EyeRis: Visual Image Recognition using Machine Learning for the Visually-Impaired

Abstract

Visually impaired people struggle daily and have difficulty recognizing and distinguishing objects around them. Thus, they mainly depend on supervision from other people to assist them. Since smartphones have become a necessity in this modern world, the researchers formulated a solution to help the visually impaired through a machine learning-based mobile application for object recognition. Nowadays, software applications can provide accurate findings in picture classification and processing processes thanks to machine learning techniques and algorithms. In this study, the researchers use a convolutional neural network (CNN)-based system on TensorFlow Lite to create a mobile version of a visual information system employing a machine learning strategy and deep learning framework. The main objectives of the smartphone application, EyeRis, is to recognize and categorize items in real-time and to separate photographs from the user-selected scenarios. Results were analyzed and contrasted based on the app’s obtained recognition accuracy data. It demonstrated the utility of CNN as a model for image recognition algorithms.