Advanced Weapon Detection and Classification Using Fine-Tuned Transfer Learning Models

D'souza, Brandon Craig (2024) Advanced Weapon Detection and Classification Using Fine-Tuned Transfer Learning Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (2MB) | Preview

Abstract

Safety and security are big problems in today’s modern world. With the increasing number of criminal activities, automatic control systems have become very important for security purposes. One of the most serious activity is use of illegal weapons. Many systems today have a number of disadvantages as they tend to employ a lot of manpower in the monitoring and controlling of processes and also in managing of knowledge in this field. This project will present a weapon detection system by using Deep Learning (DL) models and a Flask-based web application. The dataset has been sourced from images.cv which includes labeled weapon images categorized into four classes: knife, pistol, rifle, and sword. This study uses different models which includes Convolutional Neural Networks (CNN), Xception, EfficientNet-B2 and EfficientNet-B2 with Attention Baseline. The performance of these models have been evaluated based on accuracy and F1-score. Among them EfficientNet-B2 with Attention Baseline has been achieved the highest accuracy of 95% and a macro average F1-score of 0.95 which has an increased performance and accuracy. A web app has been developed using Flask for a user interface, which allows users to upload weapon images and obtain real-time predictions. The application interface gives users an easy image upload and displays the classification results which includes the weapon type detected. This investigation will thus help the law enforcement and security personnel to easily identify and classify weapons quickly so as to increase safety of people and decrease the extent to which human intervention is required

Item Type:	Thesis (Masters)
Supervisors:	Name Email Qayum, Abdul UNSPECIFIED
Uncontrolled Keywords:	Weapon Detection; Deep Learning; CNN; Xception; EfficientNet-B2
Subjects:	H Social Sciences > HV Social pathology. Social and public welfare Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	15 Aug 2025 18:06
Last Modified:	15 Aug 2025 18:06
URI:	https://norma.ncirl.ie/id/eprint/8556

Actions (login required)

View Item