Next: The tracker Up: The acquisition of prototypes Previous: The acquisition of prototypes Index

Introduction

In order to study behaviour, we need a model of motion. As the motion of objects like faces are usually complex, learning is the only efficient way of obtaining a model of motion. In order to learn motion, we need a set of typical motion from sequences. Thus, we need to be able to track a face in a video sequence. This chapter describes the face tracker that we are going to use. It also describes how movements can be split into prototypes in order to be learned later by a variable length Markov model.

franck 2006-10-16