International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 10 Issue: 07 | July 2023
p-ISSN: 2395-0072
www.irjet.net
An Enhancement of Braille Character Perception Using Deep Learning and Artificial Intelligence Techniques Ms. A.Varshini 1, Dr.S. Gopinathan2 1 M.Phil Research Scholar, Department of Computer Science, University of Madras, Tamil Nadu, India. 2 Professor, Department of Computer Science, University of Madras, Tamil Nadu, India.
---------------------------------------------------------------------***--------------------------------------------------------------------marks, and even special symbols can be represented in Abstract - India is the largest popular country in the world. Braille.
Due to shortage of vitamins, by birth 1% of the children are born with blindness. Blind people start their career with Braille. Braille makes reading and writing possible for visually impaired people. Braille-to-text translation serves as a bridge between the two worlds. The communication process has been simplified by advances in technology. This research article describes how to convert Braille to English using deep learning. In this work, 26 English Braille Images are used as a dataset after the segmentation process. The proposed method converts braille visuals to English text using convolutional neural network CNN models such as LeNet, VGG-16, DenseNet121, ResNet50, and Inceptionv3. The implemented system Inceptionv3 achieves a high level of prediction accuracy of 92%. Experiments show that the proposed Braille character recognition gives an accurate result.
Understanding the patterns of raised dots and their corresponding characters is fundamental in recognizing and interpreting Braille text accurately.
Key Words: Braille, Deep Learning, Convolutional Neural Network, LeNet, VGG-16, DenseNet121, ResNet50, and Inceptionv3.
1. INTRODUCTION Over 2.2 billion people worldwide are blind or visually impaired, according to the World Health Organization (WHO) [1]. The people having impairments of vision utilize Braille, a system of pointed dots that may be read by finger contact, as it is difficult for them to read and write text [2]. A Braille cell or block is made up of six dots that are organized in two columns and three rows. Each dot within the cell can be either raised or not raised, representing different combinations of dots that correspond to specific characters. Counting starts at the top and goes down to the bottom.
Fig. 2. Braille Alphabet Board 1.1 Challenges In Braille images, it is challenging to preprocess the images to remove all the noise that is introduced. After all the braille cells have been recognized, they need to be translated into the appropriate English alphabet. When a single alphabet is represented by recognized braille cells, there is a chance that they might have significant variations.
Fig. 1. Braille Cell By combining different combinations of raised dots within the cell, all the letters of the alphabet, numbers, punctuation
© 2023, IRJET
|
Impact Factor value: 8.226
Fig. 3. Empty Circle Braille Dot
|
ISO 9001:2008 Certified Journal
|
Page 797