Combining human body shape and pose estimation for robust upper body tracking using a depth sensor

Abstract

Rapid and accurate estimation of a person’s upper body shape and real-time tracking of the pose in the presence of occlusions is crucial for many future assistive technologies, health care applications and telemedicine systems. We propose to tackle this challenging problem by combining data-driven and generative methods for both body shape and pose estimation. Our strategy comprises a subspace-based method to predict body shape directly from a single depth map input, and a random forest regression approach to obtain a sound initialization for pose estimation of the upper body. We propose a model-fitting strategy in order to refine the estimated body shape and to exploit body shape information for improving pose accuracy. During tracking, we feed refinement results back into the forest-based joint position regressor to stabilize and accelerate pose estimation over time. Our tracking framework is designed to cope with viewpoint limitations and occlusions due to dynamic objects.