Inexpensive voice assisted smart eyewear for visually impaired persons in context of Bangladesh

Abstract

This paper describes the design of affordable smart eyewear for visually impaired people. Among the total population of Bangladesh, a significant number of people are visually impaired. Most of them relied on others which makes life challenging. They lag behind their fellow mates in terms of academic perspective. To help these people, the proposed system can play a vital role in today's world. Through this smart eyewear, visual input will convert into the audible signal by a Raspberry Pi 4B. Convolutional neural network (CNN) has been introduced to classify objects. Moreover, Optical character recognition (OCR) is available for recognizing Bangla and English text. Using speech recognition API users may able to control electrical gadgets and communicate with caregivers through a mobile app. In the research, we focused on running multiple complex algorithms on a Raspberry Pi in an optimized way and found satisfactory result which can be an affordable solution for developing country like Bangladesh.