SD-GAN uses the EMA generator as a teacher to distill perceptual knowledge to the training generator, improving FID scores, stabilizing training, and providing guidance uncorrelated with standard adversarial loss.
Title resolution pending
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
fields
cs.CV 4roles
background 2polarities
background 2representative citing papers
SemGANs is a novel GAN architecture for pixel-accurate semantic image generation that outperforms standard GANs both quantitatively and qualitatively.
SPADE-LDM conditional synthesis from composite semantic masks produces realistic 3D LGE MRI that raises LA cavity Dice from 0.908 to 0.936.
MSDS computes DeepSSIM at multiple pyramid scales and fuses the scores with learned weights, producing consistent improvements over single-scale DeepSSIM on IQA benchmarks with negligible extra cost.
citing papers explorer
-
Improving Generative Adversarial Networks with Self-Distillation
SD-GAN uses the EMA generator as a teacher to distill perceptual knowledge to the training generator, improving FID scores, stabilizing training, and providing guidance uncorrelated with standard adversarial loss.
-
Adversarial Pixel-Level Generation of Semantic Images
SemGANs is a novel GAN architecture for pixel-accurate semantic image generation that outperforms standard GANs both quantitatively and qualitatively.
-
3D Conditional Image Synthesis of Left Atrial LGE MRI from Composite Semantic Masks
SPADE-LDM conditional synthesis from composite semantic masks produces realistic 3D LGE MRI that raises LA cavity Dice from 0.908 to 0.936.
-
MSDS: Deep Structural Similarity with Multiscale Representation
MSDS computes DeepSSIM at multiple pyramid scales and fuses the scores with learned weights, producing consistent improvements over single-scale DeepSSIM on IQA benchmarks with negligible extra cost.