Adversarial Style Augmentation for Domain Generalization

Yabin Zhang, Bin Deng, Ruihuang Li +2

It is well-known that the performance of well-trained deep neural networks may degrade significantly when they are applied to data with even slightly shifted distributions. Recent…