An Audio-Aided Face and Text Recognition System for Visually Impaired

Abstract

One of the biggest difficulties for a visually impaired person is to identify people and read text. This limits the visually impaired to interact socially and puts them at risk. In recent years, several deep learning techniques have been used for face and text recognition. This paper proposes a portable system for face and text recognition using the Raspberry Pi 3B+, Raspberry Pi camera module, customized dataset, few push button switches, and earphones. Multi-task cascaded convolutional neural network and support vector machine are used for face detection and face recognition, respectively. The proposed idea involves efficient and accurate scene text detector for text detection and Tesseract optical character recognition (OCR) for text recognition. Finally, converting the text or face labels to speech by e-Speak tool, a process which allows visually impaired people hear the name of the person/text-written in front of them.