Learning for Visual Data Compression

<!DOCTYPE html> <html lang="en-US"> <head> <meta charset="UTF-8">  <title>Learning for Visual Data Compression | CVPR 2021 Tutorial</title> <meta name="generator" content="Jekyll v3.9.0" /> <meta property="og:title" content="Learning for Visual Data Compression" /> <meta property="og:locale" content="en_US" /> <meta name="description" content="CVPR 2021 Tutorial" /> <meta property="og:description" content="CVPR 2021 Tutorial" /> <link rel="canonical" href="https://guolu-home.github.io/cvpr21-tutorial.html" /> <meta property="og:url" content="https://guolu-home.github.io/cvpr21-tutorial.html" /> <meta property="og:site_name" content="Learning for Visual Data Compression" /> <meta name="twitter:card" content="summary" /> <meta property="twitter:title" content="Learning for Visual Data Compression" /> <script type="application/ld+json"> {"description":"CVPR 2021 Tutorial","url":"https://guolu-home.github.io/cvpr21-tutorial.html","@type":"WebPage","headline":"Learning for Visual Data Compression","@context":"https://schema.org"}</script>  <meta name="viewport" content="width=device-width, initial-scale=1"> <meta name="theme-color" content="#157878"> <link href='https://fonts.googleapis.com/css?family=Open+Sans:400,700' rel='stylesheet' type='text/css'> <link rel="stylesheet" href="/assets/css/style.css?v=b146dbc1a71218c5d8e3e7e20bf368863ecc077a"> </head> <body> <section class="page-header"> <h1 class="project-name">Learning for Visual Data Compression</h1> <h2 class="project-tagline">CVPR 2021 Tutorial<br> Time: 10AM-2PM EDT, June 19, 2021</h2> </section> <section class="main-content"> <h1 id="learning-for-visual-data-compression">Learning for Visual Data Compression</h1> <hr /> <h2 id="organizers">Organizers</h2> <table> <tr> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/xudong.png" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="https://www.sydney.edu.au/engineering/about/our-people/academic-staff/dong-xu.html">Dong Xu<br />University of Sydney</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/luguo.jpg" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="https://guolusjtu.github.io/guoluhomepage/">Guo Lu<br />Beijing Institute of Technology</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/liushan.png" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="https://www.linkedin.com/in/shanliu/">Shan Liu<br />Tencent</a></td> </tr> <tr> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/wangshenlong.jpg" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="http://www.cs.toronto.edu/~slwang/">Shenlong Wang<br />UIUC</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/raquel.jpg" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="http://www.cs.toronto.edu/~urtasun/">Raquel Urtasun<br />Uber ATG / University of Toronto</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/radu.png" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="http://people.ee.ethz.ch/~timofter/">Radu Timofte<br />ETH Zurich</a></td> </tr> </table> <h2 id="guest-speakers">Guest Speakers</h2> <table> <tr> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <img src="images/yangren.jpg" width="150px" height="150px" style="border-radius:100%; position:relative;" /><br /><a href="https://renyang-home.github.io/">Ren Yang<br />ETH Zurich</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <br /><a>TBD<br />TBD</a></td> <td align="center" valign="top" style="width:300px; border-color: transparent; overflow: hidden;"> <br /><a>TBD<br />TBD</a></td> </tr> </table> <h2 id="schedule-edt">Schedule (EDT)</h2> <div> <table class="alt"> <tbody> <col width="18%" /> <col width="62%" /> <col width="20%" /> <tr> <td><span class="announce_date">10:00 - 10:05</span></td> <td class="tabletext" style="text-align: left">Introduction to the speakers and our tutorial</td> <td class="tabletext">Dong Xu</td> </tr> <tr> <td><span class="announce_date">10:05 - 10:55</span></td> <td class="tabletext" style="text-align: left">Deep image compression [<a href="https://drive.google.com/file/d/1rgsPeRsipyGJmEMpC2YDsiT2gXEDwHx_/view?usp=sharing">slides</a>] </td> <td class="tabletext">Ren Yang</td> </tr> <tr> <td><span class="announce_date">10:55 - 11:45</span></td> <td class="tabletext" style="text-align: left">Deep video compression [<a