Public
FTIR Spectroscopic Imaging — Public Tier
(5 scenes)Full-access development tier with all data visible.
What you get
Measurements (y), ideal forward operator (H), spec ranges, ground truth (x_true), and true mismatch spec.
How to use
Load HDF5 → compare reconstruction vs x_true → check consistency → iterate.
What to submit
Reconstructed signals (x_hat) and corrected spec as HDF5.
Parameter Specifications
✓
True spec visible — use these exact values for Scenario III oracle reconstruction.
| Parameter | Spec Range | True Value | Unit |
|---|---|---|---|
| wavenumber_calibration | -0.4 – 0.8 | 0.2 | cm^-1 |
| water_vapor_absorption | -0.15 – 0.15 | 0.0 | - |
| detector_nonlinearity | -1.0 – 2.0 | 0.5 | - |
| atr_crystal_ri_error | -0.2 – 0.4 | 0.1 | - |
InverseNet Baseline Scores
Method: CPU_baseline — Mismatch parameter: nominal
Scenario I (Ideal)
15.88 dB
SSIM 0.2490
Scenario II (Mismatch)
15.81 dB
SSIM 0.2102
Scenario III (Oracle)
18.87 dB
SSIM 0.3863
Per-scene breakdown (4 scenes)
| Scene | PSNR I | SSIM I | PSNR II | SSIM II | PSNR III | SSIM III |
|---|---|---|---|---|---|---|
| scene_00 | 20.06 | 0.3267 | 19.25 | 0.2176 | 21.97 | 0.3704 |
| scene_01 | 17.59 | 0.3158 | 17.58 | 0.2708 | 20.70 | 0.4916 |
| scene_02 | 12.32 | 0.1756 | 12.60 | 0.1770 | 15.84 | 0.3413 |
| scene_03 | 13.55 | 0.1778 | 13.80 | 0.1754 | 16.97 | 0.3418 |
Public Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | PINN-Spectra + gradient | 0.781 | 31.81 | 0.941 | 0.93 | ✓ Certified | Physics-informed neural network |
| 2 | Cascade-UNet + gradient | 0.776 | 31.41 | 0.936 | 0.94 | ✓ Certified | Physics-informed UNet, 2025 |
| 3 | CDAE + gradient | 0.755 | 30.3 | 0.921 | 0.92 | ✓ Certified | Zhang et al., Sensors 2024 |
| 4 | SpectraFormer + gradient | 0.751 | 30.49 | 0.924 | 0.88 | ✓ Certified | Spectroscopy transformer, 2024 |
| 5 | U-Net-Spectra + gradient | 0.718 | 28.76 | 0.896 | 0.86 | ✓ Certified | Spectral U-Net variant |
| 6 | DiffusionSpectra + gradient | 0.703 | 27.5 | 0.87 | 0.9 | ✓ Certified | Zhang et al., 2024 |
| 7 | ScoreSpectra + gradient | 0.699 | 27.58 | 0.872 | 0.87 | ✓ Certified | Wei et al., 2025 |
| 8 | PnP-DnCNN + gradient | 0.666 | 26.04 | 0.833 | 0.86 | ✓ Certified | Zhang et al., 2017 |
| 9 | SG-ALS + gradient | 0.613 | 23.26 | 0.741 | 0.92 | ✓ Certified | Savitzky-Golay + ALS baseline |
| 10 | SVD + gradient | 0.596 | 22.97 | 0.73 | 0.87 | ✓ Certified | Singular Value Decomposition |
| 11 | Baseline Correction + gradient | 0.591 | 22.64 | 0.717 | 0.89 | ✓ Certified | Polynomial fitting baseline |
Visible Data Fields
y
H_ideal
spec_ranges
x_true
true_spec
Dataset
Format: HDF5
Scenes: 5
Scoring Formula
0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)
PSNR: 40%
SSIM: 40%
Consistency: 20%