Public

CDI — 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
support -3.0 – 6.0 1.5 pixels
saturation -5.0 – 10.0 2.5 %
missing_center -3.0 – 6.0 1.5 pixels

InverseNet Baseline Scores

Method: CPU_baseline — Mismatch parameter: nominal

Scenario I (Ideal)

13.90 dB

SSIM 0.2859

Scenario II (Mismatch)

4.25 dB

SSIM 0.0166

Scenario III (Oracle)

3.54 dB

SSIM 0.0335

Per-scene breakdown (4 scenes)
Scene PSNR I SSIM I PSNR II SSIM II PSNR III SSIM III
scene_00 8.88 0.1320 1.87 -0.0194 1.14 0.0155
scene_01 14.52 0.1742 6.24 0.0165 5.11 0.0467
scene_02 11.72 0.2823 3.97 0.0271 3.20 0.0308
scene_03 20.47 0.5553 4.94 0.0423 4.72 0.0411

Public Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 DiffusionPhase + gradient 0.814 34.46 0.964 0.92 ✓ Certified Song et al., NeurIPS 2024
2 ScorePhase + gradient 0.792 33.06 0.953 0.9 ✓ Certified Wei et al., ECCV 2025
3 HolographyViT + gradient 0.783 32.44 0.947 0.9 ✓ Certified Wang et al., ICCV 2024
4 AutoPhase++ + gradient 0.783 32.73 0.95 0.88 ✓ Certified Rivenson et al., ECCV 2024
5 PhaseFormer + gradient 0.775 31.84 0.941 0.9 ✓ Certified Tian et al., ICCV 2024
6 CyclePhase + gradient 0.771 31.03 0.931 0.94 ✓ Certified Ge et al., IEEE Photonics 2023
7 PhaseResNet + gradient 0.753 30.42 0.923 0.9 ✓ Certified Baoqing et al., Optica 2023
8 LRGS + gradient 0.752 30.59 0.925 0.88 ✓ Certified Choi et al., 2023
9 PhaseNet + gradient 0.727 29.18 0.903 0.87 ✓ Certified Rivenson et al., LSA 2018
10 deep-PR + gradient 0.672 25.59 0.82 0.94 ✓ Certified Asif et al., ICCP 2017
11 prDeep + gradient 0.651 25.17 0.808 0.88 ✓ Certified Metzler et al., ICML 2018
12 GS/HIO + gradient 0.546 20.84 0.638 0.91 ✓ Certified Fienup, Appl. Opt. 1982
13 Error Reduction + gradient 0.536 20.77 0.635 0.87 ✓ Certified Fienup, J. Opt. Soc. Am. 1982
14 Gerchberg-Saxton + gradient 0.535 20.4 0.618 0.92 ✓ Certified Gerchberg & Saxton, Optik 1972

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 CDI