Dataset Multi Braille Characters

Published: 2 January 2026| Version 1 | DOI: 10.17632/s4pmywvwbc.1
Contributor:
Made Ayu Dusea Widyadara

Description

This dataset contains a collection of Braille character images (Multi-Braille Characters) designed to support the development of pattern recognition systems. Dataset Details: Contents: A collection of single and combined (multi-character) Braille character images. Format: Images in .jpg format, both unlabeled and with appropriate labels. Source: Data obtained through scanning Braille books and data augmentation.

Files

Steps to reproduce

1. Acquire raw Braille images using a camera 12 MP, f/1.9, 1/2.55", 1.4µm, dual pixel PDAF 5 MP, f/2.0, (depth) with side lighting to emphasize Braille dot shadows. 2. Convert all captured images to JPG format and standardize image resolution. 3. Perform image cropping and grid-based segmentation according to the defined dataset structure. 4. Apply image augmentation including grayscale conversion, brightness (±25%), exposure (±20%), saturation (±40%), and hue (±30°). 5. Annotate all images into 105 Braille classes

Categories

Computer Vision, Image Processing, Assistive Technology, Optical Character Recognition

Licence