Toward Device Assisted Identification of Grocery Store Sections and Items for the Visually Impaired

摘要

There are systems developed to assist the visually impaired in grocery shopping. The developed systems require physical work from the users, wireless connections and products database to obtain the products information. Here comes the need for a system that assist the visually impaired person in grocery shopping without any additional devices. The system should use object recognition algorithms instead of wireless connections and database to recognize products. In the paper, a software system was created to solve the problem of assisting the visually impaired in grocery shopping. The created software accomplished only the mission of recognizing the grocery products on shelves. The paper recommends implementing a shopping cart consists of three cameras installed vertically on one side of the cart. We conducted a comparison between two object recognition algorithms to recognize products on the aisle shelves. A multimodal system created to fuse the results of the used object recognition algorithms.