| Syllabus / LearningBitByBitITPFacesLab

ITP faces

Download images.zip and pyfacesCmdLineVer.zip eigenface library from pyfaces: http://code.google.com/p/pyfaces/downloads/list

Extract files and read the usage instructions.
Test functionality as described in usage instructions by handpicking a few images from the probe set and comparing with the gallery folder.
Experiment with different parameters and compare the accuracy of the recognition result.
Write a frontend program to automate the process of testing each image in the probe set and evaluating its accuracy.

Advanced:

Download ITP face set of first year students from 2010: here.
Download the complete set of Rice images for facial recognition here.

Normalize a subset of the ITP facial images converting images to grayscale and resizing to 300x250 pixels. This will be the ITP gallery. Take the same subset and distort the images to create a test or probe set. Try adding noise, distortion, changing colors, etc.

Run eigenface analysis on both the ITP and Rice face sets using the pyfaces source code. Analyze how the program does on different sets of faces and why.

Going further:

Ask me for the FERET database of professional facial recognition images. Write a program to evaluate recognition accuracy on the 3 facial sets.
More info about FERET:
http://en.wikipedia.org/wiki/FERET_database
http://www.frvt.org/FERET/default.htm
http://www.itl.nist.gov/iad/humanid/feret/feret_master.html