Convolutional neural network-based rating system design for evaluating Taijiquan teaching

Abstract

This paper presents a novel system utilizing convolutional neural networks (CNNs) to evaluate the quality of Taijiquan instruction. This innovative system incorporates the capture of users’ movement images and bone information, employing the dynamic time warping (DTW) algorithm to precisely assess their performance based on established Taijiquan movement evaluation criteria. Furthermore, the integration of deep learning algorithms enables real-time movement recognition, instantaneously providing scoring feedback. In contrast to traditional teaching methods, this system effectively mitigates issues pertaining to movement inaccuracies and unscientific scoring, revolutionizing the approach to Taijiquan instruction. Through the seamless integration of artificial intelligence technology, this system not only enhances the evaluation process of Taijiquan teaching but also significantly boosts the efficiency of course instruction, paving the way for the future advancement and development of the art of Taijiquan.