Research map / Image generation / Image restoration
ShadowHack: Hacking Shadows via Luminance-Color Divide and Conquer
Jin Hu, Mingjia Li, Xiaojie Guo
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 |
|---|---|---|---|
| luminance-color decoupling instead of RGB end-to-end (ablation) | ShadowHack · ISTD+ (decoupling ablation) | PSNR all: 36.16 -> 36.31 | Improved |
| luminance-color decoupling instead of RGB end-to-end (ablation) | ShadowHack · ISTD+ (decoupling ablation) | RMSE: 2.53 -> 2.48 | Improved |
| luminance-color decoupling instead of YCbCr end-to-end (ablation) | ShadowHack · ISTD+ (decoupling ablation) | PSNR all: 35.62 -> 36.31 | Improved |
| YCbCr decoupling instead of Retinex decomposition (ablation) | ShadowHack · ISTD+ (decoupling ablation) | PSNR all: 35.85 -> 36.31 | Improved |
| YCbCr decoupling instead of Retinex decomposition (ablation) | ShadowHack · ISTD+ (decoupling ablation) | RMSE: 2.63 -> 2.48 | Improved |
| add color features to Q,K in attention (ablation) | LRNet · SRD (rectified attention ablation, luminance) | PSNR all: 36.60 -> 36.76 | Improved |
| add rectification (subtract Att2) on [Ft;Fc] (ablation) | LRNet · SRD (rectified attention ablation, luminance) | PSNR all: 36.76 -> 36.87 | Improved |
| rectify with [Fc] instead of [Ft;Fc] (ablation) | LRNet · SRD (rectified attention ablation, luminance) | PSNR all: 36.87 -> 36.90 | Improved |
| rectify luminance-only [Ft]&[Ft] instead of [Ft] (ablation) | LRNet · SRD (rectified attention ablation, luminance) | PSNR all: 36.60 -> 36.58 | No clear effect |
| cross-attention color injection instead of concatenation (ablation) | CRNet · SRD (CRNet ablation) | RMSE all: 3.39 -> 2.90 | Improved |
| add multiple checkpoint training (ablation) | CRNet · SRD (CRNet ablation) | RMSE all: 3.28 -> 2.90 | Improved |
| ShadowHack instead of RASM | RASM · ISTD+ | PSNR all: 36.16 -> 36.31 | Improved |
Closest papers by meaning
- Recasting Regional Lighting for Shadow Removal (AAAI 2024 · similarity 0.87)
- Towards Ghost-Free Shadow Removal via Dual Hierarchical Aggregation Network and Shadow Matting GAN (AAAI 2020 · similarity 0.87)
- Bijective Mapping Network for Shadow Removal (CVPR 2022 · similarity 0.86)
- OmniSR: Shadow Removal Under Direct and Indirect Lighting (AAAI 2025 · similarity 0.84)
- When Shadow Removal Meets Intrinsic Image Decomposition: A Joint Learning Framework Using Unpaired Data (AAAI 2025 · similarity 0.83)
- StyLitGAN: Image-Based Relighting via Latent Control (CVPR 2024 · similarity 0.83)
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.
