MATLAB Application to Userfriendly Design Fully Convolutional Data Description Models for Defect Detection of Industrial Products and Its Concurrent Visualization

In this paper, fully convolutional data description (FCDD) approach is applied to the defect detection and its concurrent visualization for industrial products and materials. Our developed MATLAB application has already allowed users to efficiently and user-friendly design, train and test various kinds of neural network (NN) models for defect detection. Supported models have been originally designed convolutional newral network (CNN), transfer learning-based CNN, NN-based suport vector machine (SVM), convolutional auto encoder (CAE), variational auto encoder (VAE), fully convolution network (FCN) such as U-Net, and YOLO, however, FCDD has not been provided yet. This paper includes the software development of the MATLAB application extended to be able to build FCDD models. In particular, a systematic threshold determination method is proposed to get the best performance for defect detection from FCDD models. Also, through three different kinds of defect detection experiments, the usefulness and effectiveness of FCDD models in terms of defect detection and its concurrent visualization of understanding are quantitatively and qualitatively evaluated by comparing conventional transfer learning-based CNN models.