DocumentCode :
3005522
Title :
Locally time-invariant models of human activities using trajectories on the grassmannian
Author :
Turaga, Pavan ; Chellappa, Rama
Author_Institution :
Center for Autom. Res., Univ. of Maryland, College Park, MD, USA
fYear :
2009
fDate :
20-25 June 2009
Firstpage :
2435
Lastpage :
2441
Abstract :
Human activity analysis is an important problem in computer vision with applications in surveillance and summarization and indexing of consumer content. Complex human activities are characterized by non-linear dynamics that make learning, inference and recognition hard. In this paper, we consider the problem of modeling and recognizing complex activities which exhibit time-varying dynamics. To this end, we describe activities as outputs of linear dynamic systems (LDS) whose parameters vary with time, or a time-varying linear dynamic system (TV-LDS). We discuss parameter estimation methods for this class of models by assuming that the parameters are locally time-invariant. Then, we represent the space of LDS models as a Grassmann manifold. Then, the TV-LDS model is defined as a trajectory on the Grassmann manifold. We show how trajectories on the Grassmannian can be characterized using appropriate distance metrics and statistical methods that reflect the underlying geometry of the manifold. This results in more expressive and powerful models for complex human activities. We demonstrate the strength of the framework for activity-based summarization of long videos and recognition of complex human actions on two datasets.
Keywords :
computer vision; image recognition; indexing; parameter estimation; statistical analysis; video surveillance; Grassmann manifold; activity-based summarization; computer vision; consumer content indexing; distance metrics; human action recognition; human activity analysis; local time-invariant model; parameter estimation method; statistical method; surveillance; time-varying linear dynamic system; Application software; Character recognition; Computer vision; Humans; Indexing; Nonlinear dynamical systems; Parameter estimation; Power system modeling; Surveillance; Time varying systems;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Vision and Pattern Recognition, 2009. CVPR 2009. IEEE Conference on
Conference_Location :
Miami, FL
ISSN :
1063-6919
Print_ISBN :
978-1-4244-3992-8
Type :
conf
DOI :
10.1109/CVPR.2009.5206710
Filename :
5206710
Link To Document :
بازگشت