Research map / 3D vision / Depth and stereo
From RGB to Spectrum for Natural Scenes via Manifold-Based Mapping
Yan Jia, Yinqiang Zheng, Lin Gu, Art Subpa-Asa, Antony Lam, Yoichi Sato, Imari Sato
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 |
|---|---|---|---|
| train on 60 images instead of 100 images (ablation) | manifold-based RGB-to-spectrum mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 4.18% | Mixed |
| raise embedding dimensionality from 3 to 4 (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 3.94% | Hurt |
| raise embedding dimensionality from 3 to 5 (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 4.01% | Hurt |
| swap camera response (Canon5D Mark II to Nikon D90) at same train/test camera (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 4.42% | Hurt |
| use Olympus EPL2 response instead of Canon 5D Mark II (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 3.94% | Hurt |
| train with one camera, test with another from same manufacturer (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 7.87% | Hurt |
| train with one camera, test with another from a different manufacturer (ablation) | manifold-based mapping · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 3.60% -> 13.9% | Hurt |
| replace direct RGB-to-spectrum RBF/sparse-dictionary mapping with manifold-based 3D embedding mapping | RGB to spectrum reconstruction · Arad & Ben-Shahar hyperspectral DB | NRMSD (%): 6.54 -> 3.60 | Improved |
| replace direct mapping with manifold-based mapping | RGB to spectrum reconstruction · Harvard outdoor DB | NRMSD (%): 19.30 -> 6.77 | Improved |
| replace direct mapping with manifold-based mapping | RGB to spectrum reconstruction · CAVE indoor DB | NRMSD (%): 16.17 -> 13.41 | Improved |
Closest papers by meaning
- Semantic-Embedded Unsupervised Spectral Reconstruction From Single RGB Images in the Wild (ICCV 2021 · similarity 0.88)
- Deeply Learned Filter Response Functions for Hyperspectral Reconstruction (CVPR 2018 · similarity 0.86)
- EigenSR: Eigenimage-Bridged Pre-Trained RGB Learners for Single Hyperspectral Image Super-Resolution (AAAI 2025 · similarity 0.84)
- FRN: Fractal-Based Recursive Spectral Reconstruction Network (NeurIPS 2025 · similarity 0.84)
- S2MAE: A Spatial-Spectral Pretraining Foundation Model for Spectral Remote Sensing Data (CVPR 2024 · similarity 0.83)
- Hyperspectral Image Reconstruction via Combinatorial Embedding of Cross-Channel Spatio-Spectral Clues (AAAI 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.
