With the continuous deepening and progress of the research in the physical education discipline, the school and the state put the landing point on the integration of information technology and the development of the discipline. In this paper, we design the system framework and perform data module design, on the basis of which we conduct detection research on teacher-student interaction behavior. The article employs an intelligent analysis system based on the YOLOv5 model to establish a model of teacher-student interaction behavior. This model is based on the movement and facial observations of both teachers and students in a physical education classroom. The intelligent analysis system, based on the YOLOv5 model and independent t-test, observed the behavior of teacher-student interaction based on the ratio of teacher-student verbal interaction and analyzed it accordingly. In teacher- and student-type speech, the ratio of teacher-direct influence speech is 45.62%, which is higher than indirect influence speech, while the ratio of student-passive speech is 5.54%, which is higher than student-active speech. Finally, the interactive intelligent system was evaluated for teaching and learning to understand students’ feedback on the intelligent analyzing system based on the YOLOv5 model from two parts: physical education test scores and students’ satisfaction surveys. In the physical education classroom, the teacher primarily uses their own words and interacts with the students less frequently. However, the system has the potential to significantly enhance students’ performance in Physical Education, with the majority of students reporting improvements in their PE skills and other areas. In terms of participation in the physical education classroom, 32 students thought that there was a “great improvement,” and 22 students thought that there was an “improvement,” for a total percentage of 98.18%, which is a significant effect.
