Research map / Image generation / Super-resolution
Parallel Multi-Resolution Fusion Network for Image Inpainting
Wentao Wang, Jianfu Zhang, Li Niu, Haoyu Ling, Xue Yang, Liqing Zhang
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 |
|---|---|---|---|
| add partial convolution with inpainting priority (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | FID: 10.49 -> 8.56 (HRNet to w/o PConv) | Improved |
| add partial convolution with inpainting priority (full vs w/o PConv) (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | PSNR: 25.007 -> 25.475 | Improved |
| add attention-guided representation fusion (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | FID: 8.41 -> 6.98 | Improved |
| add attention-guided representation fusion (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | PSNR: 25.283 -> 25.475 | Improved |
| keep all five mask-aware fusion modules instead of removing all (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | FID: 11.32 -> 6.98 | Improved |
| keep the first three mask-aware fusion modules (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | FID: 7.95 -> 6.98 | Improved |
| keep the first mask-aware fusion module (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | FID: 7.43 -> 6.98 | Improved |
| keep all mask-aware fusion modules (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | relative l1 (%): 3.85 -> 3.54 | Improved |
| full multi-resolution network vs HRNet baseline (ablation) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | SSIM: 0.831 -> 0.841 | Improved |
| proposed network vs EC (strongest PSNR baseline) | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | PSNR / FID: PSNR 25.700 -> 25.475; FID 13.58 -> 6.98 | Mixed |
| proposed network vs MEDFE | parallel multi-resolution fusion network · Places2 (average over 6 mask ratios) | relative l1 (%): 4.60 -> 3.54 | Improved |
Closest papers by meaning
- High-Resolution Image Inpainting Using Multi-Scale Neural Patch Synthesis (CVPR 2017 · similarity 0.89)
- Rethinking Image Inpainting via a Mutual Encoder-Decoder with Feature Equalizations (ECCV 2020 · similarity 0.88)
- Contextual-based Image Inpainting: Infer, Match, and Translate (ECCV 2018 · similarity 0.88)
- Video Inpainting by Jointly Learning Temporal Structure and Spatial Details (AAAI 2019 · similarity 0.88)
- Image Inpainting via Generative Multi-column Convolutional Neural Networks (NeurIPS 2018 · similarity 0.88)
- Learning Pyramid-Context Encoder Network for High-Quality Image Inpainting (CVPR 2019 · similarity 0.88)
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.
