Research map / Image generation / Image restoration
Image Inpainting via Generative Multi-column Convolutional Neural Networks
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 |
|---|---|---|---|
| Use GMCNN instead of CA | Image inpainting · Paris street view-100 | PSNR: 23.78 -> 24.65 | Helped |
| Use GMCNN instead of CA | Image inpainting · ImageNet-200 | PSNR: 22.44 -> 22.43 | No clear effect |
| Use GMCNN instead of CA | Image inpainting · Places2-2K | PSNR: 20.03 -> 20.16 | Helped |
| Use GMCNN instead of CA | Image inpainting · CelebA-HQ-2K | PSNR: 23.98 -> 25.70 | Helped |
| Use multiple parallel columns instead of a single encoder-decoder (ablation) | GMCNN · Paris street view | PSNR: 23.75 -> 24.65 | Helped |
| Use multiple parallel columns instead of coarse-to-fine processing (ablation) | GMCNN · Paris street view | PSNR: 23.63 -> 24.65 | Helped |
| Use varied rather than fixed receptive fields across branches (ablation) | GMCNN · Paris street view | PSNR: 24.36 -> 24.65 | Helped |
| Use varied rather than fixed receptive fields across branches (ablation) | GMCNN · Paris street view | SSIM: 0.8644 -> 0.8650 | Mixed |
| Add ID-MRF regularization (ablation) | GMCNN · Paris street view | PSNR: 24.62 -> 24.65 | Helped |
| Add ID-MRF regularization (ablation) | GMCNN · Paris street view | SSIM: 0.8657 -> 0.8650 | Hurt |
| Set λ_mrf to 0.02 rather than 0.002 (ablation) | GMCNN · Paris street view | PSNR: 24.36 -> 24.64 | Helped |
| Use confidence-driven rather than spatial-discounted reconstruction loss (ablation) | Single-column CNN · Paris street view | visual quality | Helped |
Closest papers by meaning
- Generative Image Inpainting With Contextual Attention (CVPR 2018 · similarity 0.90)
- Image Inpainting via Iteratively Decoupled Probabilistic Modeling (ICLR 2024 · similarity 0.86)
- Don't Look into the Dark: Latent Codes for Pluralistic Image Inpainting (CVPR 2024 · similarity 0.86)
- Paint by Inpaint: Learning to Add Image Objects by Removing Them First (CVPR 2025 · similarity 0.86)
- Reconstructing the Image Stitching Pipeline: Integrating Fusion and Rectangling into a Unified Inpainting Model (NeurIPS 2024 · similarity 0.86)
- RAD: Region-Aware Diffusion Models for Image Inpainting (CVPR 2025 · 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.