Weighted ensemble boosting for robust activity recognition in video

• Autorzy: Ivanov Y. , Hamid R.
• Konferencja: International Conference on Computer Vision and Graphics ICCVG 2006 (25-27.09.2006 ; Warsaw, Poland)
• Języki publikacji: EN

Abstrakty

EN

In this paper we introduce a novel approach to classifier combination, which we term Weighted Ensemble Boosting. We apply the proposed algorithm to the problem of activity recognition in video, and compare its performance to different classifier combination methods. These include Approximate Bayesian Combination, Boosting, Feature Stacking, and the more traditional Sum and Product rules. Our proposed Weighted Ensemble Boosting algorithm combines the Bayesian averaging strategy with the boosting framework, finding useful conjunctive feature combinations and achieving a lower error rate than the traditional boosting algorithm. The method demonstrates a comparable level of stability with respect to the classifier selection pool. We show the performance of our technique for a set of 6 types of classifiers in an office setting, detecting 7 classes of typical office activities.

Słowa kluczowe

EN

video analysis; activity recognition; boosting; ensemble methods

• Wydawca: Institute of Information Technology of the Warsaw University of Life Sciences – SGGW
• Czasopismo: Machine Graphics and Vision
• Rocznik: 2006
• Tom: Vol. 15, No. 3/4
• Strony: 415--427
• Opis fizyczny: Bibliogr. 15 poz., il., tab., wykr.

Twórcy

autor

Ivanov Y.

autor

Hamid R.

• Mitsubishi Electric Research Labs, 201 Brodway, Cambridge, MA 02139, USA,

Bibliografia

• [1] Horn Berthold K. P., Schunck Brian G.: Determining optical flow. Artificial Intelligence, 17:185-203, 1981.
• [2] Bobick A. F.: Movement, activity, and action: The role of knowledge in the perception of motion. In Philosophical Transactions Royal Society London B. Royal Philosophical Soviety, 1997.
• [3] Preund Yoav, Schapire E. Robert: A decision-theoretic generalization of on-line learning and an application to boosting. Computer and System Sciences, 55(1): 119-139, 1997.
• [4] Kittler J., Li Y. P., Matas J., Sanchez Ramos M. U.: Combining evidence in multimodal personal identity recognition systems. Intl . Conference on Audio- and Video-Based Biométrie Authentication, Crans Montana, Switzerland, 1997.
• [5] Morris R. J., Hogg D. C.: Statistical models of object interaction. In Workshop on Visual Surveillance, Bombay, India, IEEE, 1998.
• [6] Aggarwal J. K., Cai Q: Human motion analysis: A review. Computer Vision and Image Understanding, 73(3):428-440, 1999.
• [7] Bilmes J., Kirchhoff K.: Directed graphical models of classifier combination: Application to phone recognition. Intl. Conference on Spoken Language Processing, Beijing, China, 2000.
• [8] Comaniciu D., Ramesh V., Meer P.: Real-time tracking of nonrigid objects using mean shift. CVPR, pages 673-678, 2000.
• [9] Viola P., Jones M.: Robust real-time object recognition. In ICCV, Vancouver, Canada, 2001.
• [10] Pekalska E., Duin R., Skurichina M.: A discussion on the classifier projection space for classifier combining. 3rd International Workshop on Multiple Classifier Systems, pages 137-148, Cagliari, Italy, Springer Ver lag, 2002.
• [11] Ross Arun, Jain Anil K.: Information fusion in biometrics. Pattern Recognition Letters, Vol. 24, Issue 13, pp. 2115-2125, 24:2115-2125, Sep.
• [12] Viola P., Jones M. J., Snow D.: Detecting pedestrians using patterns of motion and appearance. In ICCV, pages 734-741, 2003.
• [13] Ivanov Y.: Multi-modal human identification system. In Workshop on Applications of Computer Vision, Breckenridge, CO.
• [14] Ivanov Y., Heisele B., Serre T.: Using component features for face recognition. In International Conference on Automatic Face ang Gesture Recognition, Seoul, Korea, 2004.
• [15] Smith P., Lobo N., Shah M.: Temporal boost for event recognition. In ICCV, Beijing, China, 2005.