A Web-based Application for Detection and Recognition of Armed Persons using Convolutional Neural Network

Abstract

This study investigated a web-based application for detection and recognition of armed persons using convolutional neural network. This has become imperative due to the rising security challenges, especially, on the use of illegal firearms. The study developed a model based on Convolutional Neural Networks (CNNs).  A thermal camera images which capture individuals with concealed weapons were used as the CNN input data and to analyze the transformed data. Results reveal an accuracy of 96% from the CNN model. Findings further show that the investigated model demonstrates improved detection rates compared to conventional methods, offering a unique solution for application in concealed weapon detection.  Based on these results, the study recommended that the integration of the model with existing security infrastructure and data systems is essential for its effective deployment in real-world scenarios.