Public
Polarimetric SAR (PolSAR) — Public Tier
(3 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 |
|---|---|---|---|
| cross_talk_between_polarizations | -10.0 – 5.0 | -2.5 | dB |
| channel_imbalance | -0.2 – 0.4 | 0.1 | dB |
| faraday_rotation | -1.0 – 2.0 | 0.5 | deg |
InverseNet Baseline Scores
Method: CPU_baseline — Mismatch parameter: nominal
Scenario I (Ideal)
29.33 dB
SSIM 0.8672
Scenario II (Mismatch)
28.96 dB
SSIM 0.8578
Scenario III (Oracle)
31.97 dB
SSIM 0.9007
Per-scene breakdown (4 scenes)
| Scene | PSNR I | SSIM I | PSNR II | SSIM II | PSNR III | SSIM III |
|---|---|---|---|---|---|---|
| scene_00 | 30.83 | 0.8710 | 29.32 | 0.8547 | 32.36 | 0.9000 |
| scene_01 | 29.03 | 0.8684 | 28.59 | 0.8712 | 31.59 | 0.9076 |
| scene_02 | 29.23 | 0.8516 | 28.33 | 0.8571 | 31.37 | 0.8998 |
| scene_03 | 28.23 | 0.8777 | 29.58 | 0.8480 | 32.57 | 0.8956 |
Public Tier Leaderboard
| # | Method | Score | PSNR | SSIM | Consistency | Trust | Source |
|---|---|---|---|---|---|---|---|
| 1 | SARFormer + gradient | 0.792 | 32.59 | 0.949 | 0.93 | ✓ Certified | Li et al., CVPR 2024 |
| 2 | DiffusionSAR + gradient | 0.788 | 32.8 | 0.951 | 0.9 | ✓ Certified | Wei et al., NeurIPS 2024 |
| 3 | PanSharpener++ + gradient | 0.776 | 31.9 | 0.942 | 0.9 | ✓ Certified | Zhang et al., ICCV 2024 |
| 4 | ScoreSAR + gradient | 0.773 | 31.43 | 0.936 | 0.92 | ✓ Certified | Johnson et al., ECCV 2025 |
| 5 | SAR-ResNet + gradient | 0.746 | 29.69 | 0.912 | 0.92 | ✓ Certified | Chen et al., IEEE TGRS 2022 |
| 6 | SARDenoiserViT + gradient | 0.744 | 29.94 | 0.916 | 0.89 | ✓ Certified | Wang et al., ICCV 2024 |
| 7 | SAR-CAM + gradient | 0.736 | 29.36 | 0.907 | 0.9 | ✓ Certified | Cross-attention SAR, 2024 |
| 8 | SAR-DRN + gradient | 0.715 | 28.37 | 0.888 | 0.88 | ✓ Certified | Zhang et al., RS 2018 |
| 9 | SAR-BM3D + gradient | 0.638 | 24.36 | 0.781 | 0.91 | ✓ Certified | Parrilli et al., IEEE TGRS 2012 |
| 10 | Lee Filter + gradient | 0.636 | 24.04 | 0.77 | 0.94 | ✓ Certified | Lee, IEEE TGRS 1980 |
| 11 | Range-Doppler + gradient | 0.621 | 23.97 | 0.767 | 0.87 | ✓ Certified | SAR signal processing baseline |
| 12 | Chirp Scaling + gradient | 0.582 | 22.21 | 0.699 | 0.9 | ✓ Certified | Raney et al., IEEE TGRS 1994 |
| 13 | Matched Filter + gradient | 0.580 | 21.86 | 0.684 | 0.94 | ✓ Certified | Standard SAR focusing |
Visible Data Fields
y
H_ideal
spec_ranges
x_true
true_spec
Dataset
Format: HDF5
Scenes: 3
Scoring Formula
0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)
PSNR: 40%
SSIM: 40%
Consistency: 20%