Real-Time Weapons Detection System using Computer Vision

The growing use of Closed-Circuit Television (CCTV) systems in modern security applications has driven the need for automated surveillance through computer vision. The primary aim is to reduce human intervention while enhancing early threat detection and real-time security assessments. Although advanced surveillance technologies have facilitated monitoring, constant human oversight remains challenging. This has prompted a quest for models capable of identifying unlawful activities with minimal human involvement. Real-time weapon detection, despite advancements in deep learning algorithms and dedicated CCTV cameras, remains a formidable challenge, especially with varying angles and potential obstructions. Existing detection systems are often expensive and require specialized tools, necessitating a cost-effective and reliable alternative that minimizes false positives. This research focuses on creating a secure environment by utilizing real-time resources and deep-learning algorithms for identifying dangerous weapons. Without a predefined dataset for real-time detection, the researchers compiled one from diverse sources, including camera shots, internet images, movie data, YouTube CCTV recordings, and Roboflow Computer Vision Datasets. The proposed weapon detection system employs a hybrid model of Detectron2 and YOLOv7, emphasizing precision and recall in object detection, particularly in challenging conditions like lowlight environments. This research contributes to developing an effective, reliable real-time weapon detection system tailored for diverse scenarios.