunities! Witnessing the growth of the computer vision community in India was nothing short of exhilarating. We were thrilled to see the groundbreaking research being done by these talented individuals, and we couldn’t wait to learn more.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_110":{"__typename":"Paragraph","id":"69cf10d652cf_110","name":"7592","type":"P","href":null,"layout":null,"metadata":null,"text":"The key points that I’d like to take away with me are:","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_111":{"__typename":"Paragraph","id":"69cf10d652cf_111","name":"fa8c","type":"ULI","href":null,"layout":null,"metadata":null,"text":"It is better to use problem-specific priors to aid learning in your deep networks compared to blind input-to-output mapping.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_112":{"__typename":"Paragraph","id":"69cf10d652cf_112","name":"85af","type":"ULI","href":null,"layout":null,"metadata":null,"text":"Having in-depth knowledge of the latest developments in different problem statements can help in translating ideas across different domains of AI.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_113":{"__typename":"Paragraph","id":"69cf10d652cf_113","name":"6d78","type":"ULI","href":null,"layout":null,"metadata":null,"text":"Full stack expertise of tech often pays well in designing a powerful product.","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"ImageMetadata:0*C4absgD21VLDN5jk":{"__typename":"ImageMetadata","id":"0*C4absgD21VLDN5jk","originalHeight":544,"originalWidth":652,"focusPercentX":null,"focusPercentY":null,"alt":null},"Paragraph:69cf10d652cf_114":{"__typename":"Paragraph","id":"69cf10d652cf_114","name":"dcb0","type":"IMG","href":null,"layout":"INSET_CENTER","metadata":{"__ref":"ImageMetadata:0*C4absgD21VLDN5jk"},"text":"The Fynd Research Team at IIT Gandhinagar (L-R): Bipin Gaikwad, Arnab Mishra, Prasanna Kumar, Shashank Vasisht, Vignesh Prajapati","hasDropCap":null,"dropCapImage":null,"markups":[{"__typename":"Markup","type":"EM","start":0,"end":129,"href":null,"anchorType":null,"userId":null,"linkMetadata":null}],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"Paragraph:69cf10d652cf_115":{"__typename":"Paragraph","id":"69cf10d652cf_115","name":"1e98","type":"P","href":null,"layout":null,"metadata":null,"text":"It was incredibly inspiring to see the strides being made in this field, and we feel grateful to have been a part of it. We can’t wait to attend more conferences like this and hopefully even present the groundbreaking work we’re doing here at Fynd too!","hasDropCap":null,"dropCapImage":null,"markups":[],"codeBlockMetadata":null,"iframe":null,"mixtapeMetadata":null},"CollectionViewerEdge:collectionId:91d0019cb1ab-viewerId:lo_7432eda3b414":{"__typename":"CollectionViewerEdge","id":"collectionId:91d0019cb1ab-viewerId:lo_7432eda3b414","isEditor":false,"isMuting":false},"ImageMetadata:1*qOeeVf-aNCFCtn2_qeSUFw.png":{"__typename":"ImageMetadata","id":"1*qOeeVf-aNCFCtn2_qeSUFw.png","originalWidth":2501,"originalHeight":1251},"PostViewerEdge:postId:bb3fb4c41bd8-viewerId:lo_7432eda3b414":{"__typename":"PostViewerEdge","shouldIndexPostForExternalSearch":true,"id":"postId:bb3fb4c41bd8-viewerId:lo_7432eda3b414"},"Tag:icvgip-2022":{"__typename":"Tag","id":"icvgip-2022","displayTitle":"Icvgip 