A Computer Vision-Based Approach for Assessing Student Behavioral Engagement in the Classroom

Abstract

With increasing emphasis on enhancing classroom engagement and learning outcomes, efficient and accurate assessment of student behavioral engagement has become essential. Traditional methods such as classroom observations and self-reports are often subjective, time-consuming, and lack scalability. To address these limitations, this study proposes a computer vision-based approach for automatically assessing student behavioral engagement in classroom environment. Specifically, an improved VGG16-Attn model is proposed by integrating attention mechanisms to enhance feature extraction and boost behavior recognition performance. Our results indicate that the proposed method not only accurately detects student behavior types but also, by further aggregating and calculating these behaviors, objectively and effectively captures the dynamic evolution of behavioral engagement. This method offers a promising solution for educators to gain actionable insights into student engagement, ultimately contributing to more personalized and effective teaching strategies.