Suspicious Motion Detection in Examination Hall Using Head Pose Estimation
In this paper, an attempt to reckon the suspicious activities in an examination hall has been made. It is primarily based on the careful calculation of head rotation (i.e. towards the immediate back and sideways of the student). This is tried to be achieved using the method of head pose estimation' in order to get the idea about the head rotation angle and hence predicting particular sort of suspicious activity using the concept of SVM (Support Vector Machine). The data source in case of normal activity is the live examination footage from academic institutes. Whereas in the case of suspicious activity, data has been taken from the volunteer candidates. This will mostly act as an auxiliary concept along with other means for detecting such as object detection (cheat slips, mobile phones), gaze tracking etc. In fact this proposition can be effectively used in the domains wherever surveillance is done and can help in reducing the human error involved during vigilant activities.
Keywords - Suspicious Motion Estimation, Pose Estimation, Head Motion Detection, Activity Recognition, SVM.