AI Paper Atlas

Research map / Image generation / GANs

StudioGAN: A Taxonomy and Benchmark of GANs for Image Synthesis

Minguk Kang, Joonghyuk Shin, Jaesik Park

TPAMI 2023 · Image generation › GANs · approach: GAN latent optimization and evaluation · 85 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use standing statistics instead of moving average BN statistics at evaluation (ablation)BigGAN · CIFAR10FID: 4.65 -> 4.16Improved
use standing statistics instead of moving average BN statistics at evaluation (ablation)BigGAN · ImageNetFID: 10.23 -> 8.54Improved
use standing statistics instead of batch statistics at evaluation (ablation)BigGAN · ImageNetFID: 10.86 -> 8.54Improved
add adaptive discriminator augmentation (ADA)StyleGAN2 · CIFAR10FID: 3.78 -> 2.31Improved
add DiffAugStyleGAN2 · CIFAR10FID: 3.78 -> 2.39Improved
add LeCam regularizationStyleGAN2 · CIFAR10FID: 3.78 -> 4.65Hurt
add adaptive pseudo augmentation (APA)StyleGAN2 · CIFAR10FID: 3.78 -> 4.04Hurt
add D2D-CE lossStyleGAN2 · CIFAR10FID: 3.78 -> 3.44Improved
add DiffAugBigGAN · CIFAR10FID: 4.16 -> 3.35Improved
add LeCam regularizationBigGAN · CIFAR10FID: 4.16 -> 3.23Improved
add consistency regularization (CRGAN)BigGAN · CIFAR10FID: 4.16 -> 3.16Improved
scale class-conditional StyleGAN2 from Baby-ImageNet to full ImageNetStyleGAN2 · ImageNetFID: 22.21 -> 33.40Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy