Publication detail

License Plate Recognition on Low-Cost Devices

PŘINOSIL, J. KŘÍŽ, P.

Original Title

License Plate Recognition on Low-Cost Devices

Type

conference paper

Language

English

Original Abstract

This paper presents a comprehensive approach to vehicle license plate recognition running on low-cost devices. Leveraging convolutional neural networks, we evaluate models like YOLOv7-tiny and YuNet for license plate detection, favoring YuNet's 1080×1080 resolution for the accuracy-computation trade-off. For license plate character recognition, we proposed a YOLOv4-tiny derivation model, achieving good accuracy and fast computation. The proposed approaches were validated on a test set from real traffic using Raspberry Pi 4 as the target computational device.

Keywords

license plate recognition; image analysis; convolutional neural networks; YOLO; YuNet

Authors

PŘINOSIL, J.; KŘÍŽ, P.

Released

29. 10. 2023

Publisher

IEEE Computer Society

ISBN

979-8-3503-9328-6

Book

2023 15th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)

Pages from

28

Pages to

32

Pages count

5

BibTex

@inproceedings{BUT185598,
  author="Jiří {Přinosil} and Petr {Kříž}",
  title="License Plate Recognition on Low-Cost Devices",
  booktitle="2023 15th International Congress on Ultra Modern Telecommunications and Control Systems and Workshops (ICUMT)",
  year="2023",
  pages="5",
  publisher="IEEE Computer Society",
  isbn="979-8-3503-9328-6"
}