Dynamic 3D Human Shape Modeling for Intention Prediction from Video Imagery

3D human pose and shape analysis from video imagery is an enabling technology for human activity recognition and threat monitoring in military and law enforcement applications. The problem is challenging due to (i) the high dimensionality in human pose space; (ii) lack of depth information in images; and (iii) and the large variability in human body shape, clothing, imaging condition and scene background. We propose an integrated framework that combines discriminative and generative approaches for inferring the 3D motion and anthropometric characteristics of a person. From input images, we first extract feature descriptors that encode the shape and appearance of the subject. From these descriptors, multiple plausible 3D poses are predicted using mixture-of-experts regression. For computational efficiency, we restrict visual inference to low-dimensional embedded space obtained from latent variable models. The 3D pose is then tracked probabilistically across multiple frames using a conditional graphical model. The 3D shape of the various body components are estimated by fitting a generative human model. Finally, the estimated pose and shape are analyzed to detect shape and behavioral abnormalities. We will investigate the feasibility of this approach using publicly available motion capture datasets as well as synthetic videos. BENEFIT: The proposed work addresses the technical challenges of human pose and shape estimation in video imagery. It will enable us to detect human activities and to identify suspicious and hostile human behaviors for military and law enforcement applications. The technology has wide ranging application beyond intelligent surveillance system:  a) Improved human computer interaction The framework will facilitate development of more accurate vision based systems to recognize different gesture and motion in 3D. This has vast potential use in role-playing games where the movements of the user in the physical domain are appropriately reflected as an action in the virtual environment.  b) Movement analysis For identifying the underlying causes for walking abnormalities in clinical patients. The results of gait analysis have been shown to be useful in determining the best course of treatment in these patients.     c) Intelligent training systems for sporting activities The analysis of sports-related movements often entails analyzing a variety of highly dynamic movements. Motion analysis provides the tools for the sports medicine and performance professionals to perform accurate functional evaluations/analyses for clinical and research oriented purposes. d) Realistic animation Cost effective solution to importing realistic body movements in animated characters in videos. e) Robotic locomotion Design of robot appendages and control mechanisms to allow robots to move fluidly and efficiently.