In the intelligent transportation system the automatic license plate recognition and detection plays a very important role. This application could be used for traffic control security e-payment systems in the toll pay and parking. Many algorithms have been developed to force license plate detection and recognition and all have many advantages and flaws under different situations. With the advent and rise of deep learning concepts in various fields of artificial intelligence, computer vision has developed in leaps and bounds in terms of innovations and methods. Automatic License Plate Recognition has emerged as an effective method to automate the watch keeping process for traffic systems, parking fee structures, and police surveillance. License plate recognition (LPR) is a quite used and mature technology but much work is needed to be done in order to make it perfect. In recent years, the scientific community has made major advances in methodology and performance. This paper tries to aim at summarizing and analyzing various methodologies and progress in LPR in the deep learning era using IOT sensors. Hence, in this paper, an Automatic License Plate Detection and Recognition (ALPDR) system has been proposed having four steps namely License Plate Extraction, Image Pre-processing, Character Segmentation and Character Recognition. For the first three steps (extraction, pre-processing, and segmentation), unique methods have been proposed. As the character recognition is an important step of license plate recognition and detection, four different methods for character recognition have been experimented on, which include Convolution Neural Network (CNN), MobileNet, Inception V3, ResNet 50.
Read full abstract