An unsupervised approach to detecting and isolating athletic movements

To enable automatic analysis of athletic movement, the first task is to recognize the athletic movements to be analyzed from a continuous motion data stream. Automated detection of athletic movement and the isolation of the recruited body parts would enable the analysis of sporting movements for improving sports performance and preventing possible injuries. In this paper, an unsupervised method for detecting and isolating athletic movements is proposed. Given motion capture data, the method automatically identifies when athletic movements are being performed and the body parts involved using the concepts of the manipulability and kinematic dimensionality reduction. Experiments demonstrate the ability of the proposed approach to detect and isolate athletic movements from a variety of motion data. 

Autor / Fonte:Terry Taewoong Um, Dana Kulic, Terry Taewoong Um, Dana Kulic, Dana Kulic Conference Proceedings: Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2016, 2016: 6268-6272