Design Of An Ai-Based Smart Wearable Device For Visually Impaired People

Abstract

The human visual system plays an important role in perceiving information related to the surrounding environment. Because visual cues provide more data than auditory information, visual cues are more effective than auditory cues when it comes to perceiving the information. However, in the case of visually impaired people, the lack of visual information will limit them in identifying information and people around them. Therefore, along with medical treatment technology, technology supports visually impaired people when moving is being studied and becoming urgent. Realizing the high applicability of recognition technology with artificial intelligence in helping the visually impaired, in
this article, we propose and develop a smart wearable device that recognizes faces for visually impaired people applying deep learning techniques. The device was deployed on a compact, flexible hardware platform with low cost to identify and announce the name of the person
identified to the visually impaired person through the voice. The results of experiment and evaluation show that the proposed device is extremely versatile and has achieved remarkable accuracy of 99.3% in a practical environment.