Arabic Vehicle Licence Plate Recognition Using Deep Learning Methods: Review

Abstract

Automatic vehicle identification via its license plate is proven to be a valuable solution for smart transportation and smart city applications. The most recent studies explore the implementation of deep learning techniques to improve the license plate recognition performance concerning the challenges and difficulties associated with license plates, such as languages, fonts, distortions, hazardous situations, and blurriness and illumination diversions. In many Middle East countries, vehicle plates include letters, numbers, and city names written in Arabic. Many deep learning approaches have been conducted to improve identification accuracy, with many performance issues. This study reviews the current deep learning methods used in the automatic identification system of such license plates, focusing on the process of deduction, segmentation, and recognition. Methods were analyzed and compared based on applied attributes, strengths, weaknesses, and recognition performance. The paper aims to highlight the research gaps in this area and give some insights into filling them by providing all the related information and proposing new ideas to develop the research further.