swMATH ID: 43459
Software Authors: Kwonjoon Lee, Huiwen Chang, Lu Jiang, Han Zhang, Zhuowen Tu, Ce Liu
Description: ViTGAN: Training GANs with Vision Transformers. Recently, Vision Transformers (ViTs) have shown competitive performance on image recognition while requiring less vision-specific inductive biases. In this paper, we investigate if such observation can be extended to image generation. To this end, we integrate the ViT architecture into generative adversarial networks (GANs). We observe that existing regularization methods for GANs interact poorly with self-attention, causing serious instability during training. To resolve this issue, we introduce novel regularization techniques for training GANs with ViTs. Empirically, our approach, named ViTGAN, achieves comparable performance to state-of-the-art CNN-based StyleGAN2 on CIFAR-10, CelebA, and LSUN bedroom datasets.
Homepage: https://arxiv.org/abs/2107.04589
Source Code:  https://github.com/mlpc-ucsd/ViTGAN
Dependencies: Python
Related Software: Wasserstein GAN; CIPS-3D; Swin Transformer; Deceive D; MaskGIT; VQ-Diffusion; Caffe; MMGeneration; LOGAN; CircleGAN; InfoGAN; NumPy; BigGAN; OpenCV; TensorFlow; SphereGAN; MMD GAN; AttnGAN; sngan_projection; InstaGAN
Cited in: 0 Documents