Firearm Detection in Images of Video Surveillance Cameras with Convolutional Neural Networks

Abstract

The purpose of the research is to develop a study of models of Convolutional Neural Networks using YOLOv3 and YOLOv5s (Only look once) for the detection of firearms trained with images of weapons obtained from the research database (Soft Computing and Intelligent Information Systems A University of Granada Research Group) in order to test the effectiveness of the algorithm and its training in real images of video cameras in an accessible database, to demonstrate that although the images are of low quality, the chances of identifying the firearm are high.