Time Series Modeling for Activity Prediction

Abstract

Human Motion Analysis (HMA) is a highly interdisciplinary research area which attracts great interests from computer vision, machine learning, multimedia, and medical research communities, due to the potential applications ranging from, human–computer interaction, security (intelligent surveillance), health (assistive clinical studies), information technology (content-based video retrieval), entertainments (special effects in film production and somatosensory game) to all aspects of our daily life. As an important research thread of HMA, action recognition builds the basis for all of the above-mentioned applications.