Teacher Classroom Behavior Detection Based on a Human Pose Estimation Algorithm

Teachers’ classroom behavior testing can help the understand the learning situation of students, feedback the teaching effect, and formulate corresponding measures to help. Traditional evaluation methods rely on manual testing, which is time-consuming and lacks objectivity. This article suggests a new approach to analyze the behavior of teachers by using the human skeleton posture. This method first uses the popular human posture gesture estimation technology to obtain the skeleton joint node of the teacher in the teaching process, and then remove several of the joint nodes that have a weaker effect on the classification of teachers in class. Finally, a convolutional neural network was designed to classify the teacher’s behavior based on the skeleton joint nodes and confidence obtained from processing the input. The experimental results show that the effectiveness of the method helps the dynamic management and evaluation of classroom teachers’ behavior.