Visual Event Recognition in Videos by Learning from Web Data

Speaker:	Prof. Dong XU
		Nanyang Technological University
		Singapore

Title:		"Visual Event Recognition in Videos by Learning
		 from Web Data"

Date:		Monday, 27 September 2010

Time:		10:30am - 11:30am

Venue:		Rm3598 (via lifts 27 or 28), HKUST

Abstract:

In the first part of the talk, I will introduce a visual event recognition
framework for consumer domain videos by leveraging a large amount of
loosely labeled web videos (e.g., from YouTube). Specifically, I will
present a new aligned space-time pyramid matching method to measure the
distances between two video clips as well as introduce a new cross-domain
learning method to cope with the considerable variation in feature
distributions between videos from two domains (i.e., web domain and
consumer domain). Extensive experiments demonstrate the effectiveness of
our proposed framework that requires only a small number of labeled
consumer videos by leveraging web data. In the second part of the talk, I
will introduce the ongoing research projects in our group.


********************
Biography:

Dong Xu received the B.Eng. and Ph.D. degrees from University of
Science and Technology of China in 2001 and 2005, respectively. He is
currently an Assistant Professor with Nanyang Technological University,
Singapore. During his Ph.D. study, he worked with Microsoft Research Asia
and The Chinese University of Hong Kong for more than two years. He also
spent one year at Columbia University, New York, as a Postdoctoral
Research Scientist. He was co-author (with his PhD student Lixin Duan) of
a paper that won the Best Student Paper Award in the prestigious IEEE
International Conference on Computer Vision and Pattern Recognition (CVPR
2010). His research interests include computer vision, machine learning,
and multimedia content analysis.