Efficient Multi-Object Detecting Assistive System for Visually Impaired People

Abstract Technology these days is growing drastically, and of which the expense of using this technology is increasing leading to the increase in the price of the devices used by visually impaired people. There is a need for a portable and easily accessible smart aid. This proposed system allows the unperceptive community to have access to such a device that is portable and affordable. People with deprived vision face difficulties in navigation and carrying out their daily life chores, our device aims to empower such people. ESP32CAM (Esprissif Systems 32-Bit Camera) captures images of objects/intruders, converting them into text and then into audio which alerts the user through headphones/speakers. The YOLO (You Only Look Once) package is installed, which scans the images and performs augmentation and annotation, before feature extraction to perform classification. Once classified, the image is then converted into audio. The COCO (Common Objects in Context) dataset is used for implementation. The purpose of this device is to process the images of objects as input into a voice as output. It has less error rate, less processing time, and is cost-effective. This device acts as an artificial eye for visually impaired people.