Public
CACTI — Public Tier
(6 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 |
|---|---|---|---|
| mask_dx | 0.2 – 0.8 | 0.5 | px |
| mask_dy | 0.1 – 0.5 | 0.3 | px |
| mask_rotation | -0.05 – 0.25 | 0.1 | deg |
| mask_blur | -0.25 – 0.25 | 0.0 | px |
| clock_offset | -0.05 – 0.15 | 0.05 | frames |
| gain_drift | 0.97 – 1.07 | 1.02 | |
| offset_drift | -0.018 – 0.022 | 0.002 |
InverseNet Baseline Scores
Method: CPU_baseline — Mismatch parameter: nominal
Scenario I (Ideal)
29.06 dB
SSIM 0.8718
Scenario II (Mismatch)
22.85 dB
SSIM 0.7287
Scenario III (Oracle)
28.80 dB
SSIM 0.8676
Per-scene breakdown (4 scenes)
| Scene | PSNR I | SSIM I | PSNR II | SSIM II | PSNR III | SSIM III |
|---|---|---|---|---|---|---|
| scene_00 | 28.84 | 0.8942 | 19.63 | 0.6299 | 28.66 | 0.8884 |
| scene_01 | 20.88 | 0.7034 | 16.78 | 0.4869 | 20.84 | 0.7000 |
| scene_02 | 30.69 | 0.9155 | 24.83 | 0.8474 | 30.01 | 0.9085 |
| scene_03 | 35.81 | 0.9742 | 30.15 | 0.9504 | 35.68 | 0.9735 |
Public Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | ELP-Unfolding + blind cal | 0.572 | 23.06 | 0.672 | 0.977 | ✓ Certified | InverseNet Scenario IV (blind cal, improved hybrid grid search) |
| 2 | EfficientSCI + blind cal | 0.567 | 22.65 | 0.676 | 0.974 | ✓ Certified | InverseNet Scenario IV (blind cal, improved hybrid grid search) |
| 3 | HiSViT-9 + blind cal | 0.565 | 22.58 | 0.673 | 0.974 | ✓ Certified | InverseNet Scenario IV (blind cal, improved hybrid grid search) |
| 4 | GAP-TV + blind cal | 0.522 | 21.79 | 0.59 | 0.977 | ✓ Certified | InverseNet Scenario IV (blind cal, improved hybrid grid search) |
| 5 | PnP-FFDNet + blind cal | 0.446 | 17.37 | 0.542 | 0.987 | ✓ Certified | InverseNet Scenario IV (blind cal, improved hybrid grid search) |
Visible Data Fields
y
H_ideal
spec_ranges
x_true
true_spec
Dataset
Format: HDF5
Scenes: 6
Scoring Formula
0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)
PSNR: 40%
SSIM: 40%
Consistency: 20%