AN AI-POWERED SURVEILLANCE SYSTEM FOR WEAPON AND VIOLENCE DETECTION IN REAL-TIME CCTV STREAMS

Public safety in crowded environments such as transportation hubs, public gatherings, and critical infrastructures is a significant concern. Traditional surveillance systems rely heavily on manual monitoring, which is prone to human fatigue, delayed response, and missed incidents. This paper proposes an AI-powered surveillance framework that utilizes deep learning techniques for real-time detection of weapons and violent activities in CCTV video streams. The system integrates YOLO (You Only Look Once) for object detection and CNN-LSTM models for action recognition. The proposed framework enables automated monitoring, instant threat detection, and real-time alert generation. Experimental results demonstrate improved efficiency, faster response time, and enhanced security compared to conventional surveillance systems.

Artificial Intelligence, Surveillance System, YOLO, CNN-LSTM, Weapon Detection, Violence Detection, Computer Vision