Robotics Institute
Seminar, March 26
Time
and Place | Seminar Abstract | Speaker
Biography | Speaker Appointments
Representing and Recognizing Human Actions
James W. Davis
Dept. of Computer Science and Engineering,
Center for Cognitive Science
Ohio State University
Mauldin Auditorium (NSH 1305)
Refreshments 3:15 pm
Talk 3:30 pm
The ability of
computers to detect people and reliably recognize their activities (and
ultimately their intent) is of significant importance to automatic, intelligent
video analysis and surveillance. In this talk, I will describe some of our
current research efforts toward several challenging problems in human action
recognition. First I will briefly overview our past work on the compact Motion
History Image representation. Then I will present a probabilistic
reliable-inference framework for minimal-latency classification of actions.
Next, I will present a new adaptive principal components model for representing
action "styles" and demonstrate the approach for gender recognition.
I will also briefly describe our ongoing efforts toward integrating the research
into an automatic video surveillance system currently being developed at OSU.
Support for this research is provided by the National Science Foundation, U.S.
Army Night Vision Laboratory, U.S. Air Force, Intel, and Ohio Board of Regents.
James W. Davis earned his Ph.D. from
MIT in 2000, and is currently an Assistant Professor in the Department of
Computer Science and Engineering at Ohio
State University.
He is also affiliated with the Center for Cognitive Science at OSU. His main
research interests are in computer vision, visual perception, human-computer
interaction, and artificial intelligence. Specifically, he is interested in
designing intelligent computer vision methods to reliably detect and recognize
human activities. His research was recently recognized by the National Science
Foundation with the prestigious NSF Faculty Early Career Development (CAREER)
Program award.
For appointments, please contact Bob Collins.
The Robotics Institute is part of the School of Computer Science, Carnegie Mellon University.
