Detection of Guns and Knives Images Based on YOLO v7

The threat of violence involving guns and knives presents a persistent challenge to global public safety, necessitating effective detection systems despite the small size and potential ambiguity of these weapons. This study endeavors to address this challenge by developing a computerized system tailored for identifying these prevalent weapons, with a focus on enhancing public safety in both military and civilian contexts. Leveraging recent advancements in deep learning, a gun and knife detection system is introduced by utilizing YOLO v7 and YOLO v8 frameworks. Our system provides bounding box annotations for detected weapons, facilitating efficient monitoring and detection. Evaluation results underscore the effectiveness of the YOLOv7-e6 model, exhibiting a notable increase in mean Average Precision (mAP) compared to alternative approaches, thereby enhancing the detection capability of guns and knives. This research contributes to the ongoing efforts to improve security measures and protect communities from the pervasive threat of armed violence.