Research map / Image generation / GANs
StudioGAN: A Taxonomy and Benchmark of GANs for Image Synthesis
Minguk Kang, Joonghyuk Shin, Jaesik Park
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use standing statistics instead of moving average BN statistics at evaluation (ablation) | BigGAN · CIFAR10 | FID: 4.65 -> 4.16 | Improved |
| use standing statistics instead of moving average BN statistics at evaluation (ablation) | BigGAN · ImageNet | FID: 10.23 -> 8.54 | Improved |
| use standing statistics instead of batch statistics at evaluation (ablation) | BigGAN · ImageNet | FID: 10.86 -> 8.54 | Improved |
| add adaptive discriminator augmentation (ADA) | StyleGAN2 · CIFAR10 | FID: 3.78 -> 2.31 | Improved |
| add DiffAug | StyleGAN2 · CIFAR10 | FID: 3.78 -> 2.39 | Improved |
| add LeCam regularization | StyleGAN2 · CIFAR10 | FID: 3.78 -> 4.65 | Hurt |
| add adaptive pseudo augmentation (APA) | StyleGAN2 · CIFAR10 | FID: 3.78 -> 4.04 | Hurt |
| add D2D-CE loss | StyleGAN2 · CIFAR10 | FID: 3.78 -> 3.44 | Improved |
| add DiffAug | BigGAN · CIFAR10 | FID: 4.16 -> 3.35 | Improved |
| add LeCam regularization | BigGAN · CIFAR10 | FID: 4.16 -> 3.23 | Improved |
| add consistency regularization (CRGAN) | BigGAN · CIFAR10 | FID: 4.16 -> 3.16 | Improved |
| scale class-conditional StyleGAN2 from Baby-ImageNet to full ImageNet | StyleGAN2 · ImageNet | FID: 22.21 -> 33.40 | Hurt |
Closest papers by meaning
- Are GANs Created Equal? A Large-Scale Study (NeurIPS 2018 · similarity 0.92)
- The GAN is dead; long live the GAN! A Modern GAN Baseline (NeurIPS 2024 · similarity 0.91)
- Scaling Up GANs for Text-to-Image Synthesis (CVPR 2023 · similarity 0.91)
- MSG-GAN: Multi-Scale Gradients for Generative Adversarial Networks (CVPR 2020 · similarity 0.91)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.90)
- RG-GAN: Dynamic Regenerative Pruning for Data-Efficient Generative Adversarial Networks (AAAI 2024 · similarity 0.89)
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.
