Braille Block Recognition Using Convolutional Neural Network and Guide for Visually Impaired People
Toshiaki Okamoto, Tomoyuki Shimono, Yuichi Tsuboi, Mayuko Izumi, Yousuke Takano · 2020
Braille blocks are the main tools to assist visually impaired people. However, it is difficult for visually impaired people to find braille blocks. As a matter of fact, there are many accidents such that visually impaired people fall from a platform. Therefore, it is important to detect braille blocks and inform them of where braille blocks are. This research aims to guide visually impaired people to braille blocks. In order to detect braille blocks, a camera is used and Convolutional Neural Network (CNN) is applied as the method of image recognition. In this paper, the theory of CNN and guide method are explained and the structure of CNN used in the experiment is shown. In the experiment, CNN is established learnt from training data and the recognition accuracy of braille blocks is confirmed with test data. The results of the experiments show that the test accuracy is approximately 94%.