2022","normalizedTagSlug":"icvgip-2022"},"Tag:computer-vision":{"__typename":"Tag","id":"computer-vision","displayTitle":"Computer Vision","normalizedTagSlug":"computer-vision"},"Tag:image-processing":{"__typename":"Tag","id":"image-processing","displayTitle":"Image Processing","normalizedTagSlug":"image-processing"},"Tag:computer-graphics":{"__typename":"Tag","id":"computer-graphics","displayTitle":"Computer Graphics","normalizedTagSlug":"computer-graphics"},"Tag:machine-learning":{"__typename":"Tag","id":"machine-learning","displayTitle":"Machine Learning","normalizedTagSlug":"machine-learning"},"Post:bb3fb4c41bd8":{"__typename":"Post","id":"bb3fb4c41bd8","collection":{"__ref":"Collection:91d0019cb1ab"},"content({\"postMeteringOptions\":{}})":{"__typename":"PostContent","isLockedPreviewOnly":false,"bodyModel":{"__typename":"RichText","sections":[{"__typename":"Section","name":"d898","startIndex":0,"textLayout":null,"imageLayout":null,"backgroundImage":null,"videoLayout":null,"backgroundVideo":null}],"paragraphs":[{"__ref":"Paragraph:69cf10d652cf_0"},{"__ref":"Paragraph:69cf10d652cf_1"},{"__ref":"Paragraph:69cf10d652cf_2"},{"__ref":"Paragraph:69cf10d652cf_3"},{"__ref":"Paragraph:69cf10d652cf_4"},{"__ref":"Paragraph:69cf10d652cf_5"},{"__ref":"Paragraph:69cf10d652cf_6"},{"__ref":"Paragraph:69cf10d652cf_7"},{"__ref":"Paragraph:69cf10d652cf_8"},{"__ref":"Paragraph:69cf10d652cf_9"},{"__ref":"Paragraph:69cf10d652cf_10"},{"__ref":"Paragraph:69cf10d652cf_11"},{"__ref":"Paragraph:69cf10d652cf_12"},{"__ref":"Paragraph:69cf10d652cf_13"},{"__ref":"Paragraph:69cf10d652cf_14"},{"__ref":"Paragraph:69cf10d652cf_15"},{"__ref":"Paragraph:69cf10d652cf_16"},{"__ref":"Paragraph:69cf10d652cf_17"},{"__ref":"Paragraph:69cf10d652cf_18"},{"__ref":"Paragraph:69cf10d652cf_19"},{"__ref":"Paragraph:69cf10d652cf_20"},{"__ref":"Paragraph:69cf10d652cf_21"},{"__ref":"Paragraph:69cf10d652cf_22"},{"__ref":"Paragraph:69cf10d652cf_23"},{"__ref":"Paragraph:69cf10d652cf_24"},{"__ref":"Paragraph:69cf10d652cf_25"},{"__ref":"Paragraph:69cf10d652cf_26"},{"__ref":"Paragraph:69cf10d652cf_27"},{"__ref":"Paragraph:69cf10d652cf_28"},{"__ref":"Paragraph:69cf10d652cf_29"},{"__ref":"Paragraph:69cf10d652cf_30"},{"__ref":"Paragraph:69cf10d652cf_31"},{"__ref":"Paragraph:69cf10d652cf_32"},{"__ref":"Paragraph:69cf10d652cf_33"},{"__ref":"Paragraph:69cf10d652cf_34"},{"__ref":"Paragraph:69cf10d652cf_35"},{"__ref":"Paragraph:69cf10d652cf_36"},{"__ref":"Paragraph:69cf10d652cf_37"},{"__ref":"Paragraph:69cf10d652cf_38"},{"__ref":"Paragraph:69cf10d652cf_39"},{"__ref":"Paragraph:69cf10d652cf_40"},{"__ref":"Paragraph:69cf10d652cf_41"},{"__ref":"Paragraph:69cf10d652cf_42"},{"__ref":"Paragraph:69cf10d652cf_43"},{"__ref":"Paragraph:69cf10d652cf_44"},{"__ref":"Paragraph:69cf10d652cf_45"},{"__ref":"Paragraph:69cf10d652cf_46"},{"__ref":"Paragraph:69cf10d652cf_47"},{"__ref":"Paragraph:69cf10d652cf_48"},{"__ref":"Paragraph:69cf10d652cf_49"},{"__ref":"Paragraph:69cf10d652cf_50"},{"__ref":"Paragraph:69cf10d652cf_51"},{"__ref":"Paragraph:69cf10d652cf_52"},{"__ref":"Paragraph:69cf10d652cf_53"},{"__ref":"Paragraph:69cf10d652cf_54"},{"__ref":"Paragraph:69cf10d652cf_55"},{"__ref":"Paragraph:69cf10d652cf_56"},{"__ref":"Paragraph:69cf10d652cf_57"},{"__ref":"Paragraph:69cf10d652cf_58"},{"__ref":"Paragraph:69cf10d652cf_59"},{"__ref":"Paragraph:69cf10d652cf_60"},{"__ref":"Paragraph:69cf10d652cf_61"},{"__ref":"Paragraph:69cf10d652cf_62"},{"__ref":"Paragraph:69cf10d652cf_63"},{"__ref":"Paragraph:69cf10d652cf_64"},{"__ref":"Paragraph:69cf10d652cf_65"},{"__ref":"Paragraph:69cf10d652cf_66"},{"__ref":"Paragraph:69cf10d652cf_67"},{"__ref":"Paragraph:69cf10d652cf_68"},{"__ref":"Paragraph:69cf10d652cf_69"},{"__ref":"Paragraph:69cf10d652cf_70"},{"__ref":"Paragraph:69cf10d652cf_71"},{"__ref":"Paragraph:69cf10d652cf_72"},{"__ref":"Paragraph:69cf10d652cf_73"},{"__ref":"Paragraph:69cf10d652cf_74"},{"__ref":"Paragraph:69cf10d652cf_75"},{"__ref":"Paragraph:69cf10d652cf_76"},{"__ref":"Paragraph:69cf10d652cf_77"},{"__ref":"Paragraph:69cf10d652cf_78"},{"__ref":"Paragraph:69cf10d652cf_79"},{"__ref":"Paragraph:69cf10d652cf_80"},{"__ref":"Paragraph:69cf10d652cf_81"},{"__ref":"Paragraph:69cf10d652cf_82"},{"__ref":"Paragraph:69cf10d652cf_83"},{"__ref":"Paragraph:69cf10d652cf_84"},{"__ref":"Paragraph:69cf10d652cf_85"},{"__ref":"Paragraph:69cf10d652cf_86"},{"__ref":"Paragraph:69cf10d652cf_87"},{"__ref":"Paragraph:69cf10d652cf_88"},{"__ref":"Paragraph:69cf10d652cf_89"},{"__ref":"Paragraph:69cf10d652cf_90"},{"__ref":"Paragraph:69cf10d652cf_91"},{"__ref":"Paragraph:69cf10d652cf_92"},{"__ref":"Paragraph:69cf10d652cf_93"},{"__ref":"Paragraph:69cf10d652cf_94"},{"__ref":"Paragraph:69cf10d652cf_95"},{"__ref":"Paragraph:69cf10d652cf_96"},{"__ref":"Paragraph:69cf10d652cf_97"},{"__ref":"Paragraph:69cf10d652cf_98"},{"__ref":"Paragraph:69cf10d652cf_99"},{"__ref":"Paragraph:69cf10d652cf_100"},{"__ref":"Paragraph:69cf10d652cf_101"},{"__ref":"Paragraph:69cf10d652cf_102"},{"__ref":"Paragraph:69cf10d652cf_103"},{"__ref":"Paragraph:69cf10d652cf_104"},{"__ref":"Paragraph:69cf10d652cf_105"},{"__ref":"Paragraph:69cf10d652cf_106"},{"__ref":"Paragraph:69cf10d652cf_107"},{"__ref":"Paragraph:69cf10d652cf_108"},{"__ref":"Paragraph:69cf10d652cf_109"},{"__ref":"Paragraph:69cf10d652cf_110"},{"__ref":"Paragraph:69cf10d652cf_111"},{"__ref":"Paragraph:69cf10d652cf_112"},{"__ref":"Paragraph:69cf10d652cf_113"},{"__ref":"Paragraph:69cf10d652cf_114"},{"__ref":"Paragraph:69cf10d652cf_115"}]},"validatedShareKey":"","shareKeyCreator":null},"creator":{"__ref":"User:6408aa1c1489"},"inResponseToEntityType":null,"isLocked":false,"isMarkedPaywallOnly":false,"lockedSource":"LOCKED_POST_SOURCE_NONE","mediumUrl":"https:\u002F\u002Fblog.gofynd.com\u002Fexploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8","primaryTopic":{"__ref":"Topic:1eca0103fff3"},"topics":[{"__typename":"Topic","slug":"machine-learning"}],"isPublished":true,"latestPublishedVersion":"69cf10d652cf","visibility":"PUBLIC","postResponses":{"__typename":"PostResponses","count":1},"clapCount":103,"allowResponses":true,"isLimitedState":false,"title":"Exploring the latest innovations in Computer Vision","isSeries":false,"sequence":null,"uniqueSlug":"exploring-the-latest-innovations-in-computer-vision-bb3fb4c41bd8","socialTitle":"","socialDek":"","canonicalUrl":"","metaDescription":"","latestPublishedAt":1679984698883,"readingTime":14.99622641509434,"previewContent":{"__typename":"PreviewContent","subtitle":"Insights from Fynd’s visit to The Indian Conference on Computer Vision, Graphics & Image Processing 2022"},"previewImage":{"__ref":"ImageMetadata:0*rJjyYQz-gKIc45dN"},"isShortform":false,"seoTitle":"","firstPublishedAt":1679984698883,"updatedAt":1681416328715,"shortformType":"SHORTFORM_TYPE_LINK","seoDescription":"","viewerEdge":{"__ref":"PostViewerEdge:postId:bb3fb4c41bd8-viewerId:lo_7432eda3b414"},"isSuspended":false,"license":"ALL_RIGHTS_RESERVED","tags":[{"__ref":"Tag:icvgip-2022"},{"__ref":"Tag:computer-vision"},{"__ref":"Tag:image-processing"},{"__ref":"Tag:computer-graphics"},{"__ref":"Tag:machine-learning"}],"isNewsletter":false,"statusForCollection":"APPROVED","pendingCollection":null,"detectedLanguage":"en","wordCount":3338,"layerCake":6,"responsesLocked":false}}</script><script src="https://cdn-client.medium.com/lite/static/js/manifest.bbe4ab66.js"></script><script src="https://cdn-client.medium.com/lite/static/js/9865.1496d74a.js"></script><script src="https://cdn-client.medium.com/lite/static/js/main.bbe47cad.js"></script><script src="https://cdn-client.medium.com/lite/static/js/instrumentation.d9108df7.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/reporting.ff22a7a5.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/9120.5df29668.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/5049.d1ead72d.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/4810.6318add7.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/6618.db187378.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/2707.b0942613.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/9977.5b3eb23a.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/8599.1ab63137.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/5250.9f9e01d2.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/5787.e66a3a4d.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/2648.26563adf.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/8393.826a25fb.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/7549.2176f21f.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/6589.7c500280.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/3735.afb7e926.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/5642.0a97706a.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/6546.cd03f950.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/6834.08de95de.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/7346.72622eb9.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/2420.2a5e2d95.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/839.ca7937c2.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/7975.d195c6f1.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/7394.bf599bc5.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/2961.00a48598.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/8204.c4082863.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/4391.59acaed3.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/PostPage.MainContent.902ad94b.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/8414.6565ad5f.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/3974.8d3e0217.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/2527.a0afad8a.chunk.js"></script> <script src="https://cdn-client.medium.com/lite/static/js/PostResponsesContent.36c2ecf4.chunk.js"></script><script>window.main();</script></body></html>