Project 1

Home Course Information PR Links Projects Homework

Project 1: Face Recognition using supervised pattern recognition methods

In this project, you will create a classifier for a face recognition system. This classifier must be able to recognize the faces in images in
its training set, recognize when a face is not in the dataset, and recognize when there is not a face in the image.

Project Assumptions:

bulletImages are grayscale.
bulletAll images are the same size and have been processed to make sure that there the face is in the center of the image and the picture was taken at the same range.
 

Part 1: Feature Reduction

 The feature set will be the eigenfaces with different numbers of eigenvalues. Use at least one additional paper for ideas. For extra
credit, try LDA or ICA methods for feature extraction as well.

Part 2: Pattern Recognition and Training

Design two classifiers to separate the face features (one linear, one
nonlinear, use cross validation. Validate the data on the test set. Give a confusion matrix.

Part 3: Project write-up (in the form of a website or an electronic document)

  1. Abstract or Executive summary: Summary of what will be found in your
    project and any interesting conclusions or things that you learned.
     
  2. Introduction: Introduces the topic that you are covering and what
    issues you are looking at, what is the purpose of the system you are
    analyzing?
     
  3. Background: How does your system work? Why is it done this way?
     
  4. Analysis or comparison- is there more than one way to solve the problem?
  5. Simulation Results;
  6. Conclusion
  7.  References: give websites, articles and books that you used in your
    report with comments on their content and usefulness

Part 4: Data Sets and Example Code

Project Data

The training data set will be a subset of the Yale Face DatabaseB. Contains 5760 single light source images of 10 subjects each seen under 576 viewing conditions (9 poses x 64 illumination conditions). For every subject in a particular pose, an image with ambient (background) illumination was also captured.  (Ref: Georghiades, A.S. and Belhumeur, P.N. and Kriegman, D.J., "From Few to Many: Illumination Cone Models for Face Recognition under  Variable Lighting and Pose", IEEE Trans. Pattern Anal. Mach. Intelligence, 23:6, 643-660,  2001). Project Data Set

In-Class Demo and Code
bulletYale Face Database The Yale Face Database (size 6.4MB) contains 165 grayscale images in GIF format of 15 individuals. There are 11 images per subject, one per different facial expression or configuration: center-light, w/glasses, happy, left-light, w/no glasses, normal, right-light, sad, sleepy, surprised, and wink. (images)
bulletMatlab Code: 
bulletReadinfiles.m: Reads in the images in a directory, finds mean, finds eigenvectors and values
bulletclassify_im.m Finds projection of test image onto face space

References and further information

bullet

Central Repository of information on Face Recognition: www.face-rec.org

bullet

Key Tutorial Papers (shorter paper is probably a better introduction)
bullet

M.A. Turk, A.P. Pentland, Face Recognition Using Eigenfaces, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 3-6 June 1991, Maui, Hawaii, USA, pp. 586-591
download here, 596 kB

bullet

M. Turk, A. Pentland, Eigenfaces for Recognition, Journal of Cognitive Neurosicence, Vol. 3, No. 1, 1991, pp. 71-86
download here, 10.6 MB

bullet

Data Set Reference: Georghiades, A.S. and Belhumeur, P.N. and Kriegman, D.J., "From Few to Many: Illumination Cone Models for Face Recognition under  Variable Lighting and Pose", IEEE Trans. Pattern Anal. Mach. Intelligence, 23:6, 643-660,  2001

horizontal rule

Page last edited 03/03/2006