GAMES Webinar 2019 – 123期（三维视觉前沿专题报告） |汤思宇(Max Planck Institute for Intelligent Systems),崔兆鹏 (ETH Zurich)
【GAMES Webinar 2019-123期】（三维视觉前沿专题报告）
报告嘉宾1：汤思宇，Max Planck Institute for Intelligent Systems
报告题目：Learning to see and generate people
People are often a central element of visual scenes. It has been a long-standing goal in Computer Vision to develop methods aiming at analyzing humans in visual data. An accurate acquisition of people in the scene is critical for many applications such as virtual/argumented reality, autonomous driving and gaming. In this talk, I will give an overview of a selection of projects in which we discover and propose principles, algorithms and implementations for recognising and generating 3D human bodies. I will highlight some of our recent work on learning generative models for clothed humans, I will also present a fully-automatic system that takes a 3D scene and generates plausible 3D human bodies that are posed naturally in that 3D scene. I will conclude the talk by outlining the next challenges in teaching machine to understand our complex 3D world and people in it from visual data.
Siyu Tang received an early career research grant to start her own research group at the Max Planck Institute for Intelligent Systems in November 2017. She finished her PhD (summa cum laude) at the Max Planck Institute for Informatics and Saarland University in September 2017, under the supervision of Professor Bernt Schiele. Before that, she received her Master’s degree in Computer Science at RWTH Aachen University, advised by Prof. Bastian Leibe and her Bachelor degree in Computer Science at Zhejiang University, China. She was a research intern at the Japanese National Institute of Informatics. Dr. Tang received the DAGM-MVTec Dissertation Award in 2018. She was the winner of the multi-object tracking challenge at ECCV’16 and CVPR’17. She also received a Best Paper Award for her work “Detection and tracking of occluded people” at BMVC 2012. She will be joining the Computer Science Department of ETH Zürich as a tenure track assistant professor from January 2020.
报告题目：Polarimetric 3D Reconstruction and Image Separation
Polarization is a natural characteristic of light waves, which conveys geometric and physical cues of the surrounding environment. With a rotating polarizer in front of an ordinary camera or a polarization camera, we can capture polarized images with different polarization angles, from which we can recover the polarization information. In the first part of my talk, I will introduce how we exploit the geometric cue, i.e., the surface normal, from the polarization information for 3D reconstruction, including multi-view stereo (CVPR’17), dense monocular SLAM (CVPR’18), and relative pose estimation (ICCV’19). In the second part of my talk, I will present how we utilize the physical cue in the polarization to separate the reflection and transmission layers of images captured through the glass (NeurIPS’19).
Zhaopeng Cui is currently a Senior Researcher working with Prof. Marc Pollefeys in the Computer Vision and Geometry Group at ETH Zurich. He got his Ph.D. in computer science at Simon Fraser University where he worked with Prof. Ping Tan. His research interests include 3D Computer Vision, Robotics, and Computer Graphics. In particular, his research focuses on structure from motion, multi-view stereo, SLAM, visual localization, and image and video processing.
GAMES主页的“使用教程”中有 “如何观看GAMES Webinar直播？”及“如何加入GAMES微信群？”的信息；