RECONHECIMENTO DE FALHAS POR IMAGENS: um estudo sobre o incremento de inteligência artificial no processo de detecção de não conformidades

Abstract

This work aimed to build a solution to be implemented in the quality control process in order to make the procedures for locating and classifying faults faster and more accurate. For this, neural networks are used, which is an algorithm capable of inferring about a type of problem that the input data came closest to, in addition, it is a tool that works well with a large volume of data, and the work with images the volume of information is much greater than a simple problem with inputs and outputs with a few dozen inputs. Finally, the constructed model proved to be possible, and it reached a relatively satisfactory precision for the initial objective, which was to verify if such an application would really be possible. apply the algorithm in real situations.