Abstract: Suspicious behaviour in open areas is sensitive and prone to serious consequences. There are many systems built based on video frame acquisition that monitors motion or identifies pedestrian but those systems are not smart enough to find out the unusual activities even in a real-life scenario. It is necessary to identify fugitive scenarios in real-time from video surveillance for rapid and immediate control before any casualties. To create a technology that can automatically identify suspicious behaviour using computer vision, the system focuses on distinguishing suspicious scenarios and pinpointing the precise target of the activity. The system utilizes the Open CV library for distinguishing and categorizing several kinds of tasks or actions in real-life scenario. Automated event identification, movement-based recognition, person count tracking, autonomous robot navigation, and a variety of other disciplines are some of the other topics covered. Separating objects from their background, on the other hand, is a challenging process.A difficulty arises when an object appears from the background as a result understanding the footage and its elements with portrayed scenarios becomes the most essential requirement. The utilization of a typical human behavior approach is the predictable goal in the unpredicted activity detection phase.

| DOI: 10.17148/IARJSET.2023.10565