|
|
|
|
|
• By Date • By Publication Type • By Research Category • By Author (Currently at CMU) • By Author (Past and Current CORAL Contributors) •
Raju Patil, Paul Rybski, Takeo Kanade, and Manuela Veloso. People Detection and Tracking in High Resolution Panoramic Video Mosaic. In Proceedings of the 2004 IEEE International Conference on Intelligent Robots and Systems (IROS), pp. 1323–1328, Sendai, Japan, October 2004.
(unavailable)
We are interested in the challenging scientific pursuit of how tocharacterize human activities in formatl meeting situations bytracking their positions from a computer vision system. We present ahuman activity recognition system that works within the framework ofCAMEO (the Camera Assisted Meeting Event Observer) a vision systemdesigned to operate in real-time and in uncalibrated environments.The activity classifier models human activity with a Dynamic BayesianNetwork and characterizes actions using an implementation of theViterbi algorithm modified for real-time operation through the use ofa fixed-sized backtracking window. Experimental results of theclassifier operating on tracked person data are presented. Ananalysis of the tradeoffs of choosing different window sizes is alsoincluded.
@InProceedings{04iros-cameo-tracking,
author = "Raju Patil and Paul Rybski and Takeo Kanade and Manuela Veloso",
title = "People Detection and Tracking in High Resolution Panoramic Video Mosaic",
booktitle = "Proceedings of the 2004 IEEE International Conference on Intelligent Robots and Systems (IROS)",
pages = "1323-1328",
year = 2004,
address = "Sendai, Japan",
month = "October",
abstract = "We are interested in the challenging scientific pursuit of how to
characterize human activities in formatl meeting situations by
tracking their positions from a computer vision system. We present a
human activity recognition system that works within the framework of
CAMEO (the Camera Assisted Meeting Event Observer) a vision system
designed to operate in real-time and in uncalibrated environments.
The activity classifier models human activity with a Dynamic Bayesian
Network and characterizes actions using an implementation of the
Viterbi algorithm modified for real-time operation through the use of
a fixed-sized backtracking window. Experimental results of the
classifier operating on tracked person data are presented. An
analysis of the tradeoffs of choosing different window sizes is also
included.
",
bib2html_pubtype = {Refereed Conference},
bib2html_rescat = {CAMEO},
}
Generated by bib2html.pl (written by Patrick Riley ) on Tue Oct 09, 2007 00:00:14
|
|
 | |||
| [home] [news] [people] [robosoccer] [projects] [multimedia] [download] [publications] [contact us] | |||