DVS-TBA: Research on Classroom Teaching Behavior Analysis Method Based on Dual Video Stream

This paper proposes a dual video stream-based teaching behavior analysis model (DVS-TBA) to enhance teaching quality and student engagement. By deploying dual cameras in classrooms, the system non-invasively captures comprehensive teaching activities, avoiding information loss and reducing manpower costs. Utilizing the M-CPIE model for feature extraction and Bayesian causal discrimination, the system analyzes teacher and student behaviors accurately. Experimental results demonstrate that the DVS-TBA model improves teaching mode recognition accuracy and enhances overall educational effectiveness.