A deep learning-based method for identifying differences in folk dance movements

Abstract

Folk dances are facing the problem of dissemination and teaching difficulties as the number of people watching and learning them decreases. And the models based on computer deep learning algorithms can recognize their movements effectively, which can contribute a lot to the transmission and promotion of folk dances. In this paper, a brief history of the development of human-computer interaction is introduced, and the dynamic body sensing device Azure Kinect is explained in detail. Hidden Markov algorithm is selected for modeling, and this model is combined with Azure Kinect hardware to obtain the dance movement differential recognition system used in this study. Uyghur dance was selected as the object of study and its movements were analyzed in depth. Several typical Uyghur dance movements were set up, and the angles presented by the limbs and trunk of these four movements were measured in detail. Six more movements of dance beginners were selected for analysis. The conclusion showed the difference between the two, and it can be seen that there is a gap between beginners and professional dancers. The significance of this study lies in the fact that after 3 months of systematic and continuous training, and the dance abilities of Uyghur beginners showed significant improvement, particularly in the accuracy of all four movements.