Research map / Image generation / Image restoration
VCNet: A Robust Approach to Blind Image Inpainting
Yi Wang, Ying-Cong Chen, Xin Tao, Jiaya Jia
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 mask prediction network (MPN) (ablation) | VCN · FFHQ | PSNR: 18.88 -> 20.94 | Improved |
| add MPN (ablation) | VCN · FFHQ | SSIM: 0.6222 -> 0.6999 | Improved |
| add bottleneck fusion between MPN and RIN (ablation) | VCN · FFHQ | PSNR: 20.06 -> 20.94 | Improved |
| add bottleneck fusion between MPN and RIN (ablation) | VCN · FFHQ | BCE: 0.560 -> 0.400 | Improved |
| add semantic consistency term (ablation) | VCN · FFHQ | PSNR: 20.56 -> 20.94 | Improved |
| add semantic consistency term (ablation) | VCN · FFHQ | SSIM: 0.6836 -> 0.6999 | Improved |
| add semantic consistency term (ablation) | VCN · FFHQ | BCE: 0.653 -> 0.4 | Improved |
| feed image without estimated contamination removed (instead of removing it, VCN-RM) (ablation) | VCN · FFHQ | SSIM: 0.7045 -> 0.6999 | Mixed |
| keep contamination in input rather than removing it (VCN vs VCN-RM) (ablation) | VCN · FFHQ | BCE: 0.462 -> 0.400 | Improved |
| use probabilistic context normalization network instead of concatenating mask (VCN vs GMC) | GMC · FFHQ-2K | PSNR: 20.06 -> 20.94 | Improved |
| use VCN instead of gated-convolution baseline | GC · FFHQ-2K | PSNR: 17.16 -> 20.94 | Improved |
| transfer VCN pretrained on Places2 to raindrop removal vs AttentiveGAN | AttentiveGAN · Raindrop removal (20 images) | PSNR: 24.99 -> 26.22 | Improved |
Closest papers by meaning
- Contextual-based Image Inpainting: Infer, Match, and Translate (ECCV 2018 · similarity 0.90)
- Self-Prior Guided Pixel Adversarial Networks for Blind Image Inpainting (TPAMI 2023 · similarity 0.89)
- BVINet: Unlocking Blind Video Inpainting with Zero Annotations (ICCV 2025 · similarity 0.88)
- Region Normalization for Image Inpainting (AAAI 2020 · similarity 0.88)
- Image Inpainting With Learnable Bidirectional Attention Maps (ICCV 2019 · similarity 0.87)
- Towards Enhanced Image Inpainting: Mitigating Unwanted Object Insertion and Preserving Color Consistency (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.
