ABSTRACT
Understanding movement analysis in sports can provide coaches, managers, and athletes with important information for avoiding injury risk, optimizing training menus, supporting strategic decision-making, and evaluating performance. However, existing systems are costly and are only available to a subset of professional players and the elite. Another issue is that they do not have functions to support coaches and managers. Therefore, this study aims to develop a system to support coaches and managers using inexpensive small inertial measurement devices to identify soccer movements in practice environments. We proposed a segmentation method to accurately isolate sequential soccer motions and compared classification models of soccer actions using machine learning. The results of the segmentation method showed that we could segment each movement with high accuracy. In addition, a comparison of classification models showed that SVM model achieved a high F1 score of 84%.
