In this paper, we discuss a method to detect defects in industrial products by using Denoising AutoEncoder Generative Adversarial Networks. In previous methods, a defective area is detected by restoring a defective product image which added an artificial defect to a non-defective product image by Denoising AutoEncoder (DAE). Therefore, a defective area is detected by subtracted image of them. We discuss whether further accuracy improvement is possible by introducing a framework of adversarial learning to DAE in order to restore a defective image to a non-defective image clearer.
1 January 2004 in IEEJ Transactions on Electronics, Information and Systems
Detection of bananas that ripen naturally and artificially using digital images is quite difficult because they physically have almost the same characteristics. Feature identification requires very detailed characteristic information so that a diverse feature extraction process is needed, but not necessarily the extracted ones are the optimal features. In this study, the convolution neural network (CNN) method was used and the best features were learned automatically during the classification process. Two CNN-based methods, VGG based model and ResNet18, were tested to obtain the optimal identification. The results of testing on six Indonesian banana cultivars showed that VGG based model produced better accuracy of 88.67%. However, ResNest18 achieved convergence faster.