Smart Glasses Application System for Visually Impaired People Based on Deep Learning

Abstract

The recent development of deep learning has promoted object detection to make rapid progress. In recent years, smart wearable technology is rapidly becoming part of everyday life, including watches, glasses and many other wearable items have been enriched with technology. In this study, we propose a smart glasses application system for visually impaired people based on deep learning. The system can use voice response to visually impaired people about the objects in front of them by uploading the photos to our backend object detection system through the camera function of smart glasses, and then download the text descriptions of the result and then use the text-to-speech function. According to the experimental results, the proposed system takes 3.788 seconds from uploading photos to making voice results, and the success rate of object detection is 96.3%. We hope that this application system can assistant visually impaired people to understand the surrounding environment and interact more closely with the people around them to improve the quality of life of them.