Research map / Image generation / Autoregressive and latent generation
Diverse Image Inpainting with Normalizing Flow
Cairong Wang, Yiming Zhu, Chun Yuan
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 |
|---|---|---|---|
| raise first stage resolution from 32 x 32 to 64 x 64 (ablation) | Flow-Fill · CelebA-HQ ablation (mask 20-60%) | FID: 12.44 -> 11.29 | Improved |
| raise first stage resolution from 48 x 48 to 64 x 64 (ablation) | Flow-Fill · CelebA-HQ ablation (mask 20-60%) | FID: 11.86 -> 11.29 | Improved |
| raise first stage resolution from 64 x 64 to 96 x 96 (ablation) | Flow-Fill · CelebA-HQ ablation (mask 20-60%) | FID: 11.29 -> 13.19 | Hurt |
| raise first stage resolution from 64 x 64 to 256 x 256 (ablation) | Flow-Fill · CelebA-HQ ablation (mask 20-60%) | FID: 11.29 -> 41.94 | Hurt |
| raise first stage resolution from 64 x 64 to 256 x 256 (ablation) | Flow-Fill · CelebA-HQ ablation (mask 20-60%) | PSNR: 25.60 -> 22.31 | Hurt |
| use Flow-Fill instead of BAT | BAT · CelebA-HQ (random mask 20-60%) | l1 error (%): 3.18 -> 2.34 | Improved |
| use Flow-Fill instead of BAT | BAT · CelebA-HQ (random mask 20-60%) | FID: 9.33 -> 11.29 | Hurt |
| use Flow-Fill instead of BAT | BAT · Places2 (random mask 20-60%) | FID: 22.16 -> 25.40 | Hurt |
| use Flow-Fill instead of BAT | BAT · Paris StreetView (random mask) | PSNR: 24.50 -> 24.58 | Improved |
| use Flow-Fill instead of BAT | BAT · Paris StreetView (random mask) | FID: 48.19 -> 50.18 | Hurt |
| use Flow-Fill instead of ICT for sampling | ICT · Places2 inference time per image (random mask) | time (s): 13.03s -> 0.19s | Improved |
| use Flow-Fill instead of Palette (diffusion) | Palette · Places2 inference time per image (random mask) | time (s): 27.13s -> 0.19s | Improved |
Closest papers by meaning
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.87)
- RAD: Region-Aware Diffusion Models for Image Inpainting (CVPR 2025 · similarity 0.87)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.87)
- DiverseFlow: Sample-Efficient Diverse Mode Coverage in Flows (CVPR 2025 · similarity 0.86)
- InpDiffusion: Image Inpainting Localization via Conditional Diffusion Models (AAAI 2025 · similarity 0.86)
- Video Diffusion Models Are Strong Video Inpainter (AAAI 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.
