Intelligent classroom teaching evaluation system based on artificial intelligence

Conventional classroom teaching evaluation methods, including post-class conversations and questionnaire surveys, exhibit significant delays and are unable to assist teachers in making adjustments on time. Consequently, the timely and effective provision of feedback to teachers regarding students' emotional states in the classroom has emerged as an urgent problem that must be addressed in current classroom teaching practices. This paper proposed a classroom teaching evaluation system based on YOLOv11, which focused on analyzing students' facial emotions to help teachers adjust teaching methods in a timely manner and improve the quality of classroom teaching. The system's functions included facial image acquisition, facial detection and localization, facial emotion feature extraction and recognition, classroom activity level statistics, and generation of teaching status reports. Firstly, the student emotion dataset was constructed for student classroom emotion recognition by collecting facial images of students in the classroom. Secondly, the dataset was labeled as five types of emotion. Then, the emotion recognition model was trained using the YOLOv11 algorithm and tested for its recognition accuracy. Finally, the students' classroom activity level was calculated, and the report on their classroom teaching status was generated. The results of the classroom teaching evaluation demonstrated that the student emotion recognition accuracy of the proposed system reached 90.9%. This system has been demonstrated to meet the requirements of teaching evaluation tasks based on student emotion recognition.