Raman Imaging / Microscopy

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
🥇	PINN-Spectra PINN-Spectra Physics-informed neural network 33.54 dB SSIM 0.957 Checkpoint unavailable	0.787	33.54	0.957	✓ Certified	Physics-informed neural network
🥈	Cascade-UNet Cascade-UNet Physics-informed UNet, 2025 33.0 dB SSIM 0.922 Checkpoint unavailable	0.761	33.0	0.922	✓ Certified	Physics-informed UNet, 2025
🥉	U-Net-Spectra U-Net-Spectra Spectral U-Net variant 31.86 dB SSIM 0.941 Checkpoint unavailable	0.751	31.86	0.941	✓ Certified	Spectral U-Net variant
4	DiffusionSpectra DiffusionSpectra Zhang et al., 2024 31.04 dB SSIM 0.931 Checkpoint unavailable	0.733	31.04	0.931	✓ Certified	Zhang et al., 2024
5	ScoreSpectra ScoreSpectra Wei et al., 2025 30.84 dB SSIM 0.929 Checkpoint unavailable	0.729	30.84	0.929	✓ Certified	Wei et al., 2025
6	SpectraFormer SpectraFormer Spectroscopy transformer, 2024 30.64 dB SSIM 0.926 Checkpoint unavailable	0.724	30.64	0.926	✓ Certified	Spectroscopy transformer, 2024
7	CDAE CDAE Zhang et al., Sensors 2024 31.5 dB SSIM 0.895 Checkpoint unavailable	0.723	31.5	0.895	✓ Certified	Zhang et al., Sensors 2024
8	PnP-DnCNN PnP-DnCNN Zhang et al., 2017 27.9 dB SSIM 0.800 Try in SpecLab →	0.615	27.9	0.800	✓ Certified	Zhang et al., 2017
9	SVD SVD Singular Value Decomposition 26.39 dB SSIM 0.843 Try in SpecLab →	0.611	26.39	0.843	✓ Certified	Singular Value Decomposition
10	Baseline Correction Baseline Correction Polynomial fitting baseline 24.29 dB SSIM 0.779 Try in SpecLab →	0.544	24.29	0.779	✓ Certified	Polynomial fitting baseline
11	SG-ALS SG-ALS Savitzky-Golay + ALS baseline 24.3 dB SSIM 0.670 Try in SpecLab →	0.490	24.3	0.670	✓ Certified	Savitzky-Golay + ALS 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
🥇	U-Net-Spectra + gradient U-Net-Spectra + gradient Spectral U-Net variant Score 0.679 Correct & Reconstruct →	0.679	0.758 30.24 dB / 0.920	0.660 26.27 dB / 0.839	0.619 24.64 dB / 0.791	✓ Certified	Spectral U-Net variant
🥈	Cascade-UNet + gradient Cascade-UNet + gradient Physics-informed UNet, 2025 Score 0.672 Correct & Reconstruct →	0.672	0.756 30.87 dB / 0.929	0.677 26.7 dB / 0.851	0.582 23.27 dB / 0.742	✓ Certified	Physics-informed UNet, 2025
🥉	SpectraFormer + gradient SpectraFormer + gradient Spectroscopy transformer, 2024 Score 0.632 Correct & Reconstruct →	0.632	0.710 27.75 dB / 0.875	0.625 24.76 dB / 0.794	0.561 22.2 dB / 0.698	✓ Certified	Spectroscopy transformer, 2024
4	PINN-Spectra + gradient PINN-Spectra + gradient Physics-informed neural network Score 0.624 Correct & Reconstruct →	0.624	0.758 30.57 dB / 0.925	0.609 23.9 dB / 0.765	0.506 20.56 dB / 0.625	✓ Certified	Physics-informed neural network
5	ScoreSpectra + gradient ScoreSpectra + gradient Wei et al., 2025 Score 0.614 Correct & Reconstruct →	0.614	0.720 28.64 dB / 0.894	0.590 23.32 dB / 0.743	0.532 20.7 dB / 0.632	✓ Certified	Wei et al., 2025
6	SVD + gradient SVD + gradient Singular Value Decomposition Score 0.612 Correct & Reconstruct →	0.612	0.633 24.58 dB / 0.789	0.634 24.67 dB / 0.792	0.570 22.89 dB / 0.727	✓ Certified	Singular Value Decomposition
7	CDAE + gradient CDAE + gradient Zhang et al., Sensors 2024 Score 0.606 Correct & Reconstruct →	0.606	0.730 29.19 dB / 0.904	0.567 21.75 dB / 0.679	0.520 20.72 dB / 0.633	✓ Certified	Zhang et al., Sensors 2024
8	PnP-DnCNN + gradient PnP-DnCNN + gradient Zhang et al., 2017 Score 0.599 Correct & Reconstruct →	0.599	0.659 25.44 dB / 0.816	0.594 23.47 dB / 0.749	0.543 21.82 dB / 0.682	✓ Certified	Zhang et al., 2017
9	DiffusionSpectra + gradient DiffusionSpectra + gradient Zhang et al., 2024 Score 0.583 Correct & Reconstruct →	0.583	0.747 29.52 dB / 0.909	0.514 20.08 dB / 0.603	0.489 19.64 dB / 0.581	✓ Certified	Zhang et al., 2024
10	Baseline Correction + gradient Baseline Correction + gradient Polynomial fitting baseline Score 0.547 Correct & Reconstruct →	0.547	0.580 22.44 dB / 0.708	0.556 21.79 dB / 0.681	0.505 20.25 dB / 0.611	✓ Certified	Polynomial fitting baseline
11	SG-ALS + gradient SG-ALS + gradient Savitzky-Golay + ALS baseline Score 0.517 Correct & Reconstruct →	0.517	0.565 21.61 dB / 0.673	0.511 19.97 dB / 0.597	0.476 19.41 dB / 0.570	✓ Certified	Savitzky-Golay + ALS baseline

