A system for monitoring the engagement of remote online students using eye gaze estimation

摘要

We present a system for monitoring the engagement of remote students in an online educational environment. We introduce a novel gaze estimation system that uses a client-side video-based 3D face tracking algorithm, and expands it to estimate the head pose, and intersects this pose-based gaze line of sight with the screen (coplanar with the camera) to estimate the screen position currently viewed. Our preliminary experiments indicate this approach can accurately determine the coarse location of a student's gaze and use it as a cue for engagement estimation. Our system operates in real-time and achieves comparable performance to state-of-the-art gaze estimation techniques that account for free head motion.