Project 1: Statistical and Neural Pattern Recognition

Please CLICK to view project statement

RESOURCES

  • Dataset 1 (see instructions)
  • Dataset that contains images with variations in Pose (68 individuals across 13 poses).
  • Dataset that contains images with variations in illumination (Many thanks to Aniruddha Kembhavi for compiling the set) (68 individuals across 21 Illumination).
  • YALE Dataset.
  • PIE Dataset.
  • M. Turk, A. Pentland, "Eigenfaces for recognition," Journal of Cognitive Neuroscience, vol. 3, pp 72-86, 1991.
  • P. Belhumeur, J. Hespanha, and D. Kriegman, "Eigenfaces vs. Fisherfaces: Recognition Using Class Specific Linear Projection," IEEE Trans. PAMI, vol. 19, pp. 711-720, 1997.
  • K. Etemad and R. Chellappa, "Discriminant Analysis for Recognition of Human Face Images," Journal of Optical Society of America A, pp. 1724-1733, 1997.
  • R. Chellappa, C. Wilson, and S. Sirohey, "Human and Machine Recognition of Faces: A Survey," Proceedings of IEEE, vol. 83, pp. 705-740, 1995.
  • W. Zhao, R. Chellappa, A. Rosenfeld, and J. Phillips, "Face Recognition: A Literature Survey," to appear ACM computing surveys, 2003.

    • FAQ (suitably modified) from Shaohua's website

    Q: I have questions or doubts on this project, what should I do?
    A: 1) Send your email to Prof. Chellappa (rama AT cfar DOT umd DOT edu) and cc Aswin (aswch ZAT cfar ZOT umd ZOT edu).
    2) Visit us during office hours.

    Q: How can I train FDA since there are so few samples per class i.e., within-class scatter is not robust?
    A: Read [Etemad and Chellappa, JOSA 97]. In this paper, they increase the samples for one class by constructing a mirror image, and a noisy image.

    Q: Why are my recognition rates so low?
    A: This project is not meant to be a contest. So, low recognition rates are acceptable. However, you are expected to explain a little bit about your low rates. Of course, if you can figure out a smart way to boost your rates, you can definitely publish a paper on it.

    Q: Can you specify more on the test scenarios?
    A: This is where you can use your imagination. However, a general recipe is provided in [Zhao et. al. ACM 2003]. Refer to the part on the FERET test.