Research map / Image generation / Image restoration
High-Fidelity and Efficient Pluralistic Image Completion With Transformers
Ziyu Wan, Jingbo Zhang, Dongdong Chen, Jing Liao
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 |
|---|---|---|---|
| use top-1 sampling instead of top-50 sampling | Ours · FFHQ/Places2 large masks (40-60%) | FID: FFHQ 20.024 -> 20.843; Places2 33.853 -> 34.206 | Mixed |
| use top-1 (deterministic) sampling instead of top-50 sampling | Ours · FFHQ (20%-40% masks) | PSNR: 27.922 -> 28.242 | Improved |
| replace PIC with transformer-prior plus CNN upsampling method | PIC · ImageNet (40%-60% masks) | FID: 101.278 -> 59.486 | Improved |
| replace PIC with transformer-prior plus CNN upsampling method | PIC · ImageNet (Random masks) | FID: 59.428 -> 35.842 | Improved |
| replace PIC with transformer-prior plus CNN upsampling method | PIC · Places2 (40-60% masks) | FID: 49.093 -> 33.853 | Improved |
| train on masks up to 60% then test on extremely large holes | Ours · extremely large masks | qualitative completion | Improved |
| use transformer prior instead of fully CNN completion for geometric shapes | DeepFillv1 / PIC · pentagram completion | qualitative shape reconstruction | Improved |
| use our method instead of PIC for diversity | PIC · Places2/FFHQ diversity | LPIPS diversity | Improved |
| our method versus baselines in user study | baselines · user study (30 images, 28 participants) | top-1 rank ratio: 73.70% | Improved |
Closest papers by meaning
- High-Fidelity Pluralistic Image Completion With Transformers (ICCV 2021 · similarity 0.99)
- Pluralistic Image Completion (CVPR 2019 · similarity 0.91)
- Don't Look into the Dark: Latent Codes for Pluralistic Image Inpainting (CVPR 2024 · similarity 0.86)
- Large Occluded Human Image Completion via Image-Prior Cooperating (AAAI 2024 · similarity 0.86)
- MaGIC: Multi-modality Guided Image Completion (ICLR 2024 · similarity 0.85)
- Completion as Enhancement: A Degradation-Aware Selective Image Guided Network for Depth Completion (CVPR 2025 · similarity 0.84)
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.
