Seismic Tomography

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
🥇	Domain-Adapted-CNN Domain-Adapted-CNN Domain adaptation CNN 33.73 dB SSIM 0.959 Checkpoint unavailable	0.792	33.73	0.959	✓ Certified	Domain adaptation CNN
🥈	SwinIR SwinIR Liang et al., ICCVW 2021 34.1 dB SSIM 0.942 Checkpoint unavailable	0.789	34.1	0.942	✓ Certified	Liang et al., ICCVW 2021
🥉	DiffusionExperimental DiffusionExperimental Zhang et al., 2024 33.51 dB SSIM 0.957 Checkpoint unavailable	0.787	33.51	0.957	✓ Certified	Zhang et al., 2024
4	ScoreExperimental ScoreExperimental Wei et al., 2025 33.42 dB SSIM 0.956 Checkpoint unavailable	0.785	33.42	0.956	✓ Certified	Wei et al., 2025
5	ExpFormer ExpFormer Experimental science transformer, 2024 32.6 dB SSIM 0.949 Checkpoint unavailable	0.768	32.6	0.949	✓ Certified	Experimental science transformer, 2024
6	ResUNet ResUNet Residual U-Net baseline 32.6 dB SSIM 0.915 Checkpoint unavailable	0.751	32.6	0.915	✓ Certified	Residual U-Net baseline
7	PnP-ADMM PnP-ADMM ADMM + denoiser prior 27.86 dB SSIM 0.878 Try in SpecLab →	0.653	27.86	0.878	✓ Certified	ADMM + denoiser prior
8	PnP-RED PnP-RED Romano et al., IEEE TIP 2017 28.9 dB SSIM 0.835 Try in SpecLab →	0.649	28.9	0.835	✓ Certified	Romano et al., IEEE TIP 2017
9	Matched Filter Matched Filter Optimal linear filter 26.62 dB SSIM 0.849 Try in SpecLab →	0.618	26.62	0.849	✓ Certified	Optimal linear filter
10	Wiener Filter Wiener Filter Wiener filtering baseline 26.5 dB SSIM 0.846 Try in SpecLab →	0.615	26.5	0.846	✓ Certified	Wiener filtering baseline
11	Tikhonov Tikhonov Analytical baseline 25.4 dB SSIM 0.710 Try in SpecLab →	0.528	25.4	0.710	✓ Certified	Analytical baseline

Dataset: PWM Benchmark (11 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
🥇	ExpFormer + gradient ExpFormer + gradient Experimental science transformer, 2024 Score 0.703 Correct & Reconstruct →	0.703	0.750 30.74 dB / 0.927	0.709 28.74 dB / 0.895	0.651 26.23 dB / 0.838	✓ Certified	Experimental science transformer, 2024
🥈	SwinIR + gradient SwinIR + gradient Liang et al., ICCVW 2021 Score 0.694 Correct & Reconstruct →	0.694	0.771 31.71 dB / 0.939	0.705 28.26 dB / 0.886	0.605 24.05 dB / 0.770	✓ Certified	Liang et al., ICCVW 2021
🥉	ScoreExperimental + gradient ScoreExperimental + gradient Wei et al., 2025 Score 0.679 Correct & Reconstruct →	0.679	0.782 31.69 dB / 0.939	0.655 25.65 dB / 0.822	0.601 24.24 dB / 0.777	✓ Certified	Wei et al., 2025
4	ResUNet + gradient ResUNet + gradient Residual U-Net baseline Score 0.668 Correct & Reconstruct →	0.668	0.743 29.64 dB / 0.911	0.654 25.39 dB / 0.814	0.606 23.59 dB / 0.754	✓ Certified	Residual U-Net baseline
5	Domain-Adapted-CNN + gradient Domain-Adapted-CNN + gradient Domain adaptation CNN Score 0.652 Correct & Reconstruct →	0.652	0.763 30.93 dB / 0.930	0.626 24.86 dB / 0.798	0.567 22.09 dB / 0.694	✓ Certified	Domain adaptation CNN
6	DiffusionExperimental + gradient DiffusionExperimental + gradient Zhang et al., 2024 Score 0.640 Correct & Reconstruct →	0.640	0.758 30.56 dB / 0.925	0.615 24.38 dB / 0.782	0.548 21.87 dB / 0.684	✓ Certified	Zhang et al., 2024
7	PnP-RED + gradient PnP-RED + gradient Romano et al., IEEE TIP 2017 Score 0.630 Correct & Reconstruct →	0.630	0.680 26.34 dB / 0.841	0.623 24.76 dB / 0.794	0.588 23.39 dB / 0.746	✓ Certified	Romano et al., IEEE TIP 2017
8	Wiener Filter + gradient Wiener Filter + gradient Wiener filtering baseline Score 0.594 Correct & Reconstruct →	0.594	0.659 25.03 dB / 0.803	0.570 21.94 dB / 0.687	0.553 21.89 dB / 0.685	✓ Certified	Wiener filtering baseline
9	Matched Filter + gradient Matched Filter + gradient Optimal linear filter Score 0.585 Correct & Reconstruct →	0.585	0.631 24.24 dB / 0.777	0.578 22.13 dB / 0.696	0.547 21.75 dB / 0.679	✓ Certified	Optimal linear filter
10	Tikhonov + gradient Tikhonov + gradient Tikhonov, Doklady 1963 Score 0.581 Correct & Reconstruct →	0.581	0.637 24.22 dB / 0.776	0.579 22.69 dB / 0.719	0.527 21.03 dB / 0.647	✓ Certified	Tikhonov, Doklady 1963
11	PnP-ADMM + gradient PnP-ADMM + gradient ADMM + denoiser prior Score 0.542 Correct & Reconstruct →	0.542	0.665 26.0 dB / 0.832	0.526 21.05 dB / 0.648	0.434 17.6 dB / 0.480	✓ Certified	ADMM + denoiser prior

