Published Paper Details:

Unusual human activity detection in crowded scenes. Suspicious behavior is hazardous in public settings and can result in significant casualties. Systems that detect motion or pedestrians have been developed using video frame collection, however even in real time, those systems lack the intelligence to spot unusual activity. To quickly and effectively manage a scamper issue before any casualties occur, it is necessary to spot scamper situations in real time via video monitoring. The proposed system focuses on identifying suspicious activity and aims to provide a technique that makes use of computer vision to detect suspicious behavior automatically. The system in this instance makes use of the OpenCV library to instantly classify various activities. The examination of the motion regularly shifts the location between two locations has been represented by a motion influence map. Utilizing CNN and YOLO algorithm for real-time object detection.