Generative Adversarial Networks for Synthetic Defect Generation in Assembly and Test Manufacturing

Abstract

Defect detection and classification is a critical step in any semiconductor manufacturing process. Most of the time it involves manual creation of defects which is time consuming and includes a high material and labor cost. In this paper we propose Artificial Intelligence-based synthetic defect generation techniques to augment the training image sets for Convolutional Neural Network (CNNs)-based defect detection and classification systems. Specifically, we use Generative Adversarial Networks (GANs) to create various modes of the defects which are difficult to create manually. Our results indicate that the output of our adapted GANs are images of realistic-looking defects for a wide variety of common manufacturing defects including foreign material, misplaced epoxy, scratches, and die chipping defects among others.