Design and Implementation of System of Recognition of Students’ Learning Behavior in Classroom Teaching Videos

Student learning behavior not only shows student participation in the teaching process, but also affects the quality of teaching directly. However, manual observation and coding of student behaviors exist with high workload and are susceptible to subjective factors of analysts. Combining artificial intelligence and other technologies to achieve recognition and analysis of classroom student behavior is also a prevailing research trend. The current student behavior recognition system has problems such as failing to pay attention to the interference of teacher activates for student behavior recognition, fewer behavioral categories of students that can be recognized through visual information alone. Therefore, this paper develops a system of recognition of student learning behaviors in classroom teaching videos by a method on recognition of student learning behaviors based on audio-visual information to identify student learning behaviors in teaching videos. First, this paper designs a pre-annotation module in which the user marks the teacher's image, voice, and silent clips for providing accurate identification of subsequent student behaviors. Second, the student behavior recognition module provides for the recognition of eight categories of student behaviors in real classroom teaching videos and visualizes the results so that users can understand the percentage of student learning behaviors in the current teaching clip and the temporal changes of individual behavior categories. The purpose of system is to assist teachers to be able to count student learning behaviors in the classroom quickly and scientifically and grasp student learning dynamics for data-enabled classroom teaching analysis.