An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning

Wajahat Kazmi, Ian Nabney, George Vogiatzis, Peter Rose and Alex Codd

Bristol Research (University of Bristol) · 2019

Abstract

This paper presents an industrial system to read<br/>text on tyre sidewalls. Images of vehicle tyres in motion are<br/>acquired using roadside cameras. Firstly, the tyre circularity is<br/>detected using Circular Hough Transform (CHT) with dynamic<br/>radius detection. The tyre is then unwarped into a rectangular<br/>patch and a cascade of convolutional neural network (CNN)<br/>classifiers is applied for text recognition. We introduce a novel<br/>proposal generator for localizing the tyre code by combining<br/>Histogram of Oriented Gradients (HOG) with a CNN. The<br/>proposals are then filtered using a deep network. After the code<br/>is localized, character detection and recognition are carried out<br/>using two separate deep CNNs. The end-to-end system presents<br/>impressive accuracy and efficiency proving its suitability for the<br/>intended industrial application.

Citation

Wajahat Kazmi, Ian Nabney, George Vogiatzis, Peter Rose and Alex Codd. “An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning.” Bristol Research (University of Bristol), pp. 1074–1079. 2019.

@article{kazmi2019,
  title     = {An industrial system for vehicle tyre detection and text recognition using a pipeline of conventional image processing and deep learning},
  author    = {Wajahat Kazmi and Ian Nabney and George Vogiatzis and Peter Rose and Alex Codd},
  journal   = {Bristol Research (University of Bristol)},
  pages     = {1074--1079},
  year      = {2019},
}