Improvement of a Model for Detecting Student Classroom Behaviour Based on Yolov8 via CBAM

In the evolving digital education landscape, understanding and analyzing student behaviour in the classroom is paramount for enhancing teaching and learning experiences. This study proposes an innovative approach to detect student classroom behaviour using the YOLOv8 model enhanced by the Convolutional Block Attention Module (CBAM). We hypothesize that integrating CBAM with YOLOv8 improves the model's accuracy in identifying specific student behaviours-such as hand raising, reading, and writing-in complex classroom settings. Utilizing the SCB-dataset3, which encompasses diverse classroom behaviours, our comparative experiments demonstrate a substantial improvement in detection accuracy. The YOLOv8 model, augmented with CBAM, notably increased precision, recall, and mean Average Precision (mAP), confirming our hypothesis. This research showcases the potential of applying deep learning techniques in educational settings and offers a foundation for future advancements in student behaviour analysis. Our findings suggest that attention mechanisms like CBAM can significantly enhance object detection models' performance, providing new avenues for educational technology development.