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	U-Net-Spectra + gradient	0.758	30.24	0.92
2	PINN-Spectra + gradient	0.758	30.57	0.925
3	Cascade-UNet + gradient	0.756	30.87	0.929
4	DiffusionSpectra + gradient	0.747	29.52	0.909
5	CDAE + gradient	0.730	29.19	0.904
6	ScoreSpectra + gradient	0.720	28.64	0.894
7	SpectraFormer + gradient	0.710	27.75	0.875
8	PnP-DnCNN + gradient	0.659	25.44	0.816
9	SVD + gradient	0.633	24.58	0.789
10	Baseline Correction + gradient	0.580	22.44	0.708
11	SG-ALS + gradient	0.565	21.61	0.673

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
spectral_calibration_shift	-0.4	0.8	cm^-1
fluorescence_background	-2.0	4.0	relative
laser_power_fluctuation	-1.0	2.0	-
cosmic_ray_artifact	-0.2	0.4	-

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	Cascade-UNet + gradient	0.677	26.7	0.851
2	U-Net-Spectra + gradient	0.660	26.27	0.839
3	SVD + gradient	0.634	24.67	0.792
4	SpectraFormer + gradient	0.625	24.76	0.794
5	PINN-Spectra + gradient	0.609	23.9	0.765
6	PnP-DnCNN + gradient	0.594	23.47	0.749
7	ScoreSpectra + gradient	0.590	23.32	0.743
8	CDAE + gradient	0.567	21.75	0.679
9	Baseline Correction + gradient	0.556	21.79	0.681
10	DiffusionSpectra + gradient	0.514	20.08	0.603
11	SG-ALS + gradient	0.511	19.97	0.597

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
spectral_calibration_shift	-0.48	0.72	cm^-1
fluorescence_background	-2.4	3.6	relative
laser_power_fluctuation	-1.2	1.8	-
cosmic_ray_artifact	-0.24	0.36	-

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	U-Net-Spectra + gradient	0.619	24.64	0.791
2	Cascade-UNet + gradient	0.582	23.27	0.742
3	SVD + gradient	0.570	22.89	0.727
4	SpectraFormer + gradient	0.561	22.2	0.698
5	PnP-DnCNN + gradient	0.543	21.82	0.682
6	ScoreSpectra + gradient	0.532	20.7	0.632
7	CDAE + gradient	0.520	20.72	0.633
8	PINN-Spectra + gradient	0.506	20.56	0.625
9	Baseline Correction + gradient	0.505	20.25	0.611
10	DiffusionSpectra + gradient	0.489	19.64	0.581
11	SG-ALS + gradient	0.476	19.41	0.57

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
spectral_calibration_shift	-0.28	0.92	cm^-1
fluorescence_background	-1.4	4.6	relative
laser_power_fluctuation	-0.7	2.3	-
cosmic_ray_artifact	-0.14	0.46	-

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

M → R → D

M Modulation

R Rotation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Perturbed
s_c	spectral_calibration_shift	Spectral calibration shift (cm^-1)	0.4
f_b	fluorescence_background	Fluorescence background (relative)	2.0
l_p	laser_power_fluctuation	Laser power fluctuation (-)	1.0
c_r	cosmic_ray_artifact	Cosmic ray artifact (-)	0.2

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.