Teacher Assistance System to Detect Distracted Students in Online Classroom Environment

Perceiving students' engagement accurately is important for teachers and educational institutions, especially in online classrooms. In this paper, we address the problem of automated detection of distracted students from videos recorded in online classroom environment. To tackle this difficult challenge, we propose a Teacher Assistance System that detects inattentive students using their head pose from video data. We gathered data in two sessions by using Google Meet and WebEx portals to conduct online classes for a group of nineteen undergraduate Computer Science students. The authors used the OpenCV Caffe model to detect the faces. We used the Perspective-n-Problem approach to estimate the head pose of all the pupils in the class, and we also used the distance between the nose tip and the chin point to improve the model's accuracy. In an online classroom environment, the proposed system detected distracted pupils with 90.45% accuracy. The proposed Teacher Assistance System is precise and does not demand the use of any advanced tools. As a result, it might be used in both online and offline classroom scenarios, as well as other places where people who are distracted need to be identified.