Vision and Autonomous Systems Center Seminar
- Remote Access Enabled - Zoom
- Virtual Presentation
- CE LIU
- Staff Research Scientist
- Google Resesarh
Advancing the State of the Art of Computer Vision for Billions of Users
At Google, advancing the state of the art of computer vision is very impactful as there are billions of users of Google products, many of which require high-quality, artifact-free images. I will share what we learned from successfully launching core computer vision techniques for various Google products, including PhotoScan (Photos), seamless Google Street View panorama stitching (Geo), Super Res Zoom (Pixel 4), Auto Pop-out & Uncrop (Display Ads), and Rendering4AI (Cloud AI)
We also conduct academic research and publish at top-tier conferences. I will give an overview of several representative works, including seeing through obstructions (Siggraph’15), learning the depth of moving people by watching frozen people (CVPR’19), GAN-based image uncrop (ICCV’19), and supervised contrastive learning (NeurIPS’20).
Dr. Ce Liu is a staff research scientist at Google Research, conducting research in the area of computer vision, computer graphics and machine learning. He received his B.E. and M.E. from Tsinghua University in 1999 and 2002, respectively. He received a PhD from MIT Department of Electrical Engineering and Computer Science in 2009. He worked at Microsoft Research Asia from 2002 to 2003, and Microsoft Research New England from 2009 to 2014. He has published more than 60 papers in peer-reviewed conferences and journals. He received the best student paper award at NIPS 2006 and CVPR 2009, and the best paper award honorable mention at CVPR 2019. He is a recipient of TPAMI Young Research Award in 2016. He has been serving as an area chair for CVPR/ICCV/ECCV/NeurIPS/ICLR, an associate editor for TPAMI, and served as a Program co-Chair for CVPR 2020.
Faculty Host: Fernando De La Torre Frade
Sponsored in part by Facebook Reality Labs Pittsburgh
Zoom Participation. See announcement. Virtual meetings with the speaker are possible.