Eddy Current Imaging

Eddy Current Imaging

Join the Competition Contribute to Benchmark

Standard reconstruction benchmark — forward model perfectly known, no calibration needed. Score = 0.5 × clip((PSNR−15)/30, 0, 1) + 0.5 × SSIM

# Method Score PSNR (dB) SSIM Source
🥇 DiffEC 0.882 39.3 0.955 ✓ Certified Gao et al. 2024
🥈 PhysEC 0.855 38.0 0.944 ✓ Certified Chen et al. 2024
🥉 SwinEC 0.832 36.9 0.934 ✓ Certified Wang et al. 2023
4 TransEC 0.799 35.4 0.918 ✓ Certified Li et al. 2022
5 ECNN-Defect 0.738 32.9 0.880 ✓ Certified Zhang et al. 2021
6 DnCNN-EC 0.672 30.1 0.840 ✓ Certified Gao et al. 2019
7 MUSIC-EC 0.600 27.3 0.789 ✓ Certified Skarlatos et al. 2012
8 TV-EC 0.537 24.8 0.748 ✓ Certified Sabbagh et al. 2010
9 EC-Deconv 0.471 22.1 0.705 ✓ Certified Bowler 1994

Dataset: PWM Benchmark (9 algorithms)

Blind Reconstruction Challenge — forward model has unknown mismatch, must calibrate from data. Score = 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

# Method Overall Score Public
PSNR / SSIM
 Dev
PSNR / SSIM
 Hidden
PSNR / SSIM
 Trust Source
🥇 DiffEC + gradient 0.783
0.857
37.67 dB / 0.981
0.762
32.05 dB / 0.943
0.731
30.28 dB / 0.921
✓ Certified Gao et al., NeurIPS 2024
🥈 PhysEC + gradient 0.760
0.823
35.93 dB / 0.973
0.747
30.49 dB / 0.924
0.711
28.62 dB / 0.893
✓ Certified Chen et al., IEEE Trans. Magn. 2024
🥉 SwinEC + gradient 0.758
0.809
34.5 dB / 0.964
0.757
31.82 dB / 0.941
0.707
28.45 dB / 0.890
✓ Certified Wang et al., NDT&E Int. 2023
4 TransEC + gradient 0.712
0.812
34.13 dB / 0.962
0.701
27.83 dB / 0.877
0.623
24.15 dB / 0.774
✓ Certified Li et al., IEEE Trans. Ind. Electron. 2022
5 ECNN-Defect + gradient 0.685
0.755
30.76 dB / 0.928
0.677
26.81 dB / 0.853
0.624
24.03 dB / 0.770
✓ Certified Zhang et al., NDT&E Int. 2021
6 MUSIC-EC + gradient 0.636
0.652
25.43 dB / 0.815
0.644
25.12 dB / 0.806
0.612
24.03 dB / 0.770
✓ Certified Skarlatos et al., NDT&E Int. 2012
7 DnCNN-EC + gradient 0.587
0.701
27.29 dB / 0.865
0.551
21.44 dB / 0.666
0.509
20.33 dB / 0.614
✓ Certified Gao et al., IEEE Sens. J. 2019
8 EC-Deconv + gradient 0.454
0.511
19.86 dB / 0.592
0.469
18.82 dB / 0.541
0.382
16.44 dB / 0.423
✓ Certified Bowler, J. Appl. Phys. 1994
9 TV-EC + gradient 0.429
0.617
23.25 dB / 0.741
0.367
15.65 dB / 0.385
0.304
12.96 dB / 0.267
✓ Certified Sabbagh et al., IEEE Trans. Magn. 2010

Complete score requires all 3 tiers (Public + Dev + Hidden).

Join the competition →
Scoring: 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖) PSNR 40% · SSIM 40% · Consistency 20%
Public 3 scenes

Full-access development tier with all data visible.

What you get & how to use

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.

Public Leaderboard
# Method Score PSNR SSIM
1 DiffEC + gradient 0.857 37.67 0.981
2 PhysEC + gradient 0.823 35.93 0.973
3 TransEC + gradient 0.812 34.13 0.962
4 SwinEC + gradient 0.809 34.5 0.964
5 ECNN-Defect + gradient 0.755 30.76 0.928
6 DnCNN-EC + gradient 0.701 27.29 0.865
7 MUSIC-EC + gradient 0.652 25.43 0.815
8 TV-EC + gradient 0.617 23.25 0.741
9 EC-Deconv + gradient 0.511 19.86 0.592
Spec Ranges (4 parameters)
Parameter Min Max Unit
liftoff_distance -0.2 0.4 mm
conductivity_error 57.4 59.2 MS/m
excitation_frequency_drift 99.0 102.0 kHz
probe_tilt -0.4 0.8 deg
View Details →
Dev 3 scenes

Blind evaluation tier — no ground truth available.

