A Neural Network for License Plate Detection and Recogntion

This paper presents the information about the vehicle plate detection and recognition with the character segmentation. This system depends on YOLO for detection and Optical Character Recognition (OCR) for segmentation process. YOLO is a method that provides real-time object detection using neural networks. The popularity of this algorithm is due to its accuracy and quickness. This system evaluates the different models of license plate and aims to improve the accuracy or speed at each and every stage of process. The implementation of Convolutional Neutral Network (CNN) gives more accurate and effective results for detection and segmentation stages. CNN is trained for Vehicle License Plate Recognition (VLPR) to get more accurate results in different conditions (like poor-lightening conditions) CNN gives the detected and rectified vehicle license plate images. Optical Character recognition is capable for recognizes the numbers as well as characters in detected image. By using YOLO, CNN and OCR the experimental results are very accurate without any adaption, it outperforms the detection and recognition of license plate is very accurate and provides the enhancement images and converted text format data as a final output. When compared to state-of-the-arts, the suggested YOLO architecture performs well with a licence plate identification accuracy of 98.7%.