Spatio-Temporal Motion Pattern Modeling of Extremely Crowded Scenes

Abstract : The abundance of video surveillance systems has created a dire need for computational methods that can assist or even replace human operators. Research in this field, however, has yet to tackle an important real-world scenario: extremely crowded scenes. The excessive amount of people and their activities in extremely crowded scenes present unique challenges to motion-based video analysis. In this paper, we present a novel statistical framework for modeling the motion pattern behavior of extremely crowded scenes. We construct a rich yet compact representation of the local spatio-temporal motion patterns and model their temporal behaviors with a novel, distribution-based Hidden Markov Model (HMM), exploiting the underlying statistical characteristics of the scene. We demonstrate that, by capturing the steady-state behavior of a scene, we can naturally detect unusual events as unlikely motion pattern variations. The experiments show promising results in extremely crowded real-world scenes with complex activities that are hard for even human observers to analyze.
Document type :
Conference papers
The 1st International Workshop on Machine Learning for Vision-based Motion Analysis - MLVMA'08, Oct 2008, Marseille, France. 2008
Liste complète des métadonnées

Cited literature [27 references]  Display  Hide  Download

https://hal.inria.fr/inria-00326717
Contributor : Peter Sturm <>
Submitted on : Sunday, October 5, 2008 - 12:35:53 PM
Last modification on : Monday, October 6, 2008 - 9:42:30 AM
Document(s) archivé(s) le : Monday, October 8, 2012 - 1:56:32 PM

File

mlvma08_submission_20.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : inria-00326717, version 1

Collections

Citation

Louis Kratz, Ko Nishino. Spatio-Temporal Motion Pattern Modeling of Extremely Crowded Scenes. The 1st International Workshop on Machine Learning for Vision-based Motion Analysis - MLVMA'08, Oct 2008, Marseille, France. 2008. 〈inria-00326717〉

Share

Metrics

Record views

455

Files downloads

310