Research on classroom attention algorithm for student facial expression recognition based on CF-YOLO

In classroom teaching, it is difficult for teachers to accurately grasp the actual absorption degree of each student's knowledge points, which indirectly affects the improvement of teaching quality. Although it is possible to monitor students' learning progress and psychological status by manual review of classroom videos, this method can no longer meet the changing requirements of classroom teaching. Therefore, this paper uses object detection algorithm to recognize students' classroom expressions, chooses FasterNet to replace the backbone network, improves non-maximum suppression Soft NMS, and integrates CBAM attention mechanism to cope with students' dense environment. Double tests confirmed the improvement effect, and the CF-YOLO model performed well in classroom expression recognition. Finally, this study also establishes a comprehensive evaluation model based on students' behavior and expression concentration, which provides a new idea for school management and solving problems existing in educational practice.