In contrast to typical GANs, a U-Net GAN uses a segmentation network as the discriminator. This segmentation network predicts two classes: real and fake. In doing so, the discriminator gives the generator region-specific feedback. This discriminator design also enables a CutMix-based consistency regularization on the two-dimensional output of the U-Net GAN discriminator, which further improves image synthesis quality.
How diffusion models work: the math from scratch
Video restoration and visual quality enhancment using U-net
Understanding GAN Loss Functions
Attention UNET and its Implementation in TensorFlow - Idiot Developer
Frontiers Toward Sharing Brain Images: Differentially Private
U-Net and Transformer Combined! Introducing Swin Unet, a new
Understanding Pix2Pix GAN. The name itself says “Pixel to Pixel
Diffusing Away From GANs and Transformers
Medical image segmentation using deep learning: A survey - Wang
42: A blog on A.I.
Modality specific U-Net variants for biomedical image segmentation