VIDEO ANOMALY DETECTION

Abstract

With the development of the camera technology the need of intelligent surveillance technology is required for the safety and security concern. In order to save human life and national infrastructure the anomalous pattern or activity is separated from normal one. This provides an ample time to take the necessary step in regarding to handle the critical situation. The video anomaly detection has become a very important research area in last decade. The anomaly detection find its application in area like abnormal behaviour or activity in public places, object tracking, face recognition, traffic monitoring and so on. The anomaly detection in various area like traffic monitoring, tracking, face recognition and abandoned object detection provides various aspect of challenges to find out the odd one from then normal one. We have to take the various aspects into consideration like light intensity variation, indoor observation, outdoor observation, crowded environment and object occlusion. A number of different frame model like unsupervised, supervised and semi supervised have been proposed by researcher in past years to detect the anomaly. Our aim is to build a semi-supervised frame model which provides a hand in hand coordination with human interaction to detect the anomaly using DPG, image segmentation, tracking and predicting path and abandoned object detection.