What you get & how to use

What you get: Measurements (y), ideal forward operator (H), and spec ranges only.

How to use: Apply your pipeline from the Public tier. Use consistency as self-check.

What to submit: Reconstructed signals and corrected spec. Scored server-side.

Dev Leaderboard
# Method Score PSNR SSIM
1 DiffEC + gradient 0.762 32.05 0.943
2 SwinEC + gradient 0.757 31.82 0.941
3 PhysEC + gradient 0.747 30.49 0.924
4 TransEC + gradient 0.701 27.83 0.877
5 ECNN-Defect + gradient 0.677 26.81 0.853
6 MUSIC-EC + gradient 0.644 25.12 0.806
7 DnCNN-EC + gradient 0.551 21.44 0.666
8 EC-Deconv + gradient 0.469 18.82 0.541
9 TV-EC + gradient 0.367 15.65 0.385
Spec Ranges (4 parameters)
Parameter Min Max Unit
liftoff_distance -0.24 0.36 mm
conductivity_error 57.28 59.08 MS/m
excitation_frequency_drift 98.8 101.8 kHz
probe_tilt -0.48 0.72 deg
Submit Reconstruction View Details →
Hidden 3 scenes

Fully blind server-side evaluation — no data download.

What you get & how to use

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.

Hidden Leaderboard
# Method Score PSNR SSIM
1 DiffEC + gradient 0.731 30.28 0.921
2 PhysEC + gradient 0.711 28.62 0.893
3 SwinEC + gradient 0.707 28.45 0.89
4 ECNN-Defect + gradient 0.624 24.03 0.77
5 TransEC + gradient 0.623 24.15 0.774
6 MUSIC-EC + gradient 0.612 24.03 0.77
7 DnCNN-EC + gradient 0.509 20.33 0.614
8 EC-Deconv + gradient 0.382 16.44 0.423
9 TV-EC + gradient 0.304 12.96 0.267
Spec Ranges (4 parameters)
Parameter Min Max Unit
liftoff_distance -0.14 0.46 mm
conductivity_error 57.58 59.38 MS/m
excitation_frequency_drift 99.3 102.3 kHz
probe_tilt -0.28 0.92 deg
Submit Algorithm View Details →

Blind Reconstruction Challenge

Challenge

Given measurements with unknown mismatch and spec ranges (not exact params), reconstruct the original signal. A method must be evaluated on all three tiers for a complete score. Scored on a composite metric: 0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖).

Input

Measurements y, ideal forward model H, spec ranges

Output

Reconstructed signal x̂

Spec DAG — Forward Model Pipeline

F → D

F Fourier
D Detector

Mismatch Parameters

Symbol Parameter Description Nominal Perturbed
l_d liftoff_distance Liftoff distance (mm) 0.0 0.2
c_e conductivity_error Conductivity error (MS/m) 58.0 58.6
e_f excitation_frequency_drift Excitation frequency drift (kHz) 100.0 101.0
p_t probe_tilt Probe tilt (deg) 0.0 0.4

Credits System

40%
Platform Profit Pool
Revenue allocated to benchmark rewards
30%
Winner Share
Top algorithm receives from pool
$100
Min Withdrawal
Minimum payout threshold
Spec Primitives Reference (11 primitives)
P Propagation

Free-space or medium propagation kernel (Fresnel, Rayleigh-Sommerfeld).

M Mask / Modulation

Spatial or spatio-temporal amplitude modulation (coded aperture, SLM pattern).

Π Projection

Geometric projection operator (Radon transform, fan-beam, cone-beam).

F Fourier Sampling

Sampling in the Fourier / k-space domain (MRI, ptychography).

C Convolution

Shift-invariant convolution with a point-spread function (PSF).

Σ Summation / Integration

Summation along a physical dimension (spectral, temporal, angular).

D Detector

Sensor readout with gain g and noise model η (Gaussian, Poisson, mixed).

S Structured Illumination

Patterned illumination (block, Hadamard, random) applied to the scene.

W Wavelength Dispersion

Spectral dispersion element (prism, grating) with shift α and aperture a.

R Rotation / Motion

Sample or gantry rotation (CT, electron tomography).

Λ Wavelength Selection

Spectral filter or monochromator selecting a wavelength band.