Research map / Image generation / Editing and style transfer
TuiGAN: Learning Versatile Image-to-Image Translation with Two Unpaired Images
Jianxin Lin, Yingxue Pang, Yingce Xia, Zhibo Chen, Jiebo Luo
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 cycle-consistency loss (ablation) | TuiGAN · Horse->Zebra | SIFID (x10^-4): 3.29 -> 1.03 | Improved |
| add cycle-consistency loss (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 5.61 -> 1.79 | Improved |
| add identity loss (ablation) | TuiGAN · Horse->Zebra | SIFID (x10^-4): 2.43 -> 1.03 | Improved |
| add identity loss (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 5.54 -> 1.79 | Improved |
| add total variation loss (ablation) | TuiGAN · Horse->Zebra | SIFID (x10^-4): 2.41 -> 1.03 | Improved |
| add total variation loss (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 10.85 -> 1.79 | Improved |
| add attention module in generators instead of simple addition (ablation) | TuiGAN · Horse->Zebra | SIFID (x10^-4): 1.08 -> 1.03 | Improved |
| add attention module in generators instead of simple addition (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 2.09 -> 1.79 | Improved |
| add attention module in generators instead of simple addition (ablation) | TuiGAN · Horse->Zebra | PD: 8.00 -> 6.16 | Improved |
| increase number of scales from N=0 to N=4 (ablation) | TuiGAN · Horse->Zebra | SIFID (x10^-4): 2.26 -> 1.03 | Improved |
| increase number of scales from N=0 to N=4 (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 3.75 -> 1.79 | Improved |
| increase number of scales from N=3 to N=4 (ablation) | TuiGAN · Zebra->Horse | SIFID (x10^-4): 6.30 -> 1.79 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DRIT — DRIT++: Diverse Image-to-Image Translation via Disentangled Representations (IJCV 2020) · better on 1 of 1 shared result
- SinGAN — SinGAN: Learning a Generative Model From a Single Natural Image (ICCV 2019) · better on 1 of 1 shared result
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
- Few-Shot Unsupervised Image-to-Image Translation (ICCV 2019 · similarity 0.89)
- Domain-Scalable Unpaired Image Translation via Latent Space Anchoring (TPAMI 2023 · similarity 0.89)
- Unsupervised Image-to-Image Translation With Generative Prior (CVPR 2022 · similarity 0.89)
- Unpaired Image-To-Image Translation Using Cycle-Consistent Adversarial Networks (ICCV 2017 · similarity 0.89)
- Domain Adaptive Image-to-Image Translation (CVPR 2020 · similarity 0.89)
- Unsupervised Image-to-Image Translation Using Domain-Specific Variational Information Bound (NeurIPS 2018 · similarity 0.88)
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.
