Research on students' gesture behaviour recognition and classroom management system based on computer vision

The wide application of artificial intelligence technology in the field of education has driven the demand for the transformation of traditional classroom teaching to intelligence. The university classroom is large in scale and complex in interaction, and it is difficult for teachers to fully grasp the students' participation. In this paper, we design and implement a computer vision-based student gesture behaviour recognition and classroom management system, which adopts the improved HRNet gesture estimation model and spatio-temporal graph convolutional network to achieve 15 kinds of fine-grained behaviour recognition. The system supports simultaneous processing of up to 150 people with an accuracy rate of 89.6% and a response time controlled within 215 ms, which provides teachers with objective teaching data and effectively improves the quality of classroom teaching and student participation.