Research map / Image generation / Editing and style transfer
One-Shot Adaptation of GAN in Just One CLIP
Gihyun Kwon, Jong Chul Ye
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 CLIP-space consistency loss L_con (ablation) | Ours · Sketches (one-shot) | FID: 89.77 -> 83.87 | Improved |
| add CLIP-space consistency loss L_con (ablation) | Ours · Sketches (one-shot) | identity score: 0.266 -> 0.443 | Improved |
| add patch-wise consistency loss L_patch (ablation) | Ours · Sketches (one-shot) | FID: 98.34 -> 83.87 | Improved |
| add CLIP-guided reference search instead of II2S (ablation) | Ours · Sketches (one-shot) | FID: 92.50 -> 83.87 | Improved |
| use L2 regression L_con instead of softmax with KL divergence (ablation) | Ours · Sketches (one-shot) | FID: 89.49 -> 83.87 | Improved |
| use CLIP-space L_con instead of generator-feature consistency of Ojha et al. (ablation) | Ours · Sketches (one-shot) | identity score: 0.211 -> 0.443 | Improved |
| use L_patch instead of CUT contrastive regularization (ablation) | Ours · Sketches (one-shot) | FID: 115.16 -> 83.87 | Improved |
| add reference-target matching with I_ref (ablation) | Ours · Sketches (one-shot) | FID: 88.83 -> 83.87 | Improved |
| add mixing regularization (ablation) | Ours · Sketches (one-shot) | FID: 90.88 -> 83.87 | Improved |
| add mixing regularization (ablation) | Ours · Sketches (one-shot) | LPIPS diversity: 0.554 -> 0.572 | Improved |
| use the proposed method instead of MTG in 3-shot adaptation | Ours vs MTG · 3-shot Sketches | FID: 101.2 -> 70.13 | Improved |
| use the proposed method instead of StyleGAN-NADA for text-guided adaptation (user score) | Ours vs StyleGAN-NADA · Text-guided adaptation | mean opinion score: 2.46 -> 3.64 | Improved |
Closest papers by meaning
- Generalized One-shot Domain Adaptation of Generative Adversarial Networks (NeurIPS 2022 · similarity 0.89)
- Smoothness Similarity Regularization for Few-Shot GAN Adaptation (ICCV 2023 · similarity 0.88)
- WeditGAN: Few-Shot Image Generation via Latent Space Relocation (AAAI 2024 · similarity 0.88)
- Improving Diversity in Zero-Shot GAN Adaptation with Semantic Variations (ICCV 2023 · similarity 0.88)
- One-Shot Generative Domain Adaptation (ICCV 2023 · similarity 0.88)
- E$^2$GAN: Efficient Training of Efficient GANs for Image-to-Image Translation (ICML 2024 · similarity 0.87)
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.
