Object Recognition System for Visually Impaired People

Abstract

One of the biggest problems that visually Impaired (VI) individuals face in their daily lives is object detection and recognition. A model is created for an object detector that can detect items for VI persons and other important uses by recognizing them at a specific distance. Existing object detection algorithms necessitate a huge amount of training data, which takes longer time, more complicated, and it is a difficult process. As a result, a computer vision notion for converting an object to text was developed using the Caffemodel framework by importing a pretrained dataset model. The Mobilenet SSD method is then used to translate the texts into speech. On a single screen, this system can detect many objects. It aids visually challenged people in detecting objects in real time. This technology can also be put into any portable gadget to assist visually impaired people to recognize items at a certain distance.