Research map / Image generation / Autoregressive and latent generation
Prior Guided GAN Based Semantic Inpainting
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 |
|---|---|---|---|
| replace per-image iterative latent optimization with a learned noise-prior predictor | Yeh et al. iterative inpainting · SVHN, Stanford Cars, and CelebA | FID: 3.9 -> 3.6; 4.8 -> 4.3; 4.5 -> 4.1; 5.4 -> 5.0; 5.8 -> 5.1; 6.8 -> 6.7 | Helped |
| replace per-image iterative latent optimization with a learned noise-prior predictor | Yeh et al. iterative inpainting · SVHN, Stanford Cars, and CelebA | PSNR: 21.3 -> 21.1; 16.5 -> 17.0; 15.1 -> 14.8; 12.2 -> 12.5; 23.8 -> 23.4; 20.1 -> 20.3 | Mixed |
| add structural priors (ablation) | Mz · CelebA | FID: 5.1 -> 4.9; 6.7 -> 6.0 | Helped |
| add structural priors (ablation) | Mz · CelebA | PSNR: 23.4 -> 24.1; 20.3 -> 21.7 | Helped |
| add structural priors (ablation) | Mz high-resolution · CelebA-HQ, 96x96 holes | FID: 4.1 -> 3.8 | Helped |
| use predicted rather than actual class label (ablation) | Mz high-resolution · ImageNet, 96x96 and 128x128 holes | FID: 4.5 -> 6.5; 14.5 -> 17.8 | Hurt |
| add structural priors (ablation) | Mz high-resolution · FaceForensics | video-FID: 0.738 -> 0.710 | Helped |
| add grouped LSTM prior prediction (ablation) | Mz high-resolution · FaceForensics | video-FID: 0.738 -> 0.680 | Helped |
| add structural priors and grouped LSTM prior prediction (ablation) | Mz high-resolution · FaceForensics | video-FID: 0.738 -> 0.660 | Helped |
| predict the latent prior in one feed-forward pass instead of iterative optimization | Yeh et al. iterative inpainting · 64×64 and 128×128 inpainting | inference speedup: around 800×; 780× and 820× | Helped |
Closest papers by meaning
- Semantic Image Inpainting With Deep Generative Models (CVPR 2017 · similarity 0.89)
- Learning Pyramid-Context Encoder Network for High-Quality Image Inpainting (CVPR 2019 · similarity 0.88)
- Generative Image Inpainting With Contextual Attention (CVPR 2018 · similarity 0.88)
- Image Inpainting via Iteratively Decoupled Probabilistic Modeling (ICLR 2024 · similarity 0.87)
- MTADiffusion: Mask Text Alignment Diffusion Model for Object Inpainting (CVPR 2025 · similarity 0.86)
- Reconstructing the Image Stitching Pipeline: Integrating Fusion and Rectangling into a Unified Inpainting Model (NeurIPS 2024 · 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.