Research map / Image generation / Image restoration
Intrinsic Image Transformation via Scale Space Decomposition
Lechao Cheng, Chengyi Zhang, Zicheng Liao
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 self-augmented training data (ablation) | Intrinsic Image Transformation · MPI-Sintel image split | si-MSE avg: 0.63 -> 0.59 | Improved |
| add self-augmented training data (ablation) | Intrinsic Image Transformation · MPI-Sintel scene split | si-MSE avg: 1.38 -> 1.35 | Improved |
| add web-searched augmentation data (ablation) | Intrinsic Image Transformation · MIT Intrinsic Images | LMSE Total: 0.0141 -> 0.0133 | Improved |
| use highly dataset-matched augmentation data (ablation) | Intrinsic Image Transformation · MIT Intrinsic Images | LMSE Total: 0.0133 -> 0.0121 | Improved |
| use parallel multi-branch architecture instead of sequential architecture (ablation) | Intrinsic Image Transformation · MPI-Sintel scene split | DSSIM avg: 11.07 -> 8.86 | Improved |
| use joint optimization instead of hierarchical optimization (ablation) | Intrinsic Image Transformation · MPI-Sintel scene split | si-MSE avg: 1.55 -> 1.38 | Improved |
| add the multi-band pyramid structure (ablation) | Intrinsic Image Transformation · MPI-Sintel image split | si-MSE avg: 1.15 -> 0.63 | Improved |
| use the proposed loss instead of MSE loss (ablation) | Intrinsic Image Transformation · MPI-Sintel scene split | DSSIM avg: 10.23 -> 8.86 | Improved |
| use Gaussian pyramid image components rather than CNN features as input (ablation) | Intrinsic Image Transformation · MPI-Sintel image split | DSSIM avg: 6.80 vs 6.47 | No clear effect |
Closest papers by meaning
- Self-Supervised Intrinsic Image Decomposition (NeurIPS 2017 · similarity 0.91)
- A Survey on Intrinsic Images: Delving Deep into Lambert and Beyond (IJCV 2022 · similarity 0.91)
- Revisiting Deep Intrinsic Image Decompositions (CVPR 2018 · similarity 0.91)
- ShadingNet: Image Intrinsics by Fine-Grained Shading Decomposition (IJCV 2021 · similarity 0.90)
- Unsupervised Intrinsic Image Decomposition Using Internal Self-Similarity Cues (TPAMI 2021 · similarity 0.90)
- Learning Non-Lambertian Object Intrinsics Across ShapeNet Categories (CVPR 2017 · similarity 0.89)
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.
