Empowering Accessibility: A Braille Translator App with Enhanced User Experience and Integration of Machine Learning Technologies

Abstract

This paper proposes a Braille Translation Mobile Application. It will be designed to seamlessly translate Braille content into several languages while offering speech output functionality. It aims to enable visually impaired individuals who have not yet had the chance to learn braille to read it, as well as aid them to learn it by testing their skills using the app. Moreover, it will allow any sighted individuals in their circle to be able to read the same thing they are reading. For the backend of the application, the YOLO-v8 model will identify and locate the objects in the image using the datasets. Datasets used are ‘Double-Sided Braille Image Dataset’ by Li et al., 2018, “Angelina Reader”, made by Ilya Ovodov for training and testing AI models. OCR will be used to convert the images involved into data. For braille translation, GPT-4 API has been used, and Open AI's TTS API was used for speed output functionality. As for the frontend, Flutter will be used so that the application can be accessible for both Android and iOS users all around the world. The application will have a user-friendly interface that presents both the Braille image that has been sent and detected, and its translation. Lastly, the application will provide many customization options based on the preferences of the users and accessibility needs.