Research map / Image generation / Image restoration
From Shadow Generation To Shadow Removal
Zhihao Liu, Hui Yin, Xinyi Wu, Zhenyao Wu, Yang Mi, Song Wang
Check 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 |
|---|---|---|---|
| Replace patch-based self-supervision with generated paired shadow/non-shadow regions | G2R-ShadowNet vs. Le and Samaras · ISTD test, all | RMSE: 4.0 -> 3.9 | Improved |
| Add paired shadow-free training data and skip shadow generation | G2R-ShadowNet · ISTD test, all | SSIM: 0.944 -> 0.957 | Improved |
| Add refinement subnet (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 12.3 -> 8.9 | Improved |
| Provide the shadow mask to the refinement subnet (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 12.5 -> 8.9 | Improved |
| Train the three subnetworks jointly with connected gradient flow (ablation) | G2R-ShadowNet · ISTD test, shadow region | PSNR: 33.43 -> 33.58 | Mixed |
| Connect generator and removal subnet gradients (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 9.0 -> 8.8 | Improved |
| Add adversarial generation loss L_GAN (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 42.1 -> 8.9 | Improved |
| Add identity loss L_iden (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 43.1 -> 8.9 | Improved |
| Add shadow-removal loss L_rem (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 9.8 -> 8.9 | Improved |
| Add whole-image loss L_full (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 12.3 -> 8.9 | Improved |
| Add area loss L_area (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 9.3 -> 8.9 | Improved |
| Set area-loss dilation kernel tau to 50 rather than 0, 5, 15, or 100 (ablation) | G2R-ShadowNet · ISTD test, shadow region | RMSE: 9.5 -> 8.9; 9.5 -> 8.9; 9.1 -> 8.9; 9.4 -> 8.9 | Improved |
Closest papers by meaning
- Physics-Based Shadow Image Decomposition for Shadow Removal (TPAMI 2021 · similarity 0.91)
- Leveraging Inpainting for Single-Image Shadow Removal (ICCV 2023 · similarity 0.90)
- From Shadow Segmentation to Shadow Removal (ECCV 2020 · similarity 0.90)
- OmniSR: Shadow Removal Under Direct and Indirect Lighting (AAAI 2025 · similarity 0.88)
- When Shadow Removal Meets Intrinsic Image Decomposition: A Joint Learning Framework Using Unpaired Data (AAAI 2025 · similarity 0.87)
- Polarization Guided Mask-Free Shadow Removal (AAAI 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.
