Stimulated Raman Scattering (SRS) 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 32.88 dB SSIM 0.951 Checkpoint unavailable	0.774	32.88	0.951	✓ 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
🥉	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
4	SpectraFormer SpectraFormer Spectroscopy transformer, 2024 30.48 dB SSIM 0.924 Checkpoint unavailable	0.720	30.48	0.924	✓ Certified	Spectroscopy transformer, 2024
5	DiffusionSpectra DiffusionSpectra Zhang et al., 2024 30.37 dB SSIM 0.922 Checkpoint unavailable	0.717	30.37	0.922	✓ Certified	Zhang et al., 2024
6	ScoreSpectra ScoreSpectra Wei et al., 2025 30.2 dB SSIM 0.920 Checkpoint unavailable	0.713	30.2	0.920	✓ Certified	Wei et al., 2025
7	U-Net-Spectra U-Net-Spectra Spectral U-Net variant 29.37 dB SSIM 0.907 Checkpoint unavailable	0.693	29.37	0.907	✓ Certified	Spectral U-Net variant
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	Baseline Correction Baseline Correction Polynomial fitting baseline 25.07 dB SSIM 0.804 Try in SpecLab →	0.570	25.07	0.804	✓ Certified	Polynomial fitting baseline
10	SVD SVD Singular Value Decomposition 24.99 dB SSIM 0.802 Try in SpecLab →	0.568	24.99	0.802	✓ Certified	Singular Value Decomposition
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
🥇	PINN-Spectra + gradient PINN-Spectra + gradient Physics-informed neural network Score 0.643 Correct & Reconstruct →	0.643	0.754 30.87 dB / 0.929	0.600 23.24 dB / 0.740	0.576 22.6 dB / 0.715	✓ Certified	Physics-informed neural network
🥈	Cascade-UNet + gradient Cascade-UNet + gradient Physics-informed UNet, 2025 Score 0.641 Correct & Reconstruct →	0.641	0.779 31.86 dB / 0.941	0.602 23.88 dB / 0.764	0.542 21.19 dB / 0.654	✓ Certified	Physics-informed UNet, 2025
🥉	CDAE + gradient CDAE + gradient Zhang et al., Sensors 2024 Score 0.638 Correct & Reconstruct →	0.638	0.725 28.62 dB / 0.893	0.628 24.88 dB / 0.798	0.561 21.96 dB / 0.688	✓ Certified	Zhang et al., Sensors 2024
4	SpectraFormer + gradient SpectraFormer + gradient Spectroscopy transformer, 2024 Score 0.632 Correct & Reconstruct →	0.632	0.737 29.05 dB / 0.901	0.630 24.37 dB / 0.781	0.529 21.38 dB / 0.663	✓ Certified	Spectroscopy transformer, 2024
5	PnP-DnCNN + gradient PnP-DnCNN + gradient Zhang et al., 2017 Score 0.623 Correct & Reconstruct →	0.623	0.656 25.22 dB / 0.809	0.609 23.42 dB / 0.747	0.604 23.61 dB / 0.754	✓ Certified	Zhang et al., 2017
6	DiffusionSpectra + gradient DiffusionSpectra + gradient Zhang et al., 2024 Score 0.604 Correct & Reconstruct →	0.604	0.737 29.03 dB / 0.901	0.568 22.44 dB / 0.708	0.506 20.42 dB / 0.619	✓ Certified	Zhang et al., 2024
7	ScoreSpectra + gradient ScoreSpectra + gradient Wei et al., 2025 Score 0.601 Correct & Reconstruct →	0.601	0.736 28.96 dB / 0.900	0.562 21.79 dB / 0.681	0.505 19.9 dB / 0.594	✓ Certified	Wei et al., 2025
8	Baseline Correction + gradient Baseline Correction + gradient Polynomial fitting baseline Score 0.558 Correct & Reconstruct →	0.558	0.621 23.42 dB / 0.747	0.536 20.86 dB / 0.639	0.516 20.71 dB / 0.632	✓ Certified	Polynomial fitting baseline
9	U-Net-Spectra + gradient U-Net-Spectra + gradient Spectral U-Net variant Score 0.549 Correct & Reconstruct →	0.549	0.683 26.41 dB / 0.843	0.510 19.95 dB / 0.596	0.453 18.43 dB / 0.521	✓ Certified	Spectral U-Net variant
10	SVD + gradient SVD + gradient Singular Value Decomposition Score 0.545 Correct & Reconstruct →	0.545	0.594 22.92 dB / 0.728	0.547 21.68 dB / 0.676	0.495 19.49 dB / 0.574	✓ Certified	Singular Value Decomposition
11	SG-ALS + gradient SG-ALS + gradient Savitzky-Golay + ALS baseline Score 0.544 Correct & Reconstruct →	0.544	0.572 22.01 dB / 0.690	0.547 21.68 dB / 0.676	0.512 20.03 dB / 0.600	✓ 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	Cascade-UNet + gradient	0.779	31.86	0.941
2	PINN-Spectra + gradient	0.754	30.87	0.929
3	SpectraFormer + gradient	0.737	29.05	0.901
4	DiffusionSpectra + gradient	0.737	29.03	0.901
5	ScoreSpectra + gradient	0.736	28.96	0.9
6	CDAE + gradient	0.725	28.62	0.893
7	U-Net-Spectra + gradient	0.683	26.41	0.843
8	PnP-DnCNN + gradient	0.656	25.22	0.809
9	Baseline Correction + gradient	0.621	23.42	0.747
10	SVD + gradient	0.594	22.92	0.728
11	SG-ALS + gradient	0.572	22.01	0.69

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
lock_in_phase_error	-2.0	4.0	deg
cross_phase_modulation	-1.0	2.0	-
laser_intensity_noise_(rin)	-152.0	-146.0	dBc/Hz

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	SpectraFormer + gradient	0.630	24.37	0.781
2	CDAE + gradient	0.628	24.88	0.798
3	PnP-DnCNN + gradient	0.609	23.42	0.747
4	Cascade-UNet + gradient	0.602	23.88	0.764
5	PINN-Spectra + gradient	0.600	23.24	0.74
6	DiffusionSpectra + gradient	0.568	22.44	0.708
7	ScoreSpectra + gradient	0.562	21.79	0.681
8	SVD + gradient	0.547	21.68	0.676
9	SG-ALS + gradient	0.547	21.68	0.676
10	Baseline Correction + gradient	0.536	20.86	0.639
11	U-Net-Spectra + gradient	0.510	19.95	0.596

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
lock_in_phase_error	-2.4	3.6	deg
cross_phase_modulation	-1.2	1.8	-
laser_intensity_noise_(rin)	-152.4	-146.4	dBc/Hz

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	PnP-DnCNN + gradient	0.604	23.61	0.754
2	PINN-Spectra + gradient	0.576	22.6	0.715
3	CDAE + gradient	0.561	21.96	0.688
4	Cascade-UNet + gradient	0.542	21.19	0.654
5	SpectraFormer + gradient	0