Automatic Recognition of Students' Classroom Behavior Based on Computer Vision

The performance of students' classroom behavior is an important part of the evaluation of classroom teaching, and the recognition of students' classroom behavior is of great significance to the evaluation of classroom teaching. However, due to the complexity of students' classroom behavior, it is difficult to identify intelligent students' classroom behavior. Therefore, this paper proposes a classroom behavior analysis and evaluation system based on deep learning face recognition technology. The classroom behavior analysis and evaluation system judges whether students pay attention to class from three aspects: Students' side face concentration, students' head up and head down concentration, and eyes opening and closing concentration, so as to provide an objective evaluation basis for students' classroom behavior evaluation in classroom teaching.At present, to solve the problem of performance degradation of convolutional neural network with the deepening of network layers, a deep residual network based on residual structure is proposed. Comparing the accuracy of the depth residual network with that of the depth convolution neural network on this data set, the experimental results show that the former has better network recognition performance.