Professor Xilin CHEN
Prof. Xinghuo Yu

Institute of Computing Technology
Chinese Academy of Sciences, China



Biography

Prof. Xilin Chen is a professor with the Institute of Computing Technology, Chinese Academy of Sciences. His research interests include computer vision, pattern recognition, multimedia and multimodal interfaces. His research topics cover face perception, sign language recognition, object recognition, and scene understanding. Recent works focus on explainable object recognition and scene understanding, including hierarchical modeling, vision and language, face perception for education, etc. He is currently a Senior Editor of the Journal of Visual Communication and Image Representation, and an associate editor-in-chief of the Chinese Journal of Computers, and Chinese Journal of Pattern Recognition and Artificial Intelligence. He served as an organizing committee member for multiple conferences, including general co-chair of IEEE FG13 / FG18, program co-chair of ACM ICMI 2010. He has co-authored one book and more than 300 papers. He received the Outstanding Achievement Award from Computer Vision Technical Committee of China Computer Federation (CCF) in 2019. He is a fellow of the ACM, IEEE, IAPR, and CCF.

Title

Towards Understandable Computer Vision

Abstract

In the past decades, computer vision has become the hottest area in artificial intelligence due to it reaches similar or even better results in some typical tasks, such as object recognition, than human being. However, most current computer vision systems are designed for specific task(s) in the close world, and hard to deal with open world cases. Flat structure for specific task(s) without reasoning and lack of knowledge are the major barriers toward a flexible computer vision system. For this purpose, a key factor is understandable or interpretable. Therefore, objects should be processed in a contextual environment rather than a solo one with a simple identity, and objects should also be associated with relevant concepts. A conceptual mapping of a given image brings enhanced representation, which can support versatile tasks. In this talk, I will briefly review the current state of computer vision, and talk about some open problems. I will then share my points on these relevant problems. Some of our efforts towards understandable computer vision are reported, including hierarchical object detection and categorization, scene graph construction and its application, unseen object inference, etc.