Recovering people tracking errors using enhanced covariance-based signatures

Julien Badie 1, * Slawomir Bak 1 Silviu-Tudor Serban 1 François Bremond 1
* Corresponding author
1 STARS - Spatio-Temporal Activity Recognition Systems
CRISAM - Inria Sophia Antipolis - Méditerranée
Abstract : This paper presents a new approach for tracking multiple persons in a single camera. This approach focuses on re- covering tracked individuals that have been lost and are detected again, after being miss-detected (e.g. occluded) or after leaving the scene and coming back. In order to correct tracking errors, a multi-cameras re-identification method is adapted, with a real-time constraint. The proposed ap- proach uses a highly discriminative human signature based on covariance matrix, improved using background subtrac- tion, and a people detection confidence. The problem of linking several tracklets belonging to the same individual is also handled as a ranking problem using a learned pa- rameter. The objective is to create clusters of tracklets de- scribing the same individual. The evaluation is performed on PETS2009 dataset showing promising results.
Document type :
Conference papers
Fourteenth IEEE International Workshop on Performance Evaluation of Tracking and Surveillance - 2012, Jul 2012, Beijing, China. pp.487-493, 2012, <10.1109/AVSS.2012.90>
https://hal.inria.fr/hal-00761322
Contributor : Julien Badie <>
Submitted on : Wednesday, December 5, 2012 - 11:56:46 AM
Last modification on : Tuesday, January 22, 2013 - 4:16:19 PM
Document(s) archivé(s) le : Wednesday, March 6, 2013 - 5:45:09 AM

File

4797a487.pdf
Files produced by the author(s)

Identifiers

Collections

INRIA

Citation

Julien Badie, Slawomir Bak, Silviu-Tudor Serban, François Bremond. Recovering people tracking errors using enhanced covariance-based signatures. Fourteenth IEEE International Workshop on Performance Evaluation of Tracking and Surveillance - 2012, Jul 2012, Beijing, China. pp.487-493, 2012, <10.1109/AVSS.2012.90>. <hal-00761322>

Export

BibTeX | XML | endNote

Share

Metrics

Record views

150

Document downloads

73

Contact


archive-ouverte@inria.fr