Research map / Detection and segmentation / CNN features
Unbiased Multi-Modality Guidance for Image Inpainting
Yongsheng Yu, Dawei Du, Libo Zhang, Tiejian 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 edge and segmentation auxiliary branches (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | PSNR: 25.88 -> 26.70 | Improved |
| add edge and segmentation auxiliary branches (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.36 -> 11.40 | Improved |
| add segmentation branch on top of edge branch (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | PSNR: 26.19 -> 26.70 | Improved |
| add edge branch on top of segmentation branch (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | PSNR: 26.47 -> 26.70 | Improved |
| drop biased prior guidance from predicted auxiliary structures (MMT-4 to MMT-9) (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.59 -> 11.40 | Improved |
| replace concat with unbiased MSSA+ADN (MMT-1 to MMT-9) (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 20.01 -> 11.40 | Improved |
| use MSSA fusion instead of element-wise addition (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.64 -> 11.40 | Improved |
| use MSSA fusion instead of conv fusion (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 11.90 -> 11.40 | Improved |
| use ADN instead of AdaIN within MSSA (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.81 -> 11.40 | Improved |
| use ADN instead of SPADE within MSSA (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.17 -> 11.40 | Improved |
| increase ACB layers from 2 to 8 (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 12.18 -> 11.40 | Improved |
| quadruple channels in AOT blocks (dagger AOT@8) (ablation) | Multi-modality guided inpainting transformer · CelebA-HQ 128x128 center mask | FID: 11.61 -> 14.17 | Hurt |
Closest papers by meaning
- Guidance and Evaluation: Semantic-Aware Image Inpainting for Mixed Scenes (ECCV 2020 · similarity 0.89)
- Contextual-based Image Inpainting: Infer, Match, and Translate (ECCV 2018 · similarity 0.89)
- Incremental Transformer Structure Enhanced Image Inpainting With Masking Positional Encoding (CVPR 2022 · similarity 0.88)
- SyFormer: Structure-Guided Synergism Transformer for Large-Portion Image Inpainting (AAAI 2024 · similarity 0.88)
- BrushNet: A Plug-and-Play Image Inpainting Model with Decomposed Dual-Branch Diffusion (ECCV 2024 · similarity 0.88)
- High-Resolution Image Inpainting Using Multi-Scale Neural Patch Synthesis (CVPR 2017 · 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.
