A Study on Navigation System for Visually Impairment Person based on Egocentric Vision using Deep Learning

Abstract

There are 310 thousand the visually impaired people in Japan. They use a white cane during walking outside the home. However, the white cane is difficult to get surrounding information (Ex. Traffic light, Signboard, etc...). The aim of the research is to develop a system that the impaired people are able to walk safely. Specifically, this research uses an egocentric vision and is able to recognize “traffic light”, “Signboard”, etc... using deep learning in real time. Therefore, the system is able to estimate dangerous situation, feedback the result to the user.