Lei Zhang, Microsoft Research Asia

Title: Learning to detect repeatable patterns for visual recognition
Abstract: In recent years, a large variety of efforts have been spent in computer vision on detecting local invariant features, identifying visual attributes and building a bank of object detectors. Among these efforts, a common and fundamental problem is to detect repeatable visual patterns that are invariant to intra-class variations and selective to inter-class variations. In this talk, I will introduce our latest studies along this direction and show how to employ learning algorithms to detect repeatable visual patterns. Such visual patterns are generally of different granularities, for example, local features that are semantically more repeatable and object parts that have mid-level semantics to represent a visual category. I will show their applications in image retrieval, object recognition and image parsing.

Biography: Lei Zhang is a lead researcher in the Web Search & Mining Group at Microsoft Research Asia in Beijing. He is interested in research problems in image search, Internet vision and information retrieval, and holds 20 U.S. patents for his innovation in these fields. He is an IEEE senior member and an ACM senior member, and has served as an associate editor of Multimedia System Journal, program area chair of ACM Multimedia 2012, ICPR 2012 and ICME 2011, and also served on international conference program committees, including ACM Multimedia, ICCV, CVPR, WWW, SIGIR, etc. He earned a B.S. and M.S. in Computer Science from Tsinghua University in 1993 and 1995. After two years working in industry, he later returned to Tsinghua and received his PhD degree in Computer Science in 2001.