Anomaly Detection through Video Surveillance using Machine Learning

Abstract

Anomaly Detection is identification of suspicious human behavior using real-time CCTV video. Human Anomaly Behavior has been studied as one of the main problems of computer vision for more than 15 years. It is important because of the sheer number of applications that can benefit from activity detection. For applications such as image monitoring, object tracking and formed to oversee, sign language identification, advanced human contact, and less motion capture markers, for example, human pose estimates are used. Low-cost depth sensors have disadvantages, such as restricted indoor use, and with low resolution and noisy depth information, it is difficult to estimate human poses from deep images. The proposed system therefore plans to use neural networks to solve these problems. Suspicious identification of human activity through video surveillance is an active research area in the field of image recognition and computer vision. Human activities can be monitored by video surveillance in critical and public places, such as bus stations, train stations, airports, banks, shopping malls, schools and colleges, parking lots, highways, etc. to detect terrorism, robbery, chain snatching crimes, and other suspicious activities. It is very difficult to monitor public places continuously, so it is important to have intelligent video surveillance that can track human behavior in real time and categorize it as common and unusual, and that can generate an alarm. The experimental results show that the proposed algorithm could reliably detect the unusual events in the video.