Effective Deep Learning Technique for Weapon Detection in CCTV Footage

Abstract

The modern world is highly concerned with safety and security. The level of safety and security in a country’s environment determines its capacity to draw international investment and tourism. Closed Circuit Television (CCTV) cameras are used for monitoring and to keep an eye on activities like robberies, but they still require human monitoring & engagement. A system that can instantly identify these criminal acts is required. Despite of cutting-edge deep learning algorithms, quick processing power, & sophisticated CCTV cameras, the real-time weapon detection remains a significant barrier. In this work, YOLOV5 algorithm is used to detect the weapons in Closed Circuit Television (CCTV). The proposed system exhibits 96% F1 score and 96.6% mean average precision. This is a significant improvement when comparing with existing system.