Face Detection Using an SVM Trained in Eigenfaces Space

  • Vlad Popovici
  • Jean-Philippe Thiran
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 2688)


The central problem in the case of face detectors is to build a face class model. We present a method for face class modeling in the eigenfaces space using a large-margin classifier like SVM. Two main issues are addressed: what is the required number of eigenfaces to achieve a good classification rate and how to train the SVM for a good generalization. As the experimental evidence show, generally one needs less eigenfaces than usually considered. We will present different strategies for choosing the dimensionality of the PCA space and discuss their effectiveness in the case of face-class modeling.


Face Recognition Bayesian Information Criterion Face Detection Polynomial Kernel Latent Dimensionality 
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.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Hotteling, H.: Analysis of a complex of statistical variables into principal components. Journal of Educational Psychology (1933) 417–441, 498–520Google Scholar
  2. [2]
    Sirovich, L., Kirby, M.: Low-dimensional procedure for the characterization of human faces. Journal of the Optical Society of America A 4 (1987) 519–524CrossRefGoogle Scholar
  3. [3]
    Turk, M. A., Pentland, A.P.: Face recognition using eigenfaces. In: Proc. IEEE Conf. on Computer Vision and Pattern Recognition. (1991) 586–591Google Scholar
  4. [4]
    Zhao, W., Chellappa, R., Krishaswamy, A.: Discriminant analysis of principal components for face recognition. In: Proceedings of the 3rd International Conference on Automatic Face and Gesture Recognition. (1998)Google Scholar
  5. [5]
    Tipping, M.E., Bishop, C.M.: Probabilistic principal component analysis. Technical Report NCRG/97/010, Neural Computing Research Group, Dept. of Computer Science and Applied Mathematics, Aston University (1997)Google Scholar
  6. [6]
    Moghaddam, B., Pentland, A.: Probabilistic visual learning for object detection. In: Proc. of the 5th International Conference on Computer Vision. (1995) 786–793Google Scholar
  7. [7]
    Minka, T. P.: Automatic choice of dimensionality for PCA. Technical Report 514, M. I. T. Media Laboratory Perceptual Computing Section (2000)Google Scholar
  8. [8]
    Turk, M., Pentland, A.: Eigenfaces for recognition. Journal of Cognitive Neuroscience 3 (1991) 71–86CrossRefGoogle Scholar
  9. [9]
    Vapnik, V.: The Nature of Statistical Learning Theory. Springer Verlag (1995)Google Scholar
  10. [10]
    Cristianini, N., Shawe-Taylor, J.: An Introduction to Support Vector Machines and Other Kernel-Based Learning Methods. Cambridge University Press (2000)Google Scholar
  11. [11]
    Bengio, S., Bimbot, F., Mariéthoz, J., Popovici, V., Porée, F., Bailly-Baillière, E., Matas, G., Ruiz, B.: Experimental protocol on the BANCA database. IDIAPRR 05, IDIAP (2002)Google Scholar
  12. [12]
    Sung, K., Poggio, T.: Example-based learning for view-based human face detection. IEEE Transaction on Pattern Analysis and Machine Intelligence 20 (1998) 39–51CrossRefGoogle Scholar
  13. [13]
    Osuna, E., Freund, R., Girosi, F.: Training support vector machines: an application to face detection. In: Proceedings of Intl. Conference on Computer Vision and Pattern Recognition (CVPR). (1997)Google Scholar
  14. [14]
    Penev, P. S., Sirovich, L.: The global dimensionality of face space. In: Proceedings of the 4th Intl. Conference on Automatic Face and Gesture Recognition, IEEE CS (2000) 264–270Google Scholar
  15. [15]
    Roth, V., Steinhage, V.: Nonlinear discriminant analysis using kernel functions. In Solla, S., Leen, T., Müller, K.R., eds.: Advances in Neural Information Processing Systems. Volume 12., MIT Press (1999) 568–574Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2003

Authors and Affiliations

  • Vlad Popovici
    • 1
  • Jean-Philippe Thiran
    • 1
  1. 1.Signal Processing InstituteSwiss Federal Institute of Technology LausanneLausanneSwitzerland

Personalised recommendations