Deep Learning Based a Novel Method of Classroom Behavior Recognition

Abstract

With the rapid rise of artificial intelligence, the application of artificial intelligence in classroom teaching has become more and more extensive. Combining the related technologies of artificial intelligence with the dynamic detection of classroom behavior is a hot research topic at present. In this article, a novel recognition method of students' action in classroom is proposed. The method first uses YOLOv3 to recognize a single target in the classroom, and then, according to the proposed combination method, combines each single target into a fixed student classroom behavior. The research was carried out in eight classes (215 students in total). The research results show that the proposed research method can accurately and quickly identify students' behaviors in classroom.