VISION- Wearable Speech Based Feedback System for the Visually Impaired using Computer Vision
Leo Abraham, Nikita Sara Mathew, Liza George, Shebin Sam Sajan
Abstract
With the advent of new technologies, a whole lot realm of possibilities is open to mankind which otherwise would have been either impossible or a miracle. Computer vision is one such field which has millions of possibilities and this project itself being a primary example of it. This project aims to help the blind society to experience the world independently with the help of a speech-based feedback device. This project proposes (1) identifying walkable spaces, (2) text recognition and text-to-speech, (3) identify and locate specific types of objects; and walking navigation which can be incorporated into this project as a future scope. This is implemented with the YOLO algorithm for object detection, which uses the COCO dataset. Our project will help a blind person to walk easily by finding the path, detect obstacles in front of them and thus avoid it. It will help them read texts as well, which is done using OCR which uses python and an API for the text recognition. Thereafter, gTTS is used to convert text to speech, which is the final output for the users.