Research map / Video / Video segmentation and tracking
Unveiling Deep Shadows: A Survey and Benchmark on Image and Video Shadow Detection, Removal, and Generation in the Deep Learning Era
Xiaowei Hu, Zhenghao Xing, Tianyu Wang, Chi‐Wing Fu, Pheng‐Ann Heng
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 |
|---|---|---|---|
| increase input size from 256 to 512 | SDDNet · SBU-Refine | BER: 5.39 -> 4.86 | Improved |
| increase input size from 256 to 512 | DSC · SBU-Refine | BER: 6.79 -> 6.34 | Improved |
| increase input size from 256 to 512 | ECA · SBU-Refine | BER: 7.08 -> 7.52 | Hurt |
| test on SRD after training on SBU-Refine instead of in-domain | SDDNet · SRD cross vs SBU-Refine (512) | BER: 4.86 -> 7.65 | Hurt |
| test on SRD after training on SBU-Refine instead of in-domain | ECA · SRD cross vs SBU-Refine (512) | BER: 7.52 -> 12.71 | Hurt |
| evaluate on CUHK-Shadow instead of SBU-Refine | DSDNet · 512x512 | BER: 5.04 -> 7.79 | Hurt |
| use extra supervision from DSC and BDRAR | DSDNet · SBU-Refine 256 | BER: 6.79 -> 5.37 | Improved |
| use SSISv2 instead of SSIS | SSIS · SOBA-testing | SOAP segm: 29.9 -> 35.3 | Improved |
| use SSISv2 instead of SSIS | SSIS · SOBA-challenge | SOAP segm: 12.8 -> 17.7 | Improved |
| use SSISv2 instead of SSIS | SSIS · SOBA-testing | inference speed (images/s): 5.83 -> 5.17 | Hurt |
| evaluate on SOBA-VID cross-dataset instead of SOBA-testing | SSISv2 · SOAP segm | SOAP segm: 35.3 -> 37.0 | No clear effect |
| use SCOTCH and SODA instead of ShadowSAM | ShadowSAM · ViSha | IoU: 61.72 vs 61.24 | Mixed |
Closest papers by meaning
- Cast and Attached Shadow Detection via Iterative Light and Geometry Reasoning (ECCV 2026 · similarity 0.91)
- Physics-Based Shadow Image Decomposition for Shadow Removal (TPAMI 2021 · similarity 0.90)
- OmniSR: Shadow Removal Under Direct and Indirect Lighting (AAAI 2025 · similarity 0.89)
- DeS3: Adaptive Attention-Driven Self and Soft Shadow Removal Using ViT Similarity (AAAI 2024 · similarity 0.88)
- MetaShadow: Object-Centered Shadow Detection, Removal, and Synthesis (CVPR 2025 · similarity 0.88)
- Detail-Preserving Latent Diffusion for Stable Shadow Removal (CVPR 2025 · 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.
