Research on Sports Dance Training and Teaching in Modern Colleges and Universities Combined with Deep Learning

Abstract Deep learning technology, one of the key technologies in the field of artificial intelligence, has shown a strong potential for application in many fields. The purpose of this paper is to explore the application of deep learning technology in the training and teaching of sports dance in modern colleges and universities, with a view to improving training efficiency and teaching quality. The OpenPose algorithm is used to realize the posture estimation of sports dance trainers, and the sports dance movement recognition model based on TAR-DL is constructed, and then the sports dance movement evaluation method based on the improved DTW algorithm is proposed. The sports dance movement recognition rate of the TAR-DL model is as high as 99.72%, which is significantly better than that of other 3D recognition methods. Meanwhile, the recognition rate of this paper’s method for the six basic sports dance movements is between 95% and 99%, which is better than the recognition effect. Compared with the DTW algorithm, the Improved-DTW algorithm improves the accuracy by 3.14%, while reducing the time consumed by 0.37ms, which proves the effectiveness of the algorithm improvement strategy designed in this paper. In addition, the evaluation results based on the improved DTW proposed in this paper are closer to those of the professional sport dancer teacher, which fully proves the superiority and effectiveness of the Improved-DTW algorithm in the sport dance movement recognition task, and it can be used in the movement evaluation task of sport dance training and teaching in colleges and universities.