General framework for human object detection and pose estimation in video sequences

Abstract

This paper presents a real-time object detection and pose estimation system. The main idea is to unify the detection and pose estimation processes into a tree classifier. The tree classifier uses Haar-like feature and has been trained using a boosting algorithm with a pose estimation step. The estimation step has been used only when the positive samples were not homogeneous and when the splitting improves the discriminative power compared to a single monolithic node classifier and has lower computational complexity.