Research map / Image generation / Autoregressive and latent generation
Guidance and Evaluation: Semantic-Aware Image Inpainting for Mixed Scenes
Liang Liao, Jing Xiao, Zheng Wang, Chia-Wen Lin, Shin’ichi Satoh
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 SGIM (semantic-guided inference module) to Basic-Net (ablation) | Basic-Net · Outdoor Scenes (centering holes) | PSNR: 19.14 -> 19.58 | Improved |
| add SGIM (semantic-guided inference module) to Basic-Net (ablation) | Basic-Net · Outdoor Scenes (centering holes) | SSIM: 0.71 -> 0.77 | Improved |
| add SGIM (semantic-guided inference module) to Basic-Net (ablation) | Basic-Net · Outdoor Scenes (centering holes) | FID: 43.43 -> 41.49 | Improved |
| add SCEM (segmentation confidence evaluation) to SG-Net (ablation) | SG-Net · Outdoor Scenes (centering holes) | PSNR: 19.58 -> 20.53 | Improved |
| add SCEM (segmentation confidence evaluation) to SG-Net (ablation) | SG-Net · Outdoor Scenes (centering holes) | SSIM: 0.77 -> 0.81 | Improved |
| add SCEM (segmentation confidence evaluation) to SG-Net (ablation) | SG-Net · Outdoor Scenes (centering holes) | FID: 41.49 -> 40.67 | Improved |
| add SCEM to SG-Net | SG-Net · Outdoor Scenes (irregular holes) | FID: 41.74 -> 42.47 | Hurt |
| add SCEM to SG-Net | SG-Net · Cityscapes (irregular holes) | SSIM: 0.64 -> 0.75 | Improved |
| add SCEM to SG-Net | SG-Net · Cityscapes (centering holes) | PSNR: 23.04 -> 23.41 | Improved |
| semantic guidance via segmentation-map-conditioned inference (SG-Net) instead of edge-guided two-stage EC | EC · Outdoor Scenes (irregular holes) | SSIM: 0.83 -> 0.81 | Hurt |
| replace human-annotated segmentation labels with predicted labels from DPN/Deeplab v3+ for training SGE-Net (ablation) | SGE-Net · Outdoor Scenes and Cityscapes | inpainting quality | No clear effect |
| use SGE-Net over SPG in a user preference study | SPG · User study (80 images) | preference rate (%): 73.2 | Improved |
Closest papers by meaning
- Image Inpainting Guided by Coherence Priors of Semantics and Textures (CVPR 2021 · similarity 0.91)
- Semantic Image Inpainting With Deep Generative Models (CVPR 2017 · similarity 0.90)
- Coherent Semantic Attention for Image Inpainting (ICCV 2019 · similarity 0.89)
- Contextual-based Image Inpainting: Infer, Match, and Translate (ECCV 2018 · similarity 0.89)
- Structure Matters: Tackling the Semantic Discrepancy in Diffusion Models for Image Inpainting (CVPR 2024 · similarity 0.87)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.86)
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.
