Classroom Concentration Analysis System Based on YOLOV5

This study presents a classroom attention analysis system based on YOLOV5, leveraging deep learning techniques to assess students' attention levels. The system utilizes facial recognition technology to capture facial fatigue data in the frontal view of students, incorporating various fatigue indicators such as eye open/closed and mouth open/closed states. Additionally, classroom activities, such as reading and writing, are categorized and analyzed. The facial fatigue scores and activity scores are combined using a weighted fusion approach to derive a comprehensive attention score. The scores are then normalized and classified into three levels: low, medium, and high. Experimental results demonstrate an accuracy of 97.5% for facial state recognition and 75.3% for classroom activity recognition, confirming the high accuracy and effectiveness of the proposed system. This method provides a novel solution for attention analysis in educational settings, with promising potential for practical application.