Using decision trees to recognize visual events

ABSTRACT

This paper presents a classifier-based approach to recognize events in video surveillance sequences. The aim of this work is to propose a generic event recognition system that can be used without relying on a long-term tracking procedure. It is composed of three stages. The first one aims at defining and building a set of relevant features from the foreground objects. Second, a clustering tree-based method is used to handle the features and aggregate them locally in a set of coarse to fine activity patterns. Finally, events are modeled as a sequence of structured patterns with an ensemble of randomized trees. In particular, we want this classifier to discover the temporal and causal correlations between the most discriminative patterns. Our system is tested on simulated events and in a real world context with the CAVIAR video sequences dataset. Preliminary results demonstrate the effectiveness of the proposed framework for event recognition in automated visual surveillance applications. We also prove that more flexible algorithms (i.e. deterministic classifiers) rather than probabilistic graph models are conceivable for video events analysis.

REFERENCES

• 1: http://www.tele.ucl.ac.be/~csimon/trajectories.html.
• 2: Yali Amit , Donald Geman, Shape quantization and recognition with randomized trees, Neural Computation, v.9 n.7, p.1545-1588, Oct. 1, 1997 [doi>10.1162/neco.1997.9.7.1545]
• 3: Hendrik Blockeel , Luc De Raedt , Jan Ramon, Top-Down Induction of Clustering Trees, Proceedings of the Fifteenth International Conference on Machine Learning, p.55-63, July 24-27, 1998
• 4: Aaron F. Bobick , James W. Davis, The Recognition of Human Movement Using Temporal Templates, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.23 n.3, p.257-267, March 2001 [doi>10.1109/34.910878]
• 5: Leo Breiman, Random Forests, Machine Learning, v.45 n.1, p.5-32, October 1 2001 [doi>10.1023/A:1010933404324]
• 6: H. Buxton. Learning and understanding dynamic scene activity: a review. Image and Vision Computing, 21(1):125--136, 2003.
• 7: Hannah M. Dee , Sergio A. Velastin, How close are we to solving the problem of automated visual surveillance?: A review of real-world surveillance, scientific progress and evaluative mechanisms, Machine Vision and Applications, v.19 n.5-6, p.329-343, September 2008 [doi>10.1007/s00138-007-0077-z]
• 8: A. R. Dick and M. J. Brooks. Issues in automated visual surveillance. In Proceeding of VIIth digital image computing: technique and applications, pages 195--204, Santorini, 2003.
• 9: Pierre Geurts , Damien Ernst , Louis Wehenkel, Extremely randomized trees, Machine Learning, v.63 n.1, p.3-42, April 2006 [doi>10.1007/s10994-006-6226-1]
• 10: Mohammed Waleed Kadous , Claude Sammut, Classification of Multivariate Time Series and Structured Data Using Constructive Induction, Machine Learning, v.58 n.2-3, p.179-216, February 2005 [doi>10.1007/s10994-005-5826-5]
• 11: Ivan Laptev, On Space-Time Interest Points, International Journal of Computer Vision, v.64 n.2-3, p.107-123, September 2005 [doi>10.1007/s11263-005-1838-7]
• 12: Dar-Shyang Lee, Effective Gaussian Mixture Learning for Video Background Subtraction, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.27 n.5, p.827-832, May 2005 [doi>10.1109/TPAMI.2005.102]
• 13: Gérard Medioni , Isaac Cohen , François Brémond , Somboon Hongeng , Ramakant Nevatia, Event Detection and Analysis from Video Streams, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.23 n.8, p.873-889, August 2001 [doi>10.1109/34.946990]
• 14: Nuria M. Oliver , Barbara Rosario , Alex P. Pentland, A Bayesian Computer Vision System for Modeling Human Interactions, IEEE Transactions on Pattern Analysis and Machine Intelligence, v.22 n.8, p.831-843, August 2000 [doi>10.1109/34.868684]
• 15: Ó. Pérez, M. Piccardi, J. Garcıa, and J. M. Molina. Comparison of classifiers for human activity recognition. In IWINAC (2), pages 192--201, 2007.
• 16: P. Remagnino, T. Tan, and K. Baker. Multi-agent visual surveillance of dynamic scenes. Image Vision Computing, 16(8):529--532, 1998.
• 17: P. Ribeiro and J. Santos-Victor. Human activity recognition from video: modeling, feature selection and classification arquitecture. In HAREM 2005: International Workshop on Human Activity Recognition and Modelling, Oxford, UK, 2005.
• 18: C. Simon , J. Meessen , D. Tzovaras , C. De Vleeschouwer, Using Decision Trees for Knowledge-Assisted Topologically Structured Data Analysis, Proceedings of the Eight International Workshop on Image Analysis for Multimedia Interactive Services, p.2, June 06-08, 2007 [doi>10.1109/WIAMIS.2007.84]
• 19: Louis A. Wehenkel, Automatic Learning Techniques in Power Systems, Kluwer Academic Publishers, Norwell, MA, 1998
• 20: Tao Xiang , Shaogang Gong, Beyond Tracking: Modelling Activity and Understanding Behaviour, International Journal of Computer Vision, v.67 n.1, p.21-51, April 2006 [doi>10.1007/s11263-006-4329-6]
• 21: A. Yilmaz and M. Shah. Actions as objects: A novel action representation. In proc. of IEEE International Conference on Computer Vision and Pattern Recognition (CVPR 05), pages 178--185, 2005.

INDEX TERMS

Primary Classification:
  I. Computing Methodologies
  I.2 ARTIFICIAL INTELLIGENCE
        I.2.10 Vision and Scene Understanding
            Subjects: Video analysis

General Terms:
Design, Experimentation

Keywords:
activity recognition, automated visual surveillance system, randomized decision trees