Research map / Image generation / Editing and style transfer
Unaligned Image-to-Image Translation by Learning to Reweight
Shaoan Xie, Mingming Gong, Yanwu Xu, Kun Zhang
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 learned importance reweighting (ablation) | Baseline (beta=1) · T_B->L_E | FID: 97.82 -> 84.91 | Improved |
| add learned importance reweighting (ablation) | Baseline (beta=1) · D_A->C_H | FID: 61.28 -> 53.46 | Improved |
| add learned importance reweighting (ablation) | Baseline (beta=1) · H_S->Z_D | FID: 106.84 -> 79.40 | Improved |
| add learned importance reweighting (ablation) | Baseline (beta=1) · A_U->S | FID: 115.39 -> 119.86 | Hurt |
| add learned importance reweighting (ablation) | Baseline (beta=1) · C_H->D_A | KID x100: 1.89 -> 2.07 | Hurt |
| add learned importance reweighting (ablation) | Baseline (beta=1) · Aligned-subset selection on H_S | accuracy: 0.55 -> 0.96 | Improved |
| add learned importance reweighting (ablation) | Baseline (beta=1) · Aligned-subset selection on Z_D | accuracy: 0.58 -> 0.79 | Improved |
| add learned importance reweighting (ablation) | Baseline (beta=1) · Aligned-subset selection on T_B | accuracy: 0.50 -> 0.83 | Improved |
| add effective sample size loss with lambda_ESS=0 or very low (ablation) | IrwGAN · C_H<->D_A | importance weight sparsity | Hurt |
| raise lambda_ESS to 10 (ablation) | IrwGAN · C_H<->D_A | importance weights | Hurt |
| raise unaligned anime ratio from 0% to 300% (ablation) | IrwGAN · S->A_U | FID: 95.25 -> 95.44 | No clear effect |
| replace CycleGAN with IrwGAN on unaligned domains | CycleGAN · S->A_U | FID: 100.43 -> 93.75 | Improved |
Closest papers by meaning
- Few-Shot Unsupervised Image-to-Image Translation (ICCV 2019 · similarity 0.87)
- Unpaired Image-to-Image Translation using Adversarial Consistency Loss (ECCV 2020 · similarity 0.86)
- Unsupervised Image-to-Image Translation Networks (NeurIPS 2017 · similarity 0.86)
- COCO-FUNIT: Few-Shot Unsupervised Image Translation with a Content Conditioned Style Encoder (ECCV 2020 · similarity 0.86)
- Unsupervised Image-to-Image Translation Using Domain-Specific Variational Information Bound (NeurIPS 2018 · similarity 0.86)
- Towards Identifiable Unsupervised Domain Translation: A Diversified Distribution Matching Approach (ICLR 2024 · similarity 0.85)
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.
