Research map / Image generation / Autoregressive and latent generation
Learning Pyramid-Context Encoder Network for High-Quality Image 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 |
|---|---|---|---|
| Use PEN-Net instead of the strongest baseline PConv | Image inpainting on Places2 · Places2, 128×128 square masks | MS-SSIM: 74.67% -> 78.09% | Helped |
| Use PEN-Net instead of the strongest baseline PConv | Image inpainting on Places2 · Places2, 128×128 square masks | IS: 47.00 -> 50.51 | Helped |
| Use PEN-Net instead of the strongest baseline CA | Image inpainting on Places2 · Places2, 128×128 square masks | FID: 18.34 -> 15.19 | Helped |
| Use PEN-Net instead of the strongest baseline PConv | Image inpainting on Places2 · Places2, 128×128 square masks | L1 loss: 8.92 -> 9.94 | Hurt |
| Add pyramid filling with cross-layer attention transfer instead of using a single ATN (ablation) | Single ATN on a U-Net backbone · Places2, 128×128 square masks | MS-SSIM: 71.61% -> 78.09% | Helped |
| Add pyramid filling with cross-layer attention transfer instead of using a single ATN (ablation) | Single ATN on a U-Net backbone · Places2, 128×128 square masks | IS: 37.02 -> 50.51 | Helped |
| Add pyramid filling with cross-layer attention transfer instead of using a single ATN (ablation) | Single ATN on a U-Net backbone · Places2, 128×128 square masks | FID: 26.38 -> 15.19 | Helped |
| Add pyramid filling with cross-layer attention transfer instead of using a single ATN (ablation) | Single ATN on a U-Net backbone · Places2, 128×128 square masks | L1 loss: 9.85 -> 9.94 | Hurt |
Closest papers by meaning
- Generative Image Inpainting With Contextual Attention (CVPR 2018 · similarity 0.90)
- Context Encoders: Feature Learning by Inpainting (CVPR 2016 · similarity 0.90)
- High-Resolution Image Inpainting Using Multi-Scale Neural Patch Synthesis (CVPR 2017 · similarity 0.89)
- Semantic Image Inpainting With Deep Generative Models (CVPR 2017 · similarity 0.89)
- Image Inpainting via Generative Multi-column Convolutional Neural Networks (NeurIPS 2018 · similarity 0.89)
- Reconstructing the Image Stitching Pipeline: Integrating Fusion and Rectangling into a Unified Inpainting Model (NeurIPS 2024 · similarity 0.85)
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.