Citation

Chung, Gwo Chin and Abdullah Hezam, Almaswari Osamah and Lee, It Ee and Tiang, Jun Jiat and Teong, Khan Vun (2023) Machine Learning-Based Car Specification Mismatching System for Pre-Crime Detection. International Journal of Intelligent Systems and Applications in Engineering, 11 (8S). pp. 385-392. ISSN 2147-6799

Text document.pdf - Published Version Restricted to Repository staff only Download (475kB)

Abstract

Even with the installation of security systems and video cameras in residential buildings, the number of complexes and crimes in the neighborhood continues to worry residents in the modern era. For instance, the latest statistics show that the rate of vehicle theft was the highest among the crime rates in Malaysia from 2010 to 2017. It is common for criminals to take advantage of security flaws, such as when a phony license plate is put on a car and the security system misses it, allowing the criminals to enter the facility with ease. Hence, this paper intends to close the loopholes that criminals exploit by developing a system to identify car specifications such as the vehicle type, license plate, logo, and color using machine learning. This data will then be used to match the information of the car’s owner, allowing the security system to discover and prevent any crime before it happens. Machine learning and deep learning models such as MobileNet SSD, YOLOv4, OCR and TensorFlow Lite color models are used to predict the car specifications. When mounting security cameras perpendicularly on the front-sides of vehicles to capture high-resolution photos, the proposed system is able to achieve a considerable performance accuracy of 100% for vehicle type, 97% for license plate, 74% for logo, and 68.5% for color predictions, respectively.

Item Type:	Article
Uncontrolled Keywords:	Car specification detection, machine learning, security system, vehicle crime
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions:	Faculty of Engineering (FOE)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	05 Sep 2023 01:42
Last Modified:	05 Sep 2023 01:42
URII:	http://shdl.mmu.edu.my/id/eprint/11680

Downloads

Edit (login required)