Event
Distinguished Seminar Series on Vision: Dr. Mubarak Shah
Monday, May 1, 2006
11:00 a.m.
Janice M. Perrone
301-405-1736
janice@umiacs.umd.edu
Distinguished Seminar Series on Vision
in honor of Prof. Azriel Rosenfeld
Computer Vision Laboratory
Center for Automation Research
A.V. Williams Building, Room 2460
The Fundamental Matrix in Human Action Recognition
Dr. Mubarak Shah
Computer Vision Lab
School of Computer Science
University of Central Florida
Orlando, FL 32816
http://www.cs.ucf.edu/~vision/
ABSTRACT
In this talk I will present our work on innovative use of fundamental matrix in human action recognition and its extension to independently moving cameras. I will first present our work on recognizing human actions performed by a single hand. The motion of a hand while performing an action generates a trajectory, which can be considered as an object in 3-D. Two views of the action can be matched using the fundamental matrix, resulting in view invariant action recognition.
A single point trajectory does not carry any shape or relative spatial information, which may be useful in action recognition. Instead of considering a single point, complete object contours can be used in action matching. The sequence of such object contours produces a spatio-temporal 3D volume. I will demonstrate the use of fundamental matrix in matching spatiotemporal volumes for human action recognition.
Next I will discuss the anthropometric constraint which states that a fundamental matrix can be uniquely associated between two actors (of different age, built and sex etc) if they are in the same posture and are performing the same action. I will present application of this constraint in action detection and action synchronization.
The fundamental matrix is defined for two fixed cameras observing the same scene. Next we will address the question: What happens to the fundamental matrix when cameras move independently? I will discuss temporal fundamental matrix, which captures the changes in fundamental matrix with respect to time in a single matrix, whose elements are function of time. I will present two application of this new construct: recognizing human and solving object correspondence across videos acquired by two independently moving cameras.
BIO
Dr. Mubarak Shah, Agere Chair professor of Computer Science, and the founding director of the Computer Vision Laboratory at University of Central Florida (UCF), is a researcher in computer vision. He has published two books, ten book chapters, fifty five papers in top journals and one hundred thirty papers in refereed international conferences. He has worked in several areas including activity and gesture recognition, violence detection, event ontology, object tracking (fixed camera, moving camera, multiple overlapping and non-overlapping cameras), video segmentation, story and scene segmentation, view morphing, ATR, wide-baseline matching, and video registration. .
Dr. Shah is a fellow of IEEE, was an IEEE Distinguished Visitor speaker for 1997-2000, and is often invited to present seminars, tutorials and invited talks all over the world. He received the Harris Corporation Engineering Achievement Award in 1999, the IEEE Outstanding Engineering Educator Award in 1997, TOKTEN awards from United Nations Development Program in 1995, 1997, and 2000, Teaching Incentive award in 1995 and 2003, Research Incentive Award in 2003, Millionaires' Club award in 2005, PEGASUS Professor award in 2006, an honorable mention for the ICCV 2005 Where Am I? Challenge Problem, and was nominated for the best paper award in ACM Multimedia Conference in 2005. He is an editor of international book series on Video Computing; editor in chief of Machine Vision and Applications journal, Area editor of Wiley Encyclopedia of Computer Science and Engineering, and an associate editor Pattern Recognition journal. He was an associate editor of the IEEE Transactions on PAMI, and a guest editor of the special issue of International Journal of Computer Vision on Video Computing.
