Student Engagement Detection Based on Head Pose Estimation and Facial Expressions Using Transfer Learning

The chapter delves into the critical challenge of maintaining student engagement in distance learning, which has become increasingly prevalent due to technological advancements and global events like the COVID-19 pandemic. It highlights the limitations of manual and semi-automatic engagement detection methods, which are time-consuming and rely on learner self-reporting. The proposed approach leverages two datasets: the Head Pose Image Database for tracking head orientation and the DAISEE dataset for analyzing facial expressions and interaction behaviors. The method involves a two-level verification process: first, determining if the student is looking forward using head pose estimation, and second, analyzing facial expressions to assess engagement levels. The chapter discusses the architecture and results of the models used, including a CNN for head pose estimation and pre-trained models like ResNet50 and EfficientNet for facial expression analysis. The findings demonstrate the effectiveness of this integrated approach in providing a comprehensive and real-time evaluation of student engagement, offering valuable insights for enhancing educational experiences.