Virtual Assistant for Visually Impaired

Abstract

The field of artificial intelligence has led to various virtual assistants such as Siri in iPhone, Google Allo, Microsoft Cortana, and so on. Even after such progression, very little has been done to implement these technologies to assist the visually impaired community. Recognizing a person or distinguishing an object, these tasks are straightforward for common people but can be very difficult for people that are partly or completely blind. Their lives can be made smoother by assisting them to detect what is present in front of them at that instant. We aim to develop a system/assistant that will serve to guide a visually impaired person and will indicate the person by speaking through the earpiece. The system will help the person recognize people, add new faces and detect objects that are in their vicinity. We will have a mobile application which will consist of numerous deep learning models that will help applications increase its administration. The primary working of the system will consist of the camera continuously feeding images for inputs, the core system processing this input information and the earpiece acting as the output device to provide this output to the user.

Keywords: Face Recognition, Object Detection, Cognitive Services, Text-to-Speech, Deep Learning