| G2R-ShadowNet (CVPR 2021) (256x256) | 0.46 | 21.72 | 6.08 | 0.619 | From Shadow Generation to Shadow Removal | |
| G2R-ShadowNet (CVPR 2021) (512x512) | 0.302 | 22.44 | 5.72 | 0.765 | From Shadow Generation to Shadow Removal | |
| SP+M-Net (ICCV 2019) (256x256) | 0.444 | 22.25 | 5.68 | 0.636 | Shadow Removal via Shadow Image Decomposition | |
| Auto (CVPR 2021) (256x256) | 0.37 | 23.2 | 5.37 | 0.694 | Auto-Exposure Fusion for Single-Image Shadow Removal | |
| ShadowDiffusion (CVPR 2023) (512x512) | 0.24 | 23.09 | 5.11 | 0.804 | ShadowDiffusion: When Degradation Prior Meets Diffusion Model for Shadow
Removal | |
| ShadowDiffusion (CVPR 2023) (256x256) | 0.363 | 23.26 | 4.84 | 0.684 | ShadowDiffusion: When Degradation Prior Meets Diffusion Model for Shadow
Removal | |
| Auto (CVPR 2021) (512x512) | 0.247 | 24.32 | 4.71 | 0.8 | Auto-Exposure Fusion for Single-Image Shadow Removal | |
| ShadowMaskFormer (arXiv 2024) (256x256) | 0.386 | 23.85 | 4.69 | 0.671 | ShadowMaskFormer: Mask Augmented Patch Embeddings for Shadow Removal | |
| SG-ShadowNet (ECCV 2022) (256x256) | 0.443 | 24.1 | 4.6 | 0.636 | Style-Guided Shadow Removal | - |
| ShadowFormer (AAAI 2023) (256x256) | 0.348 | 24.28 | 4.44 | 0.715 | ShadowFormer: Global Context Helps Image Shadow Removal | |
| BMNet (CVPR 2022) (256x256) | 0.327 | 24.24 | 4.39 | 0.721 | Bijective Mapping Network for Shadow Removal | - |
| SP+M-Net (ICCV 2019) (512x512) | 0.269 | 24.89 | 4.35 | 0.792 | Shadow Removal via Shadow Image Decomposition | |
| Mask-ShadowGAN (ICCV 2019) (256x256) | 0.427 | 24.67 | 4.32 | 0.662 | Mask-ShadowGAN: Learning to Remove Shadows from Unpaired Data | |
| DC-ShadowNet (ICCV 2021) (256x256) | 0.383 | 24.72 | 4.27 | 0.67 | DC-ShadowNet: Single-Image Hard and Soft Shadow Removal Using
Unsupervised Domain-Classifier Guided Network | |
| HomoFormer (CVPR 2024) (256x256) | 0.325 | 24.64 | 4.17 | 0.723 | HomoFormer: Homogenized Transformer for Image Shadow Removal | - |
| ST-CGAN (CVPR 2018)
(256x256) | 0.443 | 25.08 | 4.15 | 0.637 | Stacked Conditional Generative Adversarial Networks for Jointly Learning Shadow Detection and Shadow Removal | |
| ShadowMaskFormer (arXiv 2024) (512x512) | 0.249 | 25.13 | 4.15 | 0.798 | ShadowMaskFormer: Mask Augmented Patch Embeddings for Shadow Removal | |
| SG-ShadowNet (ECCV 2022) (512x512) | 0.279 | 25.56 | 4.01 | 0.786 | Style-Guided Shadow Removal | - |
| BMNet (CVPR 2022) (512x512) | 0.225 | 25.39 | 4 | 0.82 | Bijective Mapping Network for Shadow Removal | - |
| DSC (TPAMI 2020) (256x256) | 0.412 | 25.46 | 3.97 | 0.678 | Direction-aware Spatial Context Features for Shadow Detection and Removal | |