The Effect of Illuminance Change on Accuracy in Determining Warpage of 3D Printer Modeled Objects Using Convolutional Neural Network

Fused Deposition Modeling, one of the techniques for layered fabrication, has a problem of warpage during fabrication, and there is a need for a technique to avoid this problem. As one method to avoid this problem, a convolutional neural network (CNN) system has been developed to determine the presence or absence of warpage in real time and pause the printing process if warpage is detected. However, the system is based on the assumption that clear images are used, and it is unclear whether the system can determine warpage in an actual manufacturing environment, where the input image to determine warpage may be unclear. Therefore, the purpose of this study is to investigate whether changes in the quality of input images due to changes in illuminance affect the accuracy of warpage judgment by training the CNN proposed in the previous study using images taken in an environment where imaging conditions, such as illuminance, have been adjusted, and using images taken at different illuminance levels as test images. The results of the investigation revealed that the accuracy of warpage judgment in the test image was highest when the illuminance was the same as that of the training image, and that the accuracy of warpage judgment decreased when the illuminance was lower than that of the training image.