Automatic Number Plate Recognition Powered by Deep Learning

This paper explores the development and implementation of an Automatic Number Plate Recognition system using the YOLOv8 model, a state-of-the-art deep learning framework known for its exceptional accuracy and speed in object detection tasks. The objective of the ANPR system is to reliably identify and process vehicle number plates in real-time across various environments and traffic conditions. YOLOv8’s advanced neural network architecture allows for rapid detection and classification, making it particularly well-suited for the high demands of real-time traffic monitoring and management. We discuss the integration of YOLOv8 into comprehensive ANPR system that includes image acquisition, preprocessing, number plate detection, character segmentation, and recognition. The system is designed to be highly scalable and adaptable, with capabilities to handle different plate formats and integrate seamlessly with existing traffic and law enforcement databases. Evaluation of the system demonstrates high accuracy and efficiency in number plate recognition, showing significant improvements over previous YOLO models. The implementation details are accompanied by a performance analysis, comparing the proposed YOLOv8-based system with traditional ANPR systems in terms of detection accuracy, processing speed, and operational reliability. This study confirms the viability of using YOLOv8 for ANPR applications, proposing potential enhancements and future directions for research in intelligent traffic systems. The developed model of automatic number plate recognition system using YOLO, deep learning has obtained an accuracy of 83%, precision of 75% and F1-score of 76%.