Deep Learning-Based Refreshable Braille Display for Learning Braille Fundamentals

Learning for the visually impaired has been mainly based on Braille, a haptic assistive technology that aids in reading and writing through an embossed dot. However, the currently available braille in the market is expensive and not reusable, making it more unreasonable for its price, which results in low braille literacy in the country. These problems led to the conceptualization of a three-character refreshable braille display integrated with deep learning optical character recognition (OCR). The development of the device aims to instruct braille fundamentals not just among the visually impaired but also among all Filipinos who wish to learn braille. By following the Iterative Project Development Model, an operational device was successfully developed. The device was found to be effective in providing an introduction or teaching braille fundamentals. However, the functionality of improving braille literacy was not achieved due to the time required to test one's literacy. Test results show an unfavorable response time due to device limitations, but evaluation presents that integrating GPU into the device would greatly improve the response time.