Hidden

Polarization — Hidden Tier

(5 scenes)

Fully blind server-side evaluation — no data download.

What you get

No data downloadable. Algorithm runs server-side on hidden measurements.

How to use

Package algorithm as Docker container / Python script. Submit via link.

What to submit

Containerized algorithm accepting y + H, outputting x_hat + corrected spec.

Parameter Specifications

🔒

True spec hidden — blind evaluation, only ranges available.

Parameter Spec Range Unit
extinction_ratio -0.35 – 1.15 dB
retardance -1.4 – 4.6 nm
alignment -0.35 – 1.15 deg

Hidden Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 ScoreMicro + gradient 0.717 29.03 0.901 0.83 ✓ Certified Wei et al., ECCV 2025
2 Restormer+ + gradient 0.714 28.78 0.896 0.84 ✓ Certified Zamir et al., ICCV 2024
3 DeconvFormer + gradient 0.701 29.03 0.901 0.75 ✓ Certified Chen et al., CVPR 2024
4 ResUNet + gradient 0.698 27.56 0.871 0.87 ✓ Certified DeCelle et al., Nat. Methods 2021
5 PnP-FISTA + gradient 0.666 26.46 0.844 0.82 ✓ Certified Bai et al., 2020
6 DiffDeconv + gradient 0.657 26.3 0.84 0.79 ✓ Certified Huang et al., NeurIPS 2024
7 TV-Deconvolution + gradient 0.642 25.59 0.82 0.79 ✓ Certified Rudin et al., Phys. A 1992
8 Wiener Filter + gradient 0.637 25.38 0.814 0.79 ✓ Certified Analytical baseline
9 Restormer + gradient 0.635 24.74 0.794 0.85 ✓ Certified Zamir et al., CVPR 2022
10 CARE + gradient 0.603 23.56 0.753 0.83 ✓ Certified Weigert et al., Nat. Methods 2018
11 PnP-DnCNN + gradient 0.602 23.29 0.742 0.86 ✓ Certified Zhang et al., IEEE TIP 2017
12 U-Net + gradient 0.594 23.6 0.754 0.78 ✓ Certified Ronneberger et al., MICCAI 2015
13 Richardson-Lucy + gradient 0.570 22.31 0.703 0.83 ✓ Certified Richardson, JOSA 1972 / Lucy, AJ 1974

Dataset

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 Polarization