Design and Implementation of Image-Sensing Based Learning Assistant System for Self-Practice Dynamic Yoga on Embedded GPU Device
Shih-Wei Shen, Wan-Chia Huang, Irin Tri Anggraini, Nobuo Funabiki, Chih‐Peng Fan · 2023 IEEE International Conference on Consumer Electronics (ICCE) · 2023
In this study, by using OpenPose technology and fuzzy rules, the developed system provides an image-sensing learning system by combining the back-end with front-end functions about self-practice Yoga sports at household. In the front-end side, users can select the Yoga exercises which they want on the web page at the first level, after selecting the exercise, the website page will go into the web page at the second level, users select the options that the system provides the start, skeleton, and arrows functions. In the back-end side, firstly the the frame for initialization, the key frames, and the frames for synchronization of instructor are selected based on Euclidean distances in pre-processing. Secondly, in system processing, the frames obtained in pre-processing are utilized to compute the difference of angles between the instructor and user, then taking the angle difference value as input into the fuzzy-rule based scoring system to get each score, Finally, the system synchronizes the instructor and user at the time domain. By NVIDIA Jetson Nano device, the experimental results show that user can interact with the system through the user interface (UI), and the back-end operations work effectively by the detection of dynamic Yoga poses.