Research map / Image generation / GANs
Omni-GAN: On the Secrets of cGANs and Beyond
Peng Zhou, Lingxi Xie, Bingbing Ni, Cong Geng, Qi Tian
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 |
|---|---|---|---|
| add weight decay to Omni-GAN (ablation) | Omni-GAN · CIFAR100 | FID: 26.51 -> 8.41 | Improved |
| add weight decay to AC-GAN (ablation) | AC-GAN · CIFAR100 | FID: 34.19 -> 16.41 | Improved |
| add weight decay to projection-based BigGAN (ablation) | BigGAN · CIFAR100 | FID: 10.12 -> 13.73 | Hurt |
| add differentiable augmentation to BigGAN (ablation) | BigGAN · CIFAR100 | FID: 10.12 -> 8.63 | Improved |
| add DiffAug to AC-GAN without weight decay (ablation) | AC-GAN · CIFAR100 | FID: 34.19 -> 38.01 | No clear effect |
| add DiffAug to Omni-GAN with weight decay (ablation) | Omni-GAN · CIFAR100 | FID and IS: 8.41 -> 11.39 FID; 13.30 -> 15.37 IS | Mixed |
| replace AC-GAN loss with the improved loss (ImAC-GAN) | AC-GAN with weight decay · CIFAR100 | FID: 16.41 -> 8.11 | Improved |
| add implicit neural representation to Omni-GAN generator (ablation) | Omni-GAN with weight decay · CIFAR100 | FID: 8.41 -> 7.63 | Improved |
| add DiffAug to Omni-INR-GAN with weight decay (ablation) | Omni-INR-GAN · CIFAR10 | FID: 5.25 -> 4.32 | Improved |
| use Omni-GAN instead of BigGAN | BigGAN · ImageNet 128x128 | IS: 104.57 -> 190.94 | Improved |
| add INR to Omni-GAN generator output (ablation) | Omni-GAN · ImageNet 128x128 | IS: 190.94 -> 262.85 | Improved |
| use Omni-GAN instead of BigGAN | BigGAN · ImageNet 256x256 | IS: 187.60 -> 304.05 | Improved |
Later papers that compared with this method
2 later accepted papers report OmniGAN as a baseline in their results tables; 2 of them report a better number on most of the results they share with it.
- NICE: NoIse-modulated Consistency rEgularization for Data-Efficient GANs (NeurIPS 2023 · better on 2 of 2 shared results)
- Manifold Learning Benefits GANs (CVPR 2022 · better on 12 of 12 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- RoCGAN: Robust Conditional GAN (IJCV 2020 · similarity 0.90)
- A Unified View of cGANs with and without Classifiers (NeurIPS 2021 · similarity 0.90)
- GAN Compression: Efficient Architectures for Interactive Conditional GANs (CVPR 2020 · similarity 0.90)
- OSSGAN: Open-Set Semi-Supervised Image Generation (CVPR 2022 · similarity 0.89)
- MCGAN: Enhancing GAN Training with Regression-Based Generator Loss (AAAI 2025 · similarity 0.89)
- Diversity-Sensitive Conditional Generative Adversarial Networks (ICLR 2019 · 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.
