Proceedings of the 2nd International Conference on ICT for Digital, Smart, and Sustainable Development, ICIDSSD 2020, 27-28 February 2020, Jamia Hamdard, New Delhi, India

Research Article

Automatic License Plate Detection and Recognition (ALPDR)

Download843 downloads
  • @INPROCEEDINGS{10.4108/eai.27-2-2020.2303122,
        author={Mohd Amaan Abbasi and Anam  Saiyeda},
        title={Automatic License Plate Detection and Recognition (ALPDR)},
        proceedings={Proceedings of the 2nd International Conference on ICT for Digital, Smart, and Sustainable Development, ICIDSSD 2020, 27-28 February 2020, Jamia Hamdard, New Delhi, India},
        publisher={EAI},
        proceedings_a={ICIDSSD},
        year={2021},
        month={3},
        keywords={character recognition deep learning image recognition},
        doi={10.4108/eai.27-2-2020.2303122}
    }
    
  • Mohd Amaan Abbasi
    Anam Saiyeda
    Year: 2021
    Automatic License Plate Detection and Recognition (ALPDR)
    ICIDSSD
    EAI
    DOI: 10.4108/eai.27-2-2020.2303122
Mohd Amaan Abbasi1,*, Anam Saiyeda1
  • 1: Jamia Hamdard, Hamdard Nagar, New Delhi-110062, India
*Contact email: amaanabbasi99@gmail.com

Abstract

This work attempts to solve the problem of license plate detection and recognition of its characters. It also aims to improve various parts of the detection and recognition pipeline and proposing new state of the art techniques to make the pipeline robust. The cases with naïve approach are discussed, which could be a better option than a deep learning technique. The goal is to make the model accurate and fast, so as it can produce real-time results. The model is kept light weighted and formatted in such a way that it is easily deployable to different platforms. This paper gives an overview of when and when not to use a Deep learning model. Dataset collection is a big challenge and deep learning models rely heavily on it; a work around to this challenge is employed by implementing a naive technique. This model requires almost no data compared to that of a Deep learning model and performs better. The result of this report will give a solution to automate the task of manually writing down the characters of a license plate. The proposed solution can be used at campuses, universities, airports, shopping complexes and hotels for intelligent surveillance.