Public
Lensless — 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 |
|---|---|---|---|
| diffuser_psf | -5.0 – 10.0 | 2.5 | % |
| sensor_distance | -0.2 – 0.4 | 0.1 | mm |
| wavelength | -5.0 – 10.0 | 2.5 | nm |
InverseNet Baseline Scores
Method: CPU_baseline — Mismatch parameter: nominal
Scenario I (Ideal)
16.56 dB
SSIM 0.5739
Scenario II (Mismatch)
15.94 dB
SSIM 0.5399
Scenario III (Oracle)
16.53 dB
SSIM 0.5293
Per-scene breakdown (4 scenes)
| Scene | PSNR I | SSIM I | PSNR II | SSIM II | PSNR III | SSIM III |
|---|---|---|---|---|---|---|
| scene_00 | 16.24 | 0.4221 | 16.24 | 0.4037 | 16.34 | 0.4001 |
| scene_01 | 19.41 | 0.7507 | 18.38 | 0.7408 | 18.93 | 0.7240 |
| scene_02 | 12.42 | 0.4280 | 11.80 | 0.4154 | 12.12 | 0.4065 |
| scene_03 | 18.17 | 0.6950 | 17.33 | 0.5999 | 18.76 | 0.5867 |
Public Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | Uformer + gradient | 0.787 | 32.28 | 0.946 | 0.93 | ✓ Certified | Wang et al., CVPR 2022 |
| 2 | FlatNet + gradient | 0.739 | 30.02 | 0.917 | 0.86 | ✓ Certified | Khan et al., IEEE TPAMI 2020 |
| 3 | PnP-ADMM + gradient | 0.652 | 25.13 | 0.806 | 0.89 | ✓ Certified | Monakhova et al., Opt. Express 2019 |
| 4 | Wiener-ADMM + gradient | 0.538 | 20.51 | 0.623 | 0.92 | ✓ Certified | Antipa et al., Optica 2018 |
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%