Gun Detection in Images Using Convolution Neural Networks

The prevalence of gun-related issues has prompted increased research in gun detection systems. In this work, we propose a gun detection system for images using convolution neural networks. Our system creates models to analyze large data sets of images that either contain or do not contain a gun. Our system is able to detect all types of guns, as well as multiple guns contained in a single image. Our trained models are implemented in a web application where users are able to select an image and evaluate its contents by using our models. The system returns the message gun or not gun depending on if our model perceives a gun in the image. Preliminary results from multiple experiments are promising with our models being able to detect guns in images with high accuracy values ranging between \(88\%\) and \(95\%\) , and with low loss values.