If you are given a photograph that includes one or more faces, you seldom have trouble pointing to the faces, or counting the faces, or even recognizing whose faces they are. If you see a video with people in it, your eyes can track the movements of the people's faces, identifying without difficulty the changing positions of the eyes, the mouth, and the ears. Even if the people are only partially visible, as in an aerial shot of the Boston Marathon, you can count with reasonable accuracy the number of heads in the picture.
The problems of face tracking and face recognition have been studied extensively, with some good and some less good results. As you might imagine, law enforcement departments and the Department of Homeland Security can think of applications of a good face recognition system. A system of biometrics combining face recognition, voice recognition, etc., might be usable for providing access to computer services or to physical locations. And a face-counting or head-counting system could be used to help estimate the number of people attending an outdoor concert or demonstration.
For this project, you will implement several face-related algorithms, and develop a GUI environment to allow people to experiment with various algorithms and their parameters. The pieces of your project may include:
Reliable Face Recognition Methods: System Design, Implementation and Evaluation, by Harry Wechsler