Research map / Image generation / Editing and style transfer
Marginal Contrastive Correspondence for Guided Image Generation
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 |
|---|---|---|---|
| Add correspondence (COR) to SPADE (ablation) | SPADE · CelebA-HQ | FID: 32.02 -> 15.77 | Helped |
| Add marginal contrastive loss (MCL) to COR (ablation) | MCL-Net correspondence · CelebA-HQ | FID: 15.77 -> 12.75 | Helped |
| Add self-correlation map (SCM) to COR (ablation) | MCL-Net correspondence · CelebA-HQ | FID: 15.77 -> 13.27 | Helped |
| Add both MCL and SCM to COR (ablation) | MCL-Net correspondence · CelebA-HQ | FID: 15.77 -> 12.52 | Helped |
| Add MCL with increasing margin m from 0.1 to 0.4 (ablation) | CoCosNet correspondence · DeepFashion warped exemplars | L1: 79.32 -> 75.27 | Helped |
| Add SCM to CoCosNet correspondence (ablation) | CoCosNet correspondence · DeepFashion warped exemplars | L1: 80.14 -> 76.07 | Helped |
| Combine MCL (m=0.4) and SCM (ablation) | CoCosNet correspondence · DeepFashion warped exemplars | L1: 80.14 -> 73.72 | Helped |
| Use a large MCL margin such as m=0.5 (ablation) | MCL-Net · DeepFashion | convergence | Hurt |
| Use MCL-Net instead of CoCosNet v2 | Image translation · CelebA-HQ | FID: 12.85 -> 12.52 | Helped |
| Use MCL-Net instead of CoCosNet v2 | Image translation · DeepFashion | SWD: 16.53 -> 16.24 | Helped |
| Use MCL-Net instead of CoCosNet v2 | Image translation · ADE20K | LPIPS: 0.564 -> 0.569; UNITE: 0.571 | Hurt |
Closest papers by meaning
- Cross-Domain Correspondence Learning for Exemplar-Based Image Translation (CVPR 2020 · similarity 0.93)
- Masked and Adaptive Transformer for Exemplar Based Image Translation (CVPR 2023 · similarity 0.89)
- CoCosNet v2: Full-Resolution Correspondence Learning for Image Translation (CVPR 2021 · similarity 0.88)
- Exemplar Guided Unsupervised Image-to-Image Translation with Semantic Consistency (ICLR 2019 · similarity 0.88)
- FashionR2R: Texture-preserving Rendered-to-Real Image Translation with Diffusion Models (NeurIPS 2024 · similarity 0.86)
- Learning to Manipulate Artistic Images (AAAI 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.