Public

Matrix — 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
matrix_perturb -0.01 – 0.02 0.005
gain 0.97 – 1.06 1.015
sigma_y -0.02 – 0.04 0.01

InverseNet Baseline Scores

Method: CPU_baseline — Mismatch parameter: nominal

Scenario I (Ideal)

10.22 dB

SSIM 0.0874

Scenario II (Mismatch)

10.41 dB

SSIM 0.0770

Scenario III (Oracle)

13.43 dB

SSIM 0.1630

Per-scene breakdown (4 scenes)
Scene PSNR I SSIM I PSNR II SSIM II PSNR III SSIM III
scene_00 10.25 0.0651 10.58 0.0687 13.56 0.1508
scene_01 9.59 0.0595 9.85 0.0612 12.70 0.1494
scene_02 7.27 0.0311 7.49 0.0363 10.74 0.0941
scene_03 13.76 0.1937 13.71 0.1417 16.73 0.2576

Public Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 FlowHSI + gradient 0.851 37.3 0.979 0.93 ✓ Certified Huang et al., arXiv 2025
2 CSTrans + gradient 0.834 35.89 0.973 0.93 ✓ Certified Liu et al., CVPR 2024
3 HiSViT+ + gradient 0.830 35.72 0.972 0.92 ✓ Certified Tao et al., ECCV 2024
4 ScoreSCI + gradient 0.823 35.47 0.971 0.9 ✓ Certified Chen et al., NeurIPS 2024
5 DiffusionHSI + gradient 0.823 35.95 0.973 0.87 ✓ Certified Zhang et al., ICCV 2024
6 CST + gradient 0.816 34.3 0.963 0.94 ✓ Certified Liu et al., ICCV 2023
7 MST-L + gradient 0.812 34.11 0.962 0.93 ✓ Certified Cai et al., CVPR 2022
8 PromptSCI + gradient 0.812 34.64 0.965 0.9 ✓ Certified Bai et al., ICCV 2024
9 EfficientSCI + gradient 0.795 32.54 0.948 0.95 ✓ Certified Wang et al., IEEE TIP 2023
10 Restormer + gradient 0.792 33.08 0.953 0.9 ✓ Certified Zamir et al., CVPR 2022
11 PnP-FFDNet + gradient 0.692 26.94 0.857 0.9 ✓ Certified Zhang et al., IEEE TPAMI 2020
12 TVAL3 + gradient 0.687 26.78 0.853 0.89 ✓ Certified Li et al., SIAM J. Sci. Comput. 2009
13 FISTA-TV + gradient 0.664 25.5 0.818 0.91 ✓ Certified Beck & Teboulle, SIAM J. Imaging Sci. 2009
14 GAP-TV + gradient 0.664 25.23 0.809 0.94 ✓ Certified Yuan et al., IEEE TIP 2016

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%
Back to Matrix