Acoustic Emission Testing (AE)

Acoustic Emission Testing (AE)

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
🥇 DiffusionAE 0.817 35.5 0.950 ✓ Certified Song et al., ICLR 2021; SHM application 2024
🥈 SwinIR-AE 0.800 34.8 0.940 ✓ Certified Liang et al., ICCV 2021; AE-adapted 2024
🥉 PINN-AE 0.771 33.5 0.925 ✓ Certified Raissi et al. 2019; AE extension 2024
4 Domain-Adapted ResNet 0.736 32.0 0.905 ✓ Certified Tabian et al., Sensors 2019
5 AE-CNN 0.685 30.0 0.870 ✓ Certified Ebrahimkhanlou & Salamone, Struct. Health Monit. 2019
6 PnP-ADMM 0.608 27.5 0.800 ✓ Certified Venkatakrishnan et al., IEEE GlobalSIP 2013
7 Sparse TR (L1) 0.540 25.5 0.730 ✓ Certified Gao et al., J. Sound Vib. 2016
8 TDOA-WLS 0.432 22.0 0.630 ✓ Certified Kundu, J. Acoust. Soc. Am. 2014
9 Time-Reversal Imaging 0.382 20.5 0.580 ✓ Certified Fink, IEEE UFFC 1992; Grosse & Ohtsu 2008

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
🥇 SwinIR-AE + gradient 0.708
0.803
33.33 dB / 0.955
0.706
27.97 dB / 0.880
0.615
24.5 dB / 0.786
✓ Certified Liang et al., ICCV 2021; AE-adapted 2024
🥈 DiffusionAE + gradient 0.697
0.790
32.9 dB / 0.952
0.684
27.43 dB / 0.868
0.618
23.87 dB / 0.764
✓ Certified Song et al., ICLR 2021; SHM application 2024
🥉 PINN-AE + gradient 0.644
0.760
30.72 dB / 0.927
0.650
25.39 dB / 0.814
0.522
21.15 dB / 0.652
✓ Certified Raissi et al., J. Comput. Phys. 2019; AE extension 2024
4 Domain-Adapted ResNet + gradient 0.618
0.736
29.34 dB / 0.906
0.588
22.75 dB / 0.721
0.531
21.23 dB / 0.656
✓ Certified Tabian et al., Sensors 2019
5 AE-CNN + gradient 0.580
0.698
27.16 dB / 0.862
0.553
21.31 dB / 0.660
0.489
19.72 dB / 0.585
✓ Certified Ebrahimkhanlou & Salamone, Struct. Health Monit. 2019
6 PnP-ADMM + gradient 0.570
0.678
25.77 dB / 0.825
0.527
20.96 dB / 0.644
0.505
19.99 dB / 0.598
✓ Certified Venkatakrishnan et al., IEEE GlobalSIP 2013
7 Sparse TR (L1) + gradient 0.495
0.597
22.72 dB / 0.720
0.491
19.49 dB / 0.574
0.398
16.4 dB / 0.421
✓ Certified Gao et al., J. Sound Vib. 2016
8 TDOA-WLS + gradient 0.485
0.544
20.65 dB / 0.629
0.481
19.38 dB / 0.569
0.430
17.4 dB / 0.470
✓ Certified Kundu, J. Acoust. Soc. Am. 2014
9 Time-Reversal Imaging + gradient 0.435
0.494
18.88 dB / 0.544
0.417
17.24 dB / 0.462
0.393
16.42 dB / 0.422
✓ Certified Fink, IEEE UFFC 1992; applied to AE: Grosse & Ohtsu 2008

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 5 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 SwinIR-AE + gradient 0.803 33.33 0.955
2 DiffusionAE + gradient 0.790 32.9 0.952
3 PINN-AE + gradient 0.760 30.72 0.927
4 Domain-Adapted ResNet + gradient 0.736 29.34 0.906
5 AE-CNN + gradient 0.698 27.16 0.862
6 PnP-ADMM + gradient 0.678 25.77 0.825
7 Sparse TR (L1) + gradient 0.597 22.72 0.72
8 TDOA-WLS + gradient 0.544 20.65 0.629
9 Time-Reversal Imaging + gradient 0.494 18.88 0.544
Spec Ranges (4 parameters)
Parameter Min Max Unit
source_location_error -1.0 2.0 mm
wave_speed_error 5860.0 5980.0 m/s
sensor_coupling_gain 0.96 1.08 -
arrival_time_bias -0.1 0.2 us
View Details →
Dev 5 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 SwinIR-AE + gradient 0.706 27.97 0.88
2 DiffusionAE + gradient 0.684 27.43 0.868
3 PINN-AE + gradient 0.650 25.39 0.814
4 Domain-Adapted ResNet + gradient 0.588 22.75 0.721
5 AE-CNN + gradient 0.553 21.31 0.66
6 PnP-ADMM + gradient 0.527 20.96 0.644
7 Sparse TR (L1) + gradient 0.491 19.49 0.574
8 TDOA-WLS + gradient 0.481 19.38 0.569
9 Time-Reversal Imaging + gradient 0.417 17.24 0.462
Spec Ranges (4 parameters)
Parameter Min Max Unit
source_location_error -1.2 1.8 mm
wave_speed_error 5852.0 5972.0 m/s
sensor_coupling_gain 0.952 1.072 -
arrival_time_bias -0.12 0.18 us
Submit Reconstruction View Details →
Hidden 5 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 DiffusionAE + gradient 0.618 23.87 0.764
2 SwinIR-AE + gradient 0.615 24.5 0.786
3 Domain-Adapted ResNet + gradient 0.531 21.23 0.656
4 PINN-AE + gradient 0.522 21.15 0.652
5 PnP-ADMM + gradient 0.505 19.99 0.598
6 AE-CNN + gradient 0.489 19.72 0.585
7 TDOA-WLS + gradient 0.430 17.4 0.47
8 Sparse TR (L1) + gradient 0.398 16.4 0.421
9 Time-Reversal Imaging + gradient 0.393 16.42 0.422
Spec Ranges (4 parameters)
Parameter Min Max Unit
source_location_error -0.7 2.3 mm
wave_speed_error 5872.0 5992.0 m/s
sensor_coupling_gain 0.972 1.092 -
arrival_time_bias -0.07 0.23 us
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

P → S → D

P Propagation
S Sampling
D Detector

Mismatch Parameters

Symbol Parameter Description Nominal Perturbed
s_l source_location_error Source location error (mm) 0.0 1.0
w_s wave_speed_error Wave speed error (m/s) 5900.0 5940.0
s_c sensor_coupling_gain Sensor coupling gain (-) 1.0 1.04
a_t arrival_time_bias Arrival time bias (us) 0.0 0.1

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.