Weapon Detection Using Yolo

Abstract-In light of the increasing gun violence incidents worldwide, there is a pressing need for automated visual surveillance systems capable of detecting handguns. This paper presents a method for real-time handgun detection in video streams using the YOLO algorithm, comparing its performance in terms of false positives and false negatives against the Faster CNN algorithm. To enhance detection accuracy, we compiled a custom dataset featuring handguns from various angles and merged it with the Roboflow dataset. The YOLO model was trained on this combined dataset and validated using four different videos. The results indicate that YOLO effectively detects handguns across diverse scenes, demonstrating superior speed and comparable accuracy to Faster CNN, making it suitable for real-time applications.