href="https://drive.google.com/file/d/1_-ZUYK4pPhuqGTPVLUIl8PHFeVp5OJ8R/view?usp=sharing">slides</a>]</td> <td class="tabletext">Guo Lu</td> </tr> <tr> <td><span class="announce_date">11:45 - 12:15</span></td> <td class="tabletext" style="text-align: left">Break </td> <td class="tabletext">--</td> </tr> <tr> <td><span class="announce_date">12:15 - 13:00</span></td> <td class="tabletext" style="text-align: left">Deep point cloud compression [<a href="https://drive.google.com/file/d/1fmLyTy2RPGZoU-7NKZl0VjIpXlQAJtHp/view?usp=sharing">slides</a>]</td> <td class="tabletext">Shenlong Wang</td> </tr> <tr> <td><span class="announce_date">13:00 - 13:30</span></td> <td class="tabletext" style="text-align: left">Visual data compression standards [<a href="https://drive.google.com/file/d/1L2L2H2iWeRuQxUXcsI-TkRKGKg4EJPi1/view?usp=sharing">slides</a>]</td> <td class="tabletext">Shan Liu</td> </tr> </tbody> </table> </div> <h2 id="description">Description</h2> <p>In this tutorial, we will introduce the recent progress in deep learning based visual data compression, including image compression, video compression and point cloud compression. In the past few years, deep learning techniques have been successfully applied to various computer vision and image processing applications. However, for the data compression task, the traditional approaches (i.e., block based motion estimation and motion compensation, etc.) are still widely employed in the mainstream codecs. Considering the powerful representation capability of neural networks, it is feasible to improve the data compression performance by employing the advanced deep learning technologies. To this end, the deep leaning based compression approaches have recently received increasing attention from both academia and industry in the field of computer vision and signal processing.</p> <p>Specifically, we will first introduce the end-to-end learning based image and video compression methods and discuss the current benchmark results. Then, we will provide detailed introductions for the latest standard procedures for learning based image or video compression approaches, such as JPEG AI, JVET NNVC and IEEE FVC. After that, we will discuss the recent work on learning based point cloud compression and analyze several widely used point cloud processing methods. Finally, we will discuss the limitations of the current learning based compression methods and the future research directions, like video compression for machines. In summary, our tutorial will cover both latest works from the academic community and the standardization progress in industry, which will help the audiences with different backgrounds better understand the recent progresses in this emerging research area.</p> <h3 id="tutorial-outline">Tutorial Outline</h3> <ol> <li> <p>Standard Activities of learning based Image and Video Compression<br /> a) Brief introduction to standards involving learning based image and video compression.<br /> b) Latest progress on learning based image and video coding tools in various standards.<br /> c) Discussion and Benchmark Results<br /></p> </li> <li> <p>End-to-end Learning based Image and Video Compression<br /> a) Brief introduction of classical image and video compression frameworks<br /> b) Learning based image compression<br /> c) Learning based video compression<br /> d) Rate distortion optimization techniques for learned image and video compression<br /> e) Benchmark results and Discussions<br /></p> </li> <li> <p>Learning based Point Cloud Compression<br /> a) Existing work for traditional point cloud compression<br /> b) Learning based static point cloud geometry compression<br /> c) Learning based dynamic point cloud geometry compression<br /> d) Learning based point cloud attribute compression<br /></p> </li> <li> <p>Discussion and Future Directions<br /> a) Limitations of the current learning based approaches<br /> b) Visual data compression for machines<br /> c) Visual data compression for robotics and self-driving<br /> d) Open source projects<br /></p> </li> </ol> <footer class="site-footer"> <span class="site-footer-credits">This page was generated by <a href="https://pages.github.com">GitHub Pages</a>.</span> </footer> </section> </body> </html>

CINXE.COM

Learning for Visual Data Compression | CVPR 2021 Tutorial