Sorting algorithm for product classification using deep learning

Abstract

This article presents the development of a simulated product sorting environment through identification and localization using regional convolutional neural networks. This type of deep learning network allows us to identify three types of products and their location in the scene, which results in using a sorting algorithm by product type and allows us to determine the inventory level of each of them. The network presents 100% identification within the three trained classes and a robotic arm is used in a simulated environment to manipulate each of the products according to their coordinates, which facilitates the preparation of orders. The relocation of personnel dedicated to these tasks is proposed. reducing negative effects on their physical health, speeding up the preparation of orders, which allows a better and more timely response to the customer.