Real-time Human Movement Recognition and Interaction in Virtual Fitness using Image Recognition and Motion Analysis
Chun-Hsiu Yeh, Wei-Cheng Shen, Chi-Wei Ma, Qiu-Tong Yeh, Chung‐Wei Kuo, Jong-Shin Chen · 2023
This research focuses on the application of image recognition, motion detection, and artificial intelligence techniques to achieve real-time identification and analysis of human movements in a dynamic manner. The primary goal is to establish a seamless association between captured human movements and corresponding actions performed by virtual avatars. The implementation involves the utilization of image processing methodologies from python's OpenCV and pygame libraries. MediaPipe technology is employed to capture users' images and facilitate accurate detection of specific movement patterns. By calculating the angles between users' limbs and torso, the correctness of certain movements can be assessed. The research is practically developed and applied to a simulated fitness system, where an interactive exercise game platform is devised. The system is characterized by its engaging user interface, enhancing the ease and enjoyment of fitness activities. The seamless integration of interactive virtual avatars and real users within physical activities serves to stimulate greater participation, particularly among the elderly. The amalgamation of image recognition and motion detection technologies presents opportunities for combining fitness and entertainment in diverse life scenarios, offering an enhanced experience of physical exercise and enjoyment.