Towards Optimal Training of Cascaded Detectors

  • S. Charles Brubaker
  • Matthew D. Mullin
  • James M. Rehg
Part of the Lecture Notes in Computer Science book series (LNCS, volume 3951)


Cascades of boosted ensembles have become popular in the object detection community following their highly successful introduction in the face detector of Viola and Jones [1]. In this paper, we explore several aspects of this architecture that have not yet received adequate attention: decision points of cascade stages, faster ensemble learning, and stronger weak hypotheses. We present a novel strategy to determine the appropriate balance between false positive and detection rates in the individual stages of the cascade based on a probablistic model of the overall cascade’s performance. To improve the training time of individual stages, we explore the use of feature filtering before the application of Adaboost. Finally, we show that the use of stronger weak hypotheses based on CART can significantly improve upon the standard face detection results on the CMU-MIT data set.


False Positive Rate Operating Point Object Detection Training Time Face Detection 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Viola, P., Jones, M.J.: Robust real-time face detection. Int. J. Comput. Vision 57(2), 137–154 (2004)CrossRefGoogle Scholar
  2. 2.
    Viola, P.A., Jones, M.J., Snow, D.: Detecting pedestrians using patterns of motion and appearance. In: Proc. ICCV., vol. 2, pp. 734–741 (2003)Google Scholar
  3. 3.
    Schneiderman, H., Kanade, T.: Object detection using the statistics of parts. Int. J. Comput. Vision 56(3), 151–177 (2004)CrossRefGoogle Scholar
  4. 4.
    Freund, Y., Schapire, R.E.: A decision-theoretic generalization of on-line learning and an application to boosting. J. Comput. Syst. Sci. 55(1), 119–139 (1997)MathSciNetCrossRefzbMATHGoogle Scholar
  5. 5.
    Breiman, L., Friedman, J., Olshen, R., Stone, C.: Classification and Regression Trees. Wadsworth and Brooks, Monterey (1984)zbMATHGoogle Scholar
  6. 6.
    Viola, P., Jones, M.: Fast and robust classification using asymmetric AdaBoost and a detector cascade. In: NIPS 14, pp. 1311–1318 (2002)Google Scholar
  7. 7.
    Sun, J., Rehg, J.M., Bobick, A.F.: Automatic cascade training with perturbation bias. In: CVPR, vol. (2), pp. 276–283 (2004)Google Scholar
  8. 8.
    Luo, H.: Optimization design of cascaded classifiers. In: CVPR, vol. (1), pp. 480–485 (2005)Google Scholar
  9. 9.
    Sochman, J., Matas, J.: Waldboost-learning for time constrained sequential detection. In: CVPR, vol. (2), pp. 150–157 (2005)Google Scholar
  10. 10.
    Wu, J., Rehg, J.M., Mullin, M.D.: Learning a rare event detection cascade by direct feature selection. In: NIPS 16, pp. 1523–1530 (2004)Google Scholar
  11. 11.
    Lienhart, R., Kuranov, A., Pisarevsky, V.: Empirical analysis of detection cascades of boosted classifiers for rapid object detection. In: Michaelis, B., Krell, G. (eds.) DAGM 2003. LNCS, vol. 2781, pp. 297–304. Springer, Heidelberg (2003)CrossRefGoogle Scholar
  12. 12.
    Wu, J., Mullin, M., Rehg, J.: Linear asymmetric classifier for cascade detectors. In: Proc. 22nd International Conference on Machine Learning, pp. 993–1000 (2005)Google Scholar
  13. 13.
    Li, S.Z., Zhang, Z.Q.: Floatboost learning and statistical face detection. IEEE Trans. on PAMI 26(9), 1112–1123 (2004)CrossRefGoogle Scholar
  14. 14.
    Liu, C., Shum, H.Y.: Kullback-leibler boosting. In: CVPR, vol. (1), pp. 587–594 (2003)Google Scholar
  15. 15.
    Xiao, R., Zhu, L., Zhang, H.J.: Boosting chain learning for object detection. In: Proc. ICCV, vol. 1, pp. 709–715 (2003)Google Scholar
  16. 16.
    Levi, K., Weiss, Y.: Learning object detection from a small number of examples: The importance of good features. In: CVPR, vol. (2), pp. 53–60 (2004)Google Scholar
  17. 17.
    Brubaker, S.C., Wu, J., Sun, J., Mullin, M.D., Rehg, J.M.: On the design of cascades of boosted ensembles for face detection. Technical Report GIT-GVU-05-28, Georgia Institute of Technology (2005)Google Scholar
  18. 18.
    Yang, M.H., Kriegman, D.J., Ahuja, N.: Detecting faces in images: A survey. IEEE Trans. on PAMI 24(1), 34–58 (2002)CrossRefGoogle Scholar
  19. 19.
    Gupta, A.K., Nadarajah, S. (eds.): Handbook of Beta Distribution and its applications. Marcel Dekker, Inc., New York (2004)zbMATHGoogle Scholar
  20. 20.
    Vidal-Naquet, M., Ullman, S.: Object recognition with informative features and linear classification. In: Proc. ICCV, pp. 281–288 (2003)Google Scholar
  21. 21.
    Fleuret, F.: Fast binary feature selection with conditional mutual information. Journal of Machine Learning Research 5, 1531–1555 (2004)MathSciNetzbMATHGoogle Scholar
  22. 22.
    Opelt, A., Fussenegger, M., Pinz, A., Auer, P.: Weak hypotheses and boosting for generic object detection and recognition. In: Pajdla, T., Matas, J(G.) (eds.) ECCV 2004. LNCS, vol. 3022, pp. 71–84. Springer, Heidelberg (2004)CrossRefGoogle Scholar
  23. 23.
    Grossmann, E., Kale, A., Jaynes, C.: Towards interactive generation of ”ground-truth” in background subtraction from partially labeled examples. In: Proc. ICCV VS-PETS workshop (2005)Google Scholar
  24. 24.
    Guyon, I., Elisseeff, A.: An introduction to variable and feature selection. Journal of Machine Learning Research 3, 1157–1182 (2003)zbMATHGoogle Scholar
  25. 25.
    Schneiderman, H.: Feature-centric evaluation for efficient cascaded object detection. In: CVPR, vol. (2), pp. 29–36 (2004)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2006

Authors and Affiliations

  • S. Charles Brubaker
    • 1
  • Matthew D. Mullin
    • 1
  • James M. Rehg
    • 1
  1. 1.College of Computing and GVU CenterGeorgia Institute of TechnologyAtlantaUSA

Personalised recommendations