A Hierarchical System for Recognition, Tracking and Pose Estimation, Cognitive Vision Systems
This chapter presents a system for the recognition, tracking and pose estimation of people in video sequences. It is based on a careful selection of Haar wavelet features and uses Support Vector Machines (SVM) in spaces of reduced dimensionality for classification. Recognition is carried out hierarchically by using a set of detectors and discriminators for people and poses. The characteristic fea- tures used in the individual nodes are learned automatically. Tracking is solved via a particle filter that utilizes the SVM output and a first order kinematic model to obtain a robust scheme that successfully handles occlusion, different poses and camera zooms.