Identificación de figuras geométricas en un sistema de visión basándose en el entrenamiento de una red neuronal artificial convolucional utilizando Python

Abstract

The image processing projects through vision systems, are a great didactic support point in the mechatronics career, since they have wide application in the industry in the process lines primarily to perform assembly, inspection, selection and component placement. One of the methods used is to apply artificial neural networks for the identification of images and a factor to analyze is the evaluation of the learning capacities of these networks in the identification of geometric figures. In this article, the training of a convolutional artificial neural network using Python is presented. This type of work is focused on joining projects based on industry 4.0 that may contain link options with process systems based on these technologies. In this work, a vision system based on python programming was made and has its contribution in the libraries that were designed and can be linked to different types of applications within a manufacturing process.