Combining Transformer Generators with Convolutional Discriminators
This work addresses the computational and data efficiency challenges in generative adversarial networks for computer vision researchers, though it is incremental as it builds on existing TransGAN and CNN methods.
The paper tackles the problem of data-hungry transformer-based GANs by combining a transformer generator with a convolutional discriminator, removing the need for data augmentation and auxiliary tasks, and achieves better results in image synthesis benchmarks.
Transformer models have recently attracted much interest from computer vision researchers and have since been successfully employed for several problems traditionally addressed with convolutional neural networks. At the same time, image synthesis using generative adversarial networks (GANs) has drastically improved over the last few years. The recently proposed TransGAN is the first GAN using only transformer-based architectures and achieves competitive results when compared to convolutional GANs. However, since transformers are data-hungry architectures, TransGAN requires data augmentation, an auxiliary super-resolution task during training, and a masking prior to guide the self-attention mechanism. In this paper, we study the combination of a transformer-based generator and convolutional discriminator and successfully remove the need of the aforementioned required design choices. We evaluate our approach by conducting a benchmark of well-known CNN discriminators, ablate the size of the transformer-based generator, and show that combining both architectural elements into a hybrid model leads to better results. Furthermore, we investigate the frequency spectrum properties of generated images and observe that our model retains the benefits of an attention based generator.