International Research Journal of Education and Technology

Analysis of Human Behaviour has attracted many research attention in Computer Vision. Various applications of Computer Vision such as, education, health care, human-computer interactions and video understanding. Though many research work in this domain, the problems and challenges have remained unsolved. Some of the challenges are Occlusion, Eye Movement Metrics, Interclass variation, etc., In the part of large group, our project aims to extract behavioural information from the input videos or live streams which contains input as Crowd Environment. This method is applied in Classroom Environment to enhance the teaching quality by analysing the student activity. Many studies have focused on the physical activity of a student such as hand raising gestures and sleeping activity by Pose Estimation and Person Detection. So, we are proposing Oculus Behaviour to improve and enhance the accuracy of the existing system. Therefore, this project proposes a Behavioural Analytical Model for Crowd Attentiveness based on Skeleton Pose Estimation, Person Detection and Oculus Behaviour.

Keywords- skeleton pose estimation; crowd behaviour; person detection; eye tracking; deep learning; CNN.