Intelligent Video Surveillance Using Deep Learning

Abstract

The project Intelligent Video Surveillance Using Deep Learning is designed to build a smart real-time system that can detect serious threats like guns, fire, and suspicious activities such as firearm refilling. It combines YOLO for quick object detection with RCNN for precise classification and localization. YOLO provides speed, while RCNN ensures higher accuracy by focusing on specific regions. Once a threat is identified, the system captures the video frame and instantly sends an email alert with the frame and a warning message. This rapid alerting mechanism allows immediate action in high-risk places like schools, banks, and public areas. The model is trained on diverse datasets, enabling it to perform reliably under different lighting, camera angles, and environments. It minimizes false alarms while maintaining strong accuracy. The system is also efficient, supporting continuous monitoring without high computational cost. By sending real-