Innovation Forums: The Latest Advances in Computer Vision & Its Applications


Moderator: Ting Yu
Google


Mei Han
Director of PingAn AI Institues
Talk Title: Computer Vision R&D at PingAn Technology

Abstract:
In this talk, I’ll describe the research and development efforts in Computer Vision area at PingAn Technology, especially US Research Labs. Our projects span over the application areas of Smart City, Agriculture Supervision, Pedestrian Behavior Analysis, Medical Image Segmentation, AI Painting and beyond. I’d share my experiences of applying the technologies to making impacts to businesses, and advancing the technologies in the research community.
Biography:

Mei Han is the director of PingAn AI Institues. Her research fields are in the areas of video analysis, visual tracking, object detection, geometric modeling, image processing, computer vision and machine learning. Prior to joining PAII, she held the position of research scientist at Google and research staff member at NEC Labs America. Technologies developed by Dr. Han and her colleagues are at the core of the company Vidient's innovative surveillance products. Mei Han holds Doctorates in Robotics and Computer Science from Carnegie Mellon University and Tsinghua University.




Jason Xu
Director of DiDi Labs
Talk Title: Augmenting Reality with Computer Vision Technologies

Abstract:
Augmented Reality is an interactive experience where real-world objects are "augmented" with computer-generated virtual contents. While in general multiple senses e.g. haptics or olfaction could be involved, in most common applications AR refers to visually augmenting the surrounding world, which is realized through certain display devices such as AR eyewear or a mobile phone. Under its hood, AR is powered by a collection of technologies, among which computer vision plays an especially important role. In fact, AR is a very unique application of computer vision because both semantic and geometric understanding are of significant interest here. Semantic vision techniques like deep learning are employed to perceive what the user is looking at, in order to present the most relevant information. On the other hand, geometric vision algorithms such as SLAM are required to precisely localize the AR device, so as to render the virtual objects in a way that is geometrically consistent with the physical ones. In this talk we will dive into AR, tour through a variety of computer vision technologies being used in practical AR systems, and explore the interesting applications enabled by them.
Biography:

Dr. Jason Xu is a Director at DiDi Labs and Head of Augmented Reality, leading a multinational R&D team dedicated to exploring the frontier applications of AR in intelligent vehicles and transportation systems. Jason is a seasoned technology leader in computer vision and artificial intelligence, with more than 20 US/EU patents awarded or pending. Before joining DiDi he was the Senior Director in charge of visual recognition R&D and AI strategies at Blippar, a leading AR advertising and visual discovery platform. Prior to that, Jason was a Research Manager at Sony Research and the director of its computational intelligence lab, leading the R&D of machine learning technologies supporting a variety of Sony products, including video encoding and analysis, audio processing, and wearable computing etc. Earlier in his career, Jason worked at renowned research labs including Siemens, Google and Microsoft, on R&D projects in medical imaging, image search and multimedia communication etc. Jason received his Ph.D. degree from the University of Illinois at Urbana-Champaign (UIUC) working with Prof. Thomas S. Huang. His master and bachelor degrees were both earned at Tsinghua University, Beijing, China.



Ming-Hsuan Yang
Research Lead of Google Cloud AI