Smart Blind Assistant using Deep Learning for the Visually Impaired Users

In today’s modern world, visually impaired and blind people experience numerous difficulties in driving their daily need and life. They had undergone many challenges in strange environment without a manual aid who insist in helping them. As the technology is gradually increasing day by day, there exists various opportunities in helping out the blind and visually impaired people. The proposed work is to help the visually impaired and blind people to lead their normal life without any manual aid by using various technologies in soft computing techniques. The proposed blind assistant methodology makes use of following components namely Camera (5MP), Raspberry-Pi 3, power bank and speaker. The images are acquired from the scene using camera in the mode of video, pre-process the images using Deep Neural Network (DNN) which is a part of deep learning techniques, with the help of openCV. The outcomes of the proposed technique identify the individual and multiple obstacles in the outdoor and indoor surroundings. The obtained mean accuracy of the assistive system to identify the barriers in the proposed method is around 81.5%.