Behavior Analysis-Assisted Classroom Teaching Based on Multi-Representation Computer Vision Under the Industrial Internet of Things Framework

In traditional classroom settings, teachers predominantly rely on visual observation and verbal questioning to assess student status, limiting the ability to deliver timely and precise feedback. To address this limitation, this study introduces a multi-modal computer vision-based behavior analysis approach within an Industrial Internet of Things (IIoT) framework. The proposed system utilizes multiple cameras to capture behavioral indicators—such as speech, facial expressions, and body posture—and integrates deep learning models (e.g., YOLO, SSD) for real-time recognition of students' learning states. By leveraging IIoT's data transmission and edge computing capabilities, the system significantly enhances the accuracy and responsiveness of classroom behavior monitoring. Experimental results indicate that the method effectively detects student attention, engagement, and emotional states, thereby supporting dynamic instructional adjustments. This research contributes to advancing smart education initiatives aligned with Industry 5.0 paradigms.