A Hybrid YOLO-Based Approach for Fine-Grained Detection of Classroom Student Behaviors

This paper introduces a system for real-time identification of student behavior in classrooms, employing YOLOv8 for precise object detection and Convolutional Neural Networks (CNNs) for behavior analysis, this technology enables the identification of body language to categorize behaviors such as attentiveness and participation. The scalable solution is guar-anteed to work effectively for various classroom sizes due to the implemented approach. The system possesses the potential to revolutionize classroom management, as evidenced by its initial findings demonstrating its accuracy. Furthermore, this technology facilitates the provision of immediate feedback to teachers, while simultaneously enabling them to identify long-term behavioral patterns. It serves as a valuable tool that assists teachers in delivering tailored instruction and fostering enhanced student engagement. The source code and models are publicly released: GitHub Repository.