Deep learning based approach for vehicle license plate recognition
Loading...
Date
2020-11
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand)
Abstract
In the field of Number Plate Recognition System Artificial Intelligence and Deep Learning have played a crucial role in recent years. Therefore this work aims to establish a model of plate detection with the aid of Deep Learning. It gives us an idea of the method of object detection which gives the utmost precision. The aim of this research is to extract accurate precise and optimize the number plate image recognition as compared to the previous research work done. It is therefore necessary to implement the ANPR system model based on an effective Deep Learning algorithms. An effective License Plate Recognition System is needed to ensure the successful functioning of intelligent transportation system. During the last few years, several methods of image processing such as OpenCV have been developed. These have not been exploited in the previous researches on number plate detection. Our study is based on Image Segmentation using object detection algorithm through YOLO (You Only Look Once) object detection technique in darkflow framework and character recognition based on Convolutional Neural Network (CNN). YOLO and CNN have proved to be the most productive techniques for several supervised and unsupervised learning tasks. In this research, we study deep learning algorithms and compare their performance. We have used custom dataset to for image segmentation and character recognition. Different categories of models for object detection is classified by performing segmentation using annotated images for detecting license plates by YOLO. Further, for recognizing characters, single character recognition using CNN is used and also trained a model as a base learners. Finally, compared the performance of the model by previous research for plate detection and have drawn useful conclusions. The result shows that Deep Learning algorithms outperform with high accuracy as compared to other image processing techniques.