Research map / Image generation / Autoregressive and latent generation
Generative Image Inpainting With Contextual Attention
Read the paperCheck 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 |
|---|---|---|---|
| Add contextual attention to the two-stage baseline | Generative inpainting model · Places2 validation | ℓ1 loss: 9.4% -> 8.6% | Helped |
| Add contextual attention to the two-stage baseline | Generative inpainting model · Places2 validation | PSNR: 18.15 -> 18.91 | Helped |
| Add contextual attention to the two-stage baseline | Generative inpainting model · Places2 validation | ℓ2 loss: 2.4% -> 2.1% | Helped |
| Add contextual attention to the two-stage baseline | Generative inpainting model · Places2 validation | TV loss: 25.7% -> 25.3% | Helped |
| Replace contextual attention with spatial transformer attention (ablation) | Unified inpainting network · Image inpainting | Visual inpainting quality | Hurt |
| Replace contextual attention with appearance flow (ablation) | Unified inpainting network · Image inpainting | Visual inpainting quality | Hurt |
| Use DCGAN loss instead of WGAN-GP (ablation) | Inpainting framework · Places2 and CelebA | Training stability | Hurt |
| Use LSGAN instead of WGAN-GP (ablation) | Inpainting framework · Image inpainting | Inpainting quality | Hurt |
| Remove ℓ1 reconstruction loss from the refinement network (ablation) | Inpainting framework · Image inpainting | Content structure and training regularization | Hurt |
| Add perceptual loss (ablation) | Inpainting framework · Image inpainting | Inpainting quality | No clear effect |
| Add style loss (ablation) | Inpainting framework · Image inpainting | Inpainting quality | No clear effect |
| Add total variation loss (ablation) | Inpainting framework · Image inpainting | Inpainting quality | No clear effect |
Closest papers by meaning
- High-Resolution Image Inpainting Using Multi-Scale Neural Patch Synthesis (CVPR 2017 · similarity 0.92)
- Semantic Image Inpainting With Deep Generative Models (CVPR 2017 · similarity 0.91)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (CVPR 2025 · similarity 0.87)
- Image Inpainting via Iteratively Decoupled Probabilistic Modeling (ICLR 2024 · similarity 0.87)
- Brush2Prompt: Contextual Prompt Generator for Object Inpainting (CVPR 2024 · similarity 0.87)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · 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.