Intelligent Pose Recognition and Evaluation System for Rowing Sports
Yunlong Zhang, Yuzhi Zhou, Peirong Wang, Le He, Zhengyu Zhang, Jiuchun Ren · 2024
This paper introduces a vision-based performance evaluation system tailored for rowers. Initially, we utilize motion capture devices and video to establish a comprehensive dataset of rowers' real training actions, which is then used to train our model. Subsequently, within the algorithm, we detect key point information regarding the athlete's body during movement, proposing a set of evaluation metrics for these key points. These metrics scientifically and accurately reflect the athlete's performance, enabling a scientific assessment of the rower's technique. Experimental results validate the precision and efficacy of the system, offering more intuitive guidance and support for the technical enhancement and training regimen of rowers.