Research map / Image generation / Editing and style transfer
Image-to-Image Translation via Hierarchical Style Disentanglement
Check 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 |
|---|---|---|---|
| Organize labels as independent tags, exclusive attributes, and tag-specific styles | image-to-image translation on CelebA-HQ · CelebA-HQ multi-style | Realism (FID): 26.04 -> 21.37 (latent-guided); 22.96 -> 21.49 (reference-guided) | Helped |
| Remove feature-based local translator (ablation) | HiSD · CelebA-HQ multi-style | Disentanglement (FID): 71.85 -> 73.91 (latent-guided); 71.48 -> 71.50 (reference-guided) | Hurt |
| Use spatial-wise attention only instead of spatial- and channel-wise attention (ablation) | HiSD · CelebA-HQ multi-style | Disentanglement (FID): 71.85 -> 75.17 (latent-guided); 71.48 -> 75.17 (reference-guided) | Hurt |
| Remove tag-irrelevant conditions from the discriminator (ablation) | HiSD · CelebA-HQ multi-style | Realism (FID): 21.37 -> 20.63 (latent-guided); 21.49 -> 20.66 (reference-guided) | Helped |
| Remove tag-irrelevant conditions from the discriminator (ablation) | HiSD · CelebA-HQ multi-style | Disentanglement (FID): 71.85 -> 74.91 (latent-guided); 71.48 -> 73.61 (reference-guided) | Hurt |
| Remove adversarial objective for cycle-translation output (ablation) | HiSD · CelebA-HQ reference-guided | Realism (FID): 21.49 -> 28.39 | Hurt |
| Remove adversarial objective for cycle-translation output (ablation) | HiSD · CelebA-HQ reference-guided | Disentanglement (FID): 71.48 -> 80.34 | Hurt |
| Remove adversarial objective for cycle-translation output (ablation) | HiSD · CelebA-HQ latent-guided | Realism (FID): 21.37 -> 22.31 | No clear effect |
| Use tag-irrelevant conditional discrimination (ablation) | HiSD · CelebA-HQ multi-style | Realism (FID): 20.63 -> 21.37 (latent-guided); 20.66 -> 21.49 (reference-guided) | Hurt |
| Use tag-irrelevant conditional discrimination (ablation) | HiSD · CelebA-HQ multi-style | Disentanglement (FID): 74.91 -> 71.85 (latent-guided); 73.61 -> 71.48 (reference-guided) | Helped |
Closest papers by meaning
- Style-Guided and Disentangled Representation for Robust Image-to-Image Translation (AAAI 2022 · similarity 0.89)
- A Style-Aware Discriminator for Controllable Image Translation (CVPR 2022 · similarity 0.88)
- Smoothing the Disentangled Latent Style Space for Unsupervised Image-to-Image Translation (CVPR 2021 · similarity 0.88)
- Multi-mapping Image-to-Image Translation via Learning Disentanglement (NeurIPS 2019 · similarity 0.87)
- Free-Lunch Color-Texture Disentanglement for Stylized Image Generation (NeurIPS 2025 · similarity 0.85)
- CSGO: Content-Style Composition in Text-to-Image Generation (NeurIPS 2025 · 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.