A Framework for Multi-Pose Face Detection and Tracking

Abstract

Face detection and tracking is a key technology used in many kinds of video applications. In this paper we present a framework for multi-pose face detection and tracking by integrating the target tracking and the face detection in video sequences. Our approach involves three modules for face detection in a video: a Condensation module for face tracking, a template module for measuring the observation process in Condensation, and a correction module for restarting the tracking if the tracked face has been lost. The Condensation module handles the basic tasks of face tracking using a dynamic process to estimate the face position in video frames. The template module involves a checklist scheme to deal with the change of face poses. The correction module provides a revision process when the tracked face is lost. We also present several experiments of face detection and tracking to show the performance of our proposed approach.