Visual Impairment Assistance System

Abstract

Visually challenged people have no knowledge of outdoor hazards and require visual aids to avoid threats. Furthermore, they face many difficulties compared to other people in doing their day to day work. Even though they have guide canes to help them it becomes difficult to recognize the faces of the people. They find it difficult to identify the familiar as well as unknown faces. Recognizing familiar faces confidently without anyone’s help is a very difficult task for a blind person. This model uses deep learning techniques that can identify familiar and unknown faces for visually impaired people. Apart from recognizing faces, blind people find it difficult to read simple signboards and important instructions. A few of the examples are the menu cards of the restaurants, Direction boards, locating washrooms, etc.  The proposed model will help the visually challenged people to read the images by converting the image into a text and then converting the same text which is read into audio. For converting the image to text, OCR technology was used. Recognizing face is a very important task for the visually challenged people as a failure of this may even lead to life-threatening problems and reading sign boards and other simple text is also very important in the daily life of visually challenged people. Hence the development of this visual impairment assistance system will help the visually challenged people to live a better life.