Research map / Image generation / Editing and style transfer
U-GAT-IT: Unsupervised Generative Attentional Networks with Adaptive Layer-Instance Normalization for Image-to-Image Translation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 U-GAT-IT rather than the strongest listed baseline | selfie2anime · selfie2anime | KID ×100: 13.08 ± 0.49 -> 11.61 ± 0.57 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | horse2zebra · horse2zebra | KID ×100: 7.58 ± 0.71 -> 7.06 ± 0.8 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | cat2dog · cat2dog | KID ×100: 8.15 ± 0.48 -> 7.07 ± 0.65 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | photo2portrait · photo2portrait | KID ×100: 1.20 ± 0.31 -> 1.79 ± 0.34 | Hurt |
| Use U-GAT-IT rather than the strongest listed baseline | photo2vangogh · photo2vangogh | KID ×100: 4.26 ± 0.29 -> 4.28 ± 0.33 | No clear effect |
| Use U-GAT-IT rather than the strongest listed baseline | anime2selfie · anime2selfie | KID ×100: 11.84 ± 0.74 -> 11.52 ± 0.57 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | zebra2horse · zebra2horse | KID ×100: 8.0 ± 0.66 -> 7.47 ± 0.71 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | dog2cat · dog2cat | KID ×100: 9.45 ± 0.64 -> 8.15 ± 0.66 | Helped |
| Use U-GAT-IT rather than the strongest listed baseline | portrait2photo · portrait2photo | KID ×100: 1.42 ± 0.24 -> 1.69 ± 0.53 | Hurt |
| Use U-GAT-IT rather than the strongest listed baseline | vangogh2photo · vangogh2photo | KID ×100: 4.68 ± 0.36 -> 5.61 ± 0.32 | Hurt |
| Add AdaLIN instead of using IN only (ablation) | U-GAT-IT · selfie2anime | KID ×100: 13.64 ± 0.76 -> 11.61 ± 0.57 | Helped |
| Add AdaLIN instead of using LN only (ablation) | U-GAT-IT · selfie2anime | KID ×100: 12.39 ± 0.61 -> 11.61 ± 0.57 | Helped |
Closest papers by meaning
- Unsupervised Attention-guided Image-to-Image Translation (NeurIPS 2018 · similarity 0.90)
- Unsupervised Image-to-Image Translation Networks (NeurIPS 2017 · similarity 0.87)
- DA-GAN: Instance-Level Image Translation by Deep Attention Generative Adversarial Networks (CVPR 2018 · similarity 0.87)
- Unsupervised Image-to-Image Translation With Generative Prior (CVPR 2022 · similarity 0.87)
- TraVeLGAN: Image-To-Image Translation by Transformation Vector Learning (CVPR 2019 · similarity 0.87)
- SEAN: Image Synthesis With Semantic Region-Adaptive Normalization (CVPR 2020 · 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.