A feedback system for adolescents' intangible cultural heritage drama study based on emotion recognition model

In order to solve the problem of untimely and inaccurate feedback on the study of intangible cultural heritage drama and diversified arts for young people, this paper proposes an emotion recognition algorithm based on spatiotemporal feature fusion and multi-level attention mechanism, and constructs a study feedback system. The algorithm integrates the spatial features of facial expressions and the time series features of body movements, and combines the multi-level attention mechanism to accurately capture the emotional changes of young people in the study of intangible cultural heritage drama, museum appreciation, craft production, etc. The experiment recruited 500 young people who participated in the study of diversified arts, collected 12,000 sets of multimodal data to construct a data set, and compared the algorithm of this paper with the traditional CNN, LSTM, and CNN-LSTM algorithms. The results show that the accuracy of the algorithm of this paper is 92.3%, which is 15.6%, 18.2%, and 12.4% higher than that of CNN, LSTM, and CNN-LSTM algorithms respectively; the recall rate is 91.7%, and the F1 value is 92.0%, which are significantly better than the comparison algorithm. The actual application of the system shows that it can effectively help teachers grasp the emotional state of students' study and provide data support for optimizing the study teaching strategy of multi-form integration.