An Intelligent Framework for Crime Prediction Using Behavioural Tracking and Motion Analysis

Abstract

Closed Circuit Television Systems are now being deployed in most public spaces to make the city more secure. Manual observation of these clips for the prevention of crime would take up a lot of manpower. This paper proposes an Intelligent Framework using the power of Artificial Intelligence to ensure the safety of the surroundings. The system will use different Computer Vision techniques for video analysis. It will monitor CCTV footage for any criminal offenders, violent objects, and suspicious behavior which could lead to crime. SSD Mobilenet Model, an architecture for concealed object detection, is trained for labeling weapons in the frame. The images captured are processed using Face Detection algorithms to identify human faces. Facial Recognition API using libraries in python is implemented to recognize the offenders from criminal records. A ResNet-GRU Model was trained for human behavior analysis which detects suspicious actions. An alert is generated when there are signs of crime and concerned authorities are notified. The proposed framework aims to make societies secure by correctly identifying criminals and crime-related objects.