ARCHIVES

The increasing rate of urban crime necessitates the development of intelligent surveillance systems capable of detecting and responding to criminal activities in real time. Traditional CCTV systems depend heavily on human supervision, which is often limited by fatigue, delayed judgment, and oversight. To address these limitations, this project proposes a real-time crime and robbery detection system that combines deep learning with an intuitive web interface. The system utilizes the YOLOv8 (You Only Look Once) object detection algorithm to identify potentially dangerous or suspicious objects—such as guns, knives, or suspicious bags—in live or recorded video feeds. The system is designed with a Streamlit-powered user interface to facilitate easy deployment and monitoring by non-technical users. It features real-time object detection, alert generation, and configuration settings that allow users to customize the detection threshold and targeted object classes. Upon detecting a criminal object, the system immediately triggers an audio alert using Pygame and sends a frame snapshot via email to predefined recipients using SMTP. These automated alerts significantly enhance response times and minimize the dependency on manual surveillance efforts, especially in sensitive environments like banks, malls, traffic junctions, and public venues. Overall, this project demonstrates the practical application of artificial intelligence and computer vision technologies in public safety and crime prevention. By integrating YOLOv8 with a lightweight, user-friendly interface, the system offers a scalable and modular solution for smart surveillance. The project not only improves the efficiency of monitoring systems but also lays a strong foundation for future enhancements such as behavior analysis, facial recognition, and smart city integration.