Automatic Number Plate Recognition System using Machine Learning Techniques

The system employs video processing technology to detect the image of a vehicle's number plate from a video stream. Subsequently, various algorithms are applied to extract the number from the detected plate. This system can be used at college entrances, cantonment area and other places with strict access restrictions. The system detects the number plate whenever a vehicle passes by, converting it into both images and text. Subsequently, this information is stored in the database for future reference or analysis. This research paper introduces ANPR system which is designed to optimize accuracy, efficiency, and ease of implementation which includes combination of Python programming language, YOLOv8 (You Only Look One-level), and EasyOCR, the proposed system addresses the complexities associated with detection of license plate and recognition in diverse real-world situations. Core of the system lies in the adoption of YOLOv8, an object detection model renowned for its real-time capabilities and high accuracy. EasyOCR, a user-friendly Optical Character Recognition (OCR) library, is seamlessly integrated into the ANPR system. This addition enhances the system's ability to accurately extract alphanumeric information from detected license plates, bridging the gap between detection and recognition phases. Due to this the system ensures efficient noise reduction, image segmentation, and overall improvement in the quality of input images, contributing to the system's reliability. The paper presents an in-depth evaluation of the ANPR system's performance through datasets. Additionally, the research explores potential applications of the ANPR system in real-world scenarios, emphasizing its applicability in traffic management, law enforcement, and security systems.