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	ScoreExperimental + gradient	0.782	31.69	0.939
2	SwinIR + gradient	0.771	31.71	0.939
3	Domain-Adapted-CNN + gradient	0.763	30.93	0.93
4	DiffusionExperimental + gradient	0.758	30.56	0.925
5	ExpFormer + gradient	0.750	30.74	0.927
6	ResUNet + gradient	0.743	29.64	0.911
7	PnP-RED + gradient	0.680	26.34	0.841
8	PnP-ADMM + gradient	0.665	26.0	0.832
9	Wiener Filter + gradient	0.659	25.03	0.803
10	Tikhonov + gradient	0.637	24.22	0.776
11	Matched Filter + gradient	0.631	24.24	0.777

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
velocity_model_error	4900.0	5200.0	m/s
source_location_error	-10.0	20.0	m
receiver_coupling	0.97	1.06	-
timing_error	-0.0004	0.0008	s

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	ExpFormer + gradient	0.709	28.74	0.895
2	SwinIR + gradient	0.705	28.26	0.886
3	ScoreExperimental + gradient	0.655	25.65	0.822
4	ResUNet + gradient	0.654	25.39	0.814
5	Domain-Adapted-CNN + gradient	0.626	24.86	0.798
6	PnP-RED + gradient	0.623	24.76	0.794
7	DiffusionExperimental + gradient	0.615	24.38	0.782
8	Tikhonov + gradient	0.579	22.69	0.719
9	Matched Filter + gradient	0.578	22.13	0.696
10	Wiener Filter + gradient	0.570	21.94	0.687
11	PnP-ADMM + gradient	0.526	21.05	0.648

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
velocity_model_error	4880.0	5180.0	m/s
source_location_error	-12.0	18.0	m
receiver_coupling	0.964	1.054	-
timing_error	-0.00048	0.00072	s

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	ExpFormer + gradient	0.651	26.23	0.838
2	ResUNet + gradient	0.606	23.59	0.754
3	SwinIR + gradient	0.605	24.05	0.77
4	ScoreExperimental + gradient	0.601	24.24	0.777
5	PnP-RED + gradient	0.588	23.39	0.746
6	Domain-Adapted-CNN + gradient	0.567	22.09	0.694
7	Wiener Filter + gradient	0.553	21.89	0.685
8	DiffusionExperimental + gradient	0.548	21.87	0.684
9	Matched Filter + gradient	0.547	21.75	0.679
10	Tikhonov + gradient	0.527	21.03	0.647
11	PnP-ADMM + gradient	0.434	17.6	0.48

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
velocity_model_error	4930.0	5230.0	m/s
source_location_error	-7.0	23.0	m
receiver_coupling	0.979	1.069	-
timing_error	-0.00028	0.00092	s

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 → D

P Propagation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
v_m	velocity_model_error	Velocity model error (m/s)	5000.0	5100.0
s_l	source_location_error	Source location error (m)	0.0	10.0
r_c	receiver_coupling	Receiver coupling (-)	1.0	1.03
t_e	timing_error	Timing error (s)	0.0	0.0004

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.