Generating the IHUGUN individual weapon dataset to improve detection using the YOLO algorithm version ۱۱

Individual weapon detection is one of the major challenges in the field of machine vision and deep learning. Given the importance of this topic in security and surveillance, there is a need for accurate and fast algorithms for individual weapon detection. Individual weapons, as dangerous tools, can be a threat to public safety in many situations. Therefore, the development of intelligent systems for detecting and recognizing these weapons is of particular importance. In this research, our goal is to create a personalized dataset called IHUGUN, which includes ۴۱۵۳ images of individual weapons, which can help improve the accuracy and speed of detecting the presence of handguns. This paper examines the YOLO algorithm version ۱۱ and its impact on the accuracy and efficiency of individual weapon detection. The evaluation results of the YOLO algorithm version ۱۱ show its outstanding performance in this field. The accuracy of this algorithm is ۹۲%, the recall is ۹۸%, and the average precision of ۰.۵ times ۹۴.۴%. Also, the average accuracy in the range of ۰.۵ to ۰.۹۵ is reported to be ۷۵.۳ percent and the F-۱ criterion is ۹۴.۹ percent. Using this algorithm, we try to achieve better results in detecting individual weapons and ultimately help improve public safety. This research can be considered an important step towards the development of individual weapons detection technologies and help create more effective and intelligent security systems. This trained model was run on Tesla T۴ ۱۵GB hardware.

Keywords:

Object recognition , artificial intelligence , computer vision , deep learning , dataset