Design and Implementation of a Motion Training Assistance System Based on Body Sensing Recognition Technology
Cheng Chen · 2023
Nowadays, with the rapid development of modern information technology, traditional sports equipment is being challenged by emerging technological products, such as smart cloud gyms and intelligent sports wearables. These tech-savvy products offer users a more intelligent and convenient exercise experience. However, for non-professional sports enthusiasts, many of the current auxiliary functions on the market, especially those related to timing and counting for activities like running and swimming, are significantly limited compared to the high-end equipment provided by professional gyms. Notably, these functions lack targeted guidance for strength-based exercises. This paper introduces and designs an assistance system for sports training based on body sensing recognition technology. In this system, we opted to use the Kinectv2 intelligent body sensing device to capture human skeletal information, ensuring the data’s accuracy and real-time responsiveness. By integrating double exponential smoothing filtering and jitter removal techniques, we delved deep into the analysis of the collected skeletal data. Additionally, using the spatial vector method, we extracted essential joint skeletal features and employed supervised learning methods to optimize and train the model parameters. As a result, the system can accurately and real-time recognize and guide various exercise training movements. After a series of data analyses and tests, the findings show that the system demonstrates excellent practicality and stability across multiple real-world application scenarios, providing nonprofessional fitness enthusiasts with an efficient and convenient exercise assistance tool.