Recognition of Suspicious Action in a Video Surveillance Programme

Abstract

It is more crucial than ever to stop and catch criminal activities due to the increase in both urban and suburban areas. since it is practically difficult for people to continuously monitor these security cameras. A workforce and their constant attention are required to ascertain whether the activities captured are odd or suspicious. The most recent study tries to integrate artificial intelligence, computer vision, and image processing into video surveillance applications. Despite significant improvements in dataset gathering, methodologies, and frameworks, there aren't many articles that give an in-depth analysis of the state of video surveillance systems today. This paper offers a deep learning approach for detecting suspicious behaviour. This method identifies the frame and portion of odd activity, making it easier to determine whether the activity is suspicious or uncommon quickly. Using deep learning models in recognition or classification high movement frames, where we can activate a detection alert when aware of a threat, alerting us to unusual behaviour at a specific point in time, the objective is to identify signs of aggression and violence in actual time, allowing deviations to be distinguished from normal patterns.