<![CDATA[3D Printing is currently needed in various industries, including education in terms of research development. In this study, researchers classify 3D printing defect images to recognize images that are difficult to see with the naked eye. With limited observation, an image classification method is needed to help users detect defects in the printing process with a Deep Learning model. The printing process uses PLA and ABS-based filament materials, which are mostly used in 3D Printing objects with fused deposition modeling (FDM)-based 3D Printer machines. In this study, there are several stages, including data augmentation, model development using sequential CNN, pre-trained CNN based with pre-trained models, namely VGG-16 and VGG-19, training, validation, and model evaluation. The dataset taken for training is 1557, with a ratio of 80 percent training and 20 percent validation between defective and non-defective objects. The results of this study have a good accuracy value on Sequential CNN with an accuracy of 99.68% compared to pre-trained CNN models, namely VGG-16 and VGG-19. The classification results are also compared with other additive manufacturing classification methods with different machines and materials such as metal and 3D Food Printing which are measured based on classification model optimization analysis]]>
Copyrights © 2026
