Research on Classroom Management and Student Engagement Analysis System Based on Artificial Intelligence

This study designs and implements an artificial intelligence based classroom management and student engagement analysis system which combines computer vision, deep learning and multimodal sensor technology to analyze students’ facial expression, behavioral action and physiological signal in real time. This study uses improved convolutional neural network method (CNN) based facial expression recognition (accuracy >85%), OpenPose based posture estimation combined with support vector machine algorithm (SVM) based application for classroom behavior analysis (accuracy 90%), and eye- tracking and heart rate sensor information to measure attention level (HRV formula: \(= \sqrt {\frac{1}{{{\rm{N}} - 1}}\mathop \sum \limits_{\mathfrak{i} = 1}^{\rm{N}} {{( {\mathcal{R}{\mathcal{R}}_{\rangle} - \overline {\mathcal{R}\mathcal{R}} } )}}^2}\)) (accuracy 85%). Experimental results show that the system can improve classroom engagement (15% increase in experimental class compared with control class) and has high stability (≥85% in complex environment) based on modular design (such as Apache Kafka for real-time data stream) . This research work offers a practical application of based on artificial intelligence for educational intelligence, which has great application value.