Advancements in Weapon Detection Systems: Leveraging Deep Learning for Enhanced Security

Intelligent surveillance systems play a crucial role in enhancing security by enabling the automatic detection of weapons and armed persons in real-time. Artificial intelligence (AI) and convolutional neural networks (CNNs) have gained immense popularity due to their remarkable ability to solve complex problems across various domains. CNNs, in particular, have revolutionized fields like computer vision, enabling machines to accurately recognize images, detect objects, and even generate creative content. As a result, CNNs are now complementary to industries such as healthcare, automotive, and defense. This study focuses on developing a robust detection system by utilizing deep learning-based object detection models. A custom dataset was created by collecting and annotating images from various open sources, covering three key classes: weapons, armed person, and civilian. Three versions of the You Only Look (YOLO) model, YOLOv9, YOLOv10, and YOLOv11 were trained and evaluated using standard performance metrics such as precision, recall, and mean average precision (mAP). Among the evaluated models, YOLOv9s achieved the highest detection performance, with a precision of 0.822, a recall of 0.704, and a mAP of 0.769. The findings underline the effectiveness of YOLO-based models in real-time surveillance applications and contribute to the development of advanced systems for early threat identification and public safety.