Multinational License Plate Recognition Using Generalized Character Sequence Detection


The computer vision community considers automatic license plate recognition (ALPR) to be a solved problem. However, the majority of current ALPR research is focused on license plates (LP) from certain nations and uses country-specific data, limiting their practical application. To work on the LPs of other nations, such ALPR systems require algorithm adjustments. Previous work on multinational LP recognition has been validated using datasets from a variety of nations with similar LP layouts. To overcome this problem, this research proposes a deep ALPR system that may be used by multinational LPs. LP detection, unified character recognition, and multinational LP layout detection are the three primary processes in the proposed method. The system is mostly built on YOLO networks (you only look once). For the first stage, little YOLOv3 was utilized, while for the second step, YOLOv3-SPP was employed, which is a variant of YOLOv3 that includes the spatial pyramid pooling (SPP) block. For character recognition, the localized LP is sent into YOLOv3-SPP. The bounding boxes of the anticipated characters are returned by the character recognition network, but no information about the LP number sequence is provided. A erroneous LP number is one with an improper sequence. As a result, we offer a layout detection method that can extract the correct sequence of LP numbers from multinational LPs in order to retrieve the correct sequence. We compiled our own Korean automobile plate (KarPlate) dataset and made it available to the public. The suggested approach was tested on LP datasets from South Korea, Taiwan, Greece, the United States, and Croatia. In addition, a small dataset of LPs from 17 nations was gathered to test the global LP layout identification algorithm's performance. The suggested ALPR system takes roughly 42 milliseconds per image to extract the LP number on average. Our ALPR system's efficiency is demonstrated by experimental findings.

Did you like this research project?

To get this research project Guidelines, Training and Code... Click Here

PROJECT TITLE : Scaling Up Generalized Kernel Methods ABSTRACT: Over the course of the past two decades, kernel methods have enjoyed a great deal of success. During this era of big data, the collection of data has experienced
PROJECT TITLE : Enhancing Factorization Machines with Generalized Metric Learning ABSTRACT: The cold-start and data sparsity issues that plague recommender systems can be circumvented with the help of factorization machines (FMs),
PROJECT TITLE : Quantum Blockchain Based on Dimensional Lifting Generalized Gram-Schmidt Procedure ABSTRACT: The development of quantum computers compromises the integrity of classical blockchains, making it necessary either
PROJECT TITLE : Generalized Bayesian Model Selection for Speckle on Remote Sensing Images ABSTRACT: Coherent summation of back-scattered waves and subsequent nonlinear envelope changes introduce speckle noise into both synthetic
PROJECT TITLE :Improved Low-Complexity Sphere Decoding for Generalized Spatial Modulation - 2018ABSTRACT:During this letter, two types of improved sphere decoding (SD) algorithms for generalized spatial modulation (GSM), termed

Ready to Complete Your Academic MTech Project Work In Affordable Price ?

Project Enquiry