Research map / 3D vision / NeRF and radiance fields
UnMix-NeRF: Spectral Unmixing Meets Neural Radiance Fields
Fabian Perez, Sara Rojas, Carlos Hinojosa, Hoover Rueda-Chacón, Bernard Ghanem
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 physical constraints (softmax abundances, sigmoid scales) to unmix-only model (ablation) | UnMix-NeRF · NeSpoF | PSNR: 24.73 -> 22.36 | Hurt |
| add per-point scaling factors (ELMM) (ablation) | UnMix-NeRF · NeSpoF | PSNR: 22.36 -> 29.36 | Improved |
| add RGB loss from spectral-to-RGB projection (ablation) | UnMix-NeRF · NeSpoF | PSNR: 29.36 -> 30.92 | Improved |
| initialize endmembers with VCA instead of randomly (ablation) | UnMix-NeRF · NeSpoF | PSNR: 30.92 -> 31.09 | No clear effect |
| add specular field with tint factor (ablation) | UnMix-NeRF · NeSpoF | PSNR: 31.09 -> 33.20 | Improved |
| use UnMix-NeRF instead of HyperGS on BaySpec Caladium | UnMix-NeRF · BaySpec Caladium | PSNR: 27.70 -> 30.08 | Improved |
| use UnMix-NeRF instead of NeSpoF for training cost | UnMix-NeRF · NeSpoF (synthetic, 21 bands) | training time: 11.9 hours -> 44 min | Improved |
| use UnMix-NeRF instead of HyperGS on Surface Optics Rosemary | UnMix-NeRF · Surface Optics Rosemary | SSIM: 0.9845 -> 0.9355 | Hurt |
| use unmixing-based unsupervised material segmentation via endmember cluster probe | UnMix-NeRF · NeSpoF material segmentation (extended) | F1 / mIoU: F1 0.41, mIoU 0.28 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HyperGS — HyperGS: Hyperspectral 3D Gaussian Splatting (CVPR 2025) · better on 6 of 7 shared results
- MipNeRF360 — Mip-NeRF 360: Unbounded Anti-Aliased Neural Radiance Fields (CVPR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Ref-NeRF: Structured View-Dependent Appearance for Neural Radiance Fields (CVPR 2022 · similarity 0.89)
- NeRF-HuGS: Improved Neural Radiance Fields in Non-static Scenes Using Heuristics-Guided Segmentation (CVPR 2024 · similarity 0.88)
- NeRF-Texture: Synthesizing Neural Radiance Field Textures (TPAMI 2024 · similarity 0.87)
- SpectralNeRF: Physically Based Spectral Rendering with Neural Radiance Field (AAAI 2024 · similarity 0.87)
- NeRFPrior: Learning Neural Radiance Field as a Prior for Indoor Scene Reconstruction (CVPR 2025 · similarity 0.86)
- I2-NeRF: Learning Neural Radiance Fields Under Physically-Grounded Media Interactions (NeurIPS 2025 · similarity 0.86)
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.
