Dance Movement Recognition Based on Gesture
Ping Lei, Nana LI, Haidong Liu · Advances in computer science research · 2023
Aiming at the low accuracy of traditional dance movement recognition methods, a movement recognition algorithm based on human posture estimation is proposed.Firstly, PAFs algorithm is adopted to recognize the spatial skeleton nodes of the human body model and the connection of human body joints, thus the human movement skeleton is obtained.According to the movement skeleton, the human body posture can be estimated.After the posture information is preprocessed and features are extracted, LSTM time series algorithm is used to classify and recognize the dance movements.The results show that the algorithm can clearly identify the dance movement skeleton nodes.For different movement categories, the recognition accuracy and recall rate of different movement categories are above 85%, and the recognition accuracy of curtsey movement is up to 95.2%.It can be seen that the recognition accuracy of this algorithm is significantly improved and different dance movement categories can be accurately recognized.