Design and Implementation of Dance Personalized Teaching System Assisted by Artificial Intelligence

Current online dance learning faces problems such as the inability to capture accurate information about dance movements and the difficulty of gaining insights into the key flaws of skillful movements. In this paper, a personalized dance teaching system based on motion recognition is constructed. Firstly, RGB and infrared cameras are used to carry out 25 joint point coordinates calibration to establish the human skeletal structure, and combined with depth information to realize the representation of the human skeleton structure in three-dimensional space. The image sequence is converted into a batch of images by a sequence folding layer, and independent CNN convolutional computation is performed in the time dimension, and then the images are converted into feature vector output by a spreading layer. The backbone network is constructed by stacking three LSTM layers for dance movement recognition. In the experiment, we analyze the recognition effect of the system on six types of dance movements, the performance is verified and then applied to the dance teaching practice and invite 50 dance practitioners to subjectively evaluate the system in this paper, and the scores obtained are higher than 8, and the effectiveness of the system is affirmed. The effectiveness of the system is confirmed. This paper provides a useful exploration for the intelligentization of dance teaching.