Research map / 3D vision / Depth and stereo
Tuning IR-Cut Filter for Illumination-Aware Spectral Reconstruction From RGB
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Learn an IR-cut filter response instead of selecting from existing camera responses | RGB-to-spectrum reconstruction · ICVL | RMSE: 3.56 -> 3.33 | Helped |
| Learn an IR-cut filter response rather than design three independent spectral responses | RGB-to-spectrum reconstruction · ICVL | RMSE: 3.33 -> 3.17 | Hurt |
| Learn an IR-cut filter response instead of selecting from existing camera responses | RGB-to-spectrum reconstruction · TokyoTech-TC | RMSE: 3.78 -> 3.54 | Helped |
| Use the proposed reconstruction network instead of JCS | RGB-to-spectrum reconstruction with Point Grey Grasshopper2 14S5C CSS · TokyoTech | RMSE: 4.32 -> 3.27 | Helped |
| Use the proposed reconstruction network instead of JCS | RGB-to-spectrum reconstruction with Point Grey Grasshopper2 14S5C CSS · TokyoTech-TC | RMSE: 3.78 -> 3.49 | Helped |
| Use the proposed reconstruction network instead of JCS | RGB-to-spectrum reconstruction with Point Grey Grasshopper2 14S5C CSS · ICVL | RMSE: 3.56 -> 3.16 | Helped |
| Enable the illumination prediction branch (ablation) | Proposed RGB-to-spectrum network · TokyoTech-TC | RMSE: 3.49 -> 3.54 | Hurt |
| Enable the illumination prediction branch (ablation) | Proposed RGB-to-spectrum network · ICVL | RMSE: 3.16 -> 3.28 | Hurt |
| Allow nonzero IR-cut transmittance for 730~770nm instead of forcing it to zero (ablation) | Learned IR-cut filter for visible-spectrum reconstruction · 420~720nm visible spectral reconstruction with near-infrared input | RMSE | Helped |
| Predict illumination spectra instead of using the white-world assumption | Illumination estimation · TokyoTech-TC, truth CCT 4500K | illumination MSE: 0.2242 -> 0.0732 | Helped |
Closest papers by meaning
- Deeply Learned Filter Response Functions for Hyperspectral Reconstruction (CVPR 2018 · similarity 0.90)
- FRN: Fractal-Based Recursive Spectral Reconstruction Network (NeurIPS 2025 · similarity 0.85)
- Spectral Compressive Imaging via Chromaticity-Intensity Decomposition (NeurIPS 2025 · similarity 0.85)
- SpectralNeRF: Physically Based Spectral Rendering with Neural Radiance Field (AAAI 2024 · similarity 0.84)
- GCC: Generative Color Constancy via Diffusing a Color Checker (CVPR 2025 · similarity 0.84)
- Bright-NeRF: Brightening Neural Radiance Field with Color Restoration from Low-Light RAW Images (AAAI 2025 · 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.