Intelligent Surveillance System Using Deep Learning

Abstract

Criminal activities have become increasingly common in today’s world, creating a critical need for intelligent and proactive surveillance systems. This proposed system presents an advanced security framework that predicts abnormal activities and detects weapons using the YOLO v12 algorithm, while ResNet-101 is employed for facial verification to accurately identify individuals. YOLO v12 enables real-time detection of suspicious behaviors and dangerous objects such as guns and knives with high accuracy and low computational latency, making it suitable for continuous surveillance in complex environments. Facial verification using ResNet-101 enhances the system’s capability by matching detected faces against authorized or watchlist databases, supporting reliable person identification and threat attribution. The integration of behavioral analysis, weapon detection, and identity verification provides a comprehensive understanding of potential security threats. Experimental evaluation demonstrates improved accuracy, efficiency, and adaptability compared to traditional surveillance methods. Overall, the proposed system offers a scalable and effective solution for intelligent monitoring, proactive crime prevention, and enhanced public safety in real world surveillance applications.