Unusual Human Activity Detection

Suspicious behavior is dangerous in public and can cause serious causality. There are various systems developed based on the capture of video images with motion and pedestrian detection, but these systems are not intelligent enough to detect anomalous activity even in real time. Escape from video surveillance needs to be detected in real time in order to manage casualties quickly and quickly before they occur. The proposed system focuses on the detection of suspicious activity and aims to realize a technique that can automatically detect suspicious activity using computer vision. Here, the system uses the OpenCV library to classify different types of actions in real time. Motion influence maps were used to represent motion analysis that frequently repositions from one location to another. The system uses pixel-level representations to facilitate understanding or identification of the actual situation