Smart CCTV Surveillance Using YOLO and Deep Learning for Crime Prevention

Abstract

The rapid advancement of artificial intelligence and deep learning has significantly transformed surveillance and public safety systems. This project presents an AI-powered Suspicious Activity Detection and Crime Face Recognition System designed to enhance real-time security monitoring. By integrating YOLO-based object detection, pose estimation, and anomaly detection techniques, the system can automatically identify suspicious behaviors such as aggressive movements, loitering, and concealed weapon possession. Simultaneously, the system employs a Grassmannian-based CNN facial recognition module to accurately identify known offenders by comparing extracted facial features against a criminal database. This combination ensures high precision in detecting threats while minimizing false alarms. Designed for deployment in smart city environments, banking institutions, and law enforcement operations, the system provides automated alerts and reporting mechanisms to facilitate immediate responses from authorities. Its real-time processing capability reduces reliance on manual monitoring, enhances situational awareness, and improves crime prevention strategies. By analyzing behavior patterns and matching suspect faces with existing records, the proposed system offers a proactive approach to public safety. Overall, this AI-driven surveillance solution represents a scalable, intelligent, and efficient method to detect and respond to criminal activities, contributing significantly to safer urban environments.