EyeMath: Increasing Accessibility of Mathematics to Visually Impaired Readers

Mathematics education for visually impaired students is challenging because their learning materials are generally limited to braille books, and audiobooks. In order to increase the chance of learning mathematical content for people with visual impairment, this paper presents the design and development of a cloud-based mobile application called EyeMath, using serverless microservices in Amazon AWS. Users can provide images of page snippets for the application to process and read their content to the users. EyeMath segments an input image into smaller pieces and separates pieces that have only plain text from pieces with mathematical symbols. The mathematical-related pieces are further processed into an Abstract Syntax Tree (AST) and then parsed into Thai sentences. For plain text pieces, EyeMath relies on Tesseract OCR to convert them into text. Finally, results for all pieces are combined together systematically for the device’s screen reader program to read aloud. The performance evaluation of the application shows high correctness in reading math content within test images and our usability testing confirms the potential usefulness of the application to visually impaired readers.