Detecção de Falhas do Tipo Spaghetti em Impressoras 3D Utilizando Visão Computacional com Redes Neurais Convolucionais

Abstract

This work presents the development of an intelligent system for the automatic detection of spaghetti-type failures in FDM 3D printers. The solution uses convolutional neural networks (CNNs) applied to the analysis of images captured in real time, integrated with OctoPrint to pause the printing process and send email notifications to the user. The model was trained using a proprietary dataset and achieved high accuracy in classifying normal and defective prints. The results demonstrate that the system reduces material waste and increases the safety of the 3D printing process.