NORMA eResearch @NCI Library

Deep Learning-based Weapon Detection Using Live Cameras

Bhardwaj, Abhinav (2022) Deep Learning-based Weapon Detection Using Live Cameras. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
Preview
PDF (Master of Science)
Download (4MB) | Preview
[thumbnail of Configuration manual]
Preview
PDF (Configuration manual)
Download (4MB) | Preview

Abstract

The world is surrounded by cameras, but none of us receive alerts when something goes wrong. In the age of automation, human monitoring is still necessary to keep track of things. Images can be processed in a novel way that focuses the detection algorithm on a specific point in time. This paper explains a new approach for detecting weapons and alerting security forces. The author revisited the topic of identification and created associated confusion classes to reduce the incidence of false positives and false negatives. Deep learning-based classification and recognition algorithms were used to test the new self-generated database. Database was built by using computer camera to capture weapons like ’knife’ and other weaponlike objects ’vapes’, ’keys’, and ’pen’. This study uses a variety of algorithms, all based on deep learning and CNN architecture. Transfer learning from TensorFlow 2 Detection Model Zoo was used in the COCO dataset models that have been pre trained to reduce training time.

With this study’s groundbreaking work, it is now feasible to identify weaponry in real-time video footage. To discover the best CNN object detector for real-time weapon detection in CCTV video streams. The author has built a new dataset of 411 photos using the OpenCV package captured in various environments such as: bright light, low light, blurred, sharp, reflective surface, dark and light background. In terms of speed and accuracy, On SSD MobileNet V2 FPNLite 640X640, the trained model predicted images in nearly all orientations, aspects, and viewpoints with an accuracy of 71.8 percent.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
H Social Sciences > HV Social pathology. Social and public welfare > Criminology
Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Tamara Malone
Date Deposited: 18 Jan 2023 17:51
Last Modified: 06 Mar 2023 16:29
URI: https://norma.ncirl.ie/id/eprint/6089

Actions (login required)

View Item View Item