Vision-based human motion analysis for the development of automated rehabilitation tools

Patients requiring physical rehabilitation, such as those suffering long-term chronic disease or recovering from illness or injury are reliant upon rehabilitation programmes to recover. Tele-rehabilitation has been proposed as a promising development to remove physical barriers of locality and place rehabilitation services within the patients home. We propose an interdisciplinary research approach to generate an assistive rehabilitative platform through the development of computer vision techniques, particularly near-real-time motion and gesture analysis. In this paper, we present our initial findings on the comparison of two popular machine learning algorithms, Support Vector Machines and Random Forests for their ability to accurately predict periodic kinematic actions.