Difficulty Evaluation of Yoga Poses by Angular Velocity and Body Area Calculation for GPU-Based Yoga Self-Practice System

In this study, an extended difficulty evaluation functions of Yoga self-practice system, which is based on OpenPose, is developed by using human skeleton detection. The Yoga self-practice system implements on the NVIDIA GPU-based embedded platform and utilizes the output results for subsequent Yoga self-practice applications. In addition to providing user's Yoga motion guidance, practice options, and user's feedbacks, the self-practice system also offers a difficulty selection function. In this work, two different difficulty assessment methods are proposed in comparison with previous designs, and two indicators for evaluating the difficulty of Yoga poses are introduced. Firstly, angular velocity calculations are used to perform the dynamic and static regions of each instructor's Yoga motion by assessing the magnitude of pose changes in terms of angles during dynamic and static regions. This facilitates the evaluation of difficulty based on angular and temporal analyses. Secondly, the body area enclosed by four key joints of the upper and lower body is calculated to enhance the credibility of the difficulty analysis through the area comparisons. The research provides a series of difficulty analysis and experimental results for instructor's Yoga poses.