Intelligent evaluation and feedback mechanism based on student behavior data in the context of integrated education

Artificial intelligence has been promoted in education, but traditional teaching evaluation and feedback mechanisms still cannot start from students' classroom behavior, resulting in poor accuracy and low efficiency. In integrated education, educational goals not only focus on imparting knowledge, but also emphasize fairness and respect for individual differences among students. To improve the accuracy and fairness of teaching evaluation, this paper designs an intelligent evaluation system that integrates student behavior data. On the basis of You Only Look Once version 5 (YOLOv5) and the Deep Sort algorithm, this system has built an intelligent model that accurately tracks student behavior and evaluates teaching effectiveness by analyzing key behavioral indicators such as students' overall classroom head up rate and concentration. The results showed that the intelligent model could more effectively recognize students' behavior, with target recognition accuracy and tracking accuracy reaching a target detection mAP@0.5 of 85.13% and a tracking precision of 90.78%. In addition, the intelligent evaluation system quantified teaching effectiveness, with a mean comprehensive classroom head up rate of 71.85%, an increase of 8.69% compared to traditional quantitative methods. Intelligent evaluation is more sensitive than traditional methods. This system can accurately measure an individual's participation in learning and provide personalized learning feedback for teachers. The application of this technology in integrated education environments offers strong support for achieving more detailed evaluation and feedback and has important practical value for promoting educational equity and implementing personalized teaching.