NAYAN-Narrator for the Visually Impaired

Abstract

Worldwide there are millions of blind people and even more number of visually impaired people. Disability of optical text and image reading features a huge impact on the standard of life for visually disabled people. Although there are several devices designed for providing a helping hand to the visually disabled to work out objects using an alternating sense like sound and touch, the event of text reading and image narrating devices are still at a nascent stage. In the current scenario, systems involving image captioning or text recognition are generally limited either by explicitly hoping on color and shape identification or by requiring manual assistance or even thanks to the high cost. Therefore, a low-cost and effective system is presented that can be automatically mapped and helps to skim the text clearly yet explain images to visually impaired people. Our important objective is to acknowledge the text character and pictures present in books, novels, letters, etc. and explain the content in sequence through speech signals just like the human narration.

Keywords: Neural networks, Visually impairment, Image captioning, CNN, RNN, LSTM, OCR, Text-to-speech