Compact weapons, such as knives, handguns, and pistols, are becoming more common in a variety of criminal activities, including bank robberies, retail theft, and break-ins at homes. Another security hazard that should be considered is the possibility of fires resulting from road accidents, short circuits, or household mishaps. This emphasizes the need for strong security systems to quickly detect and stop such events. Closed-circuit television (CCTV) cameras are a mainstay of modern security systems, serving as vital tools for preventing crime, identifying incidents, and resolving them. However, human frailty and exhaustion impede manual surveillance, highlighting the need for automated systems that can quickly identify small arms or fires and alert authorities. Making the most of the latest advancements in deep and machine learning, especially in computer vision, has become a viable path. Enhancing a detection system's performance is the main goal of this study. The experimental results showed a mean average precision of 95.7%, a precision of 98%, and a recall of 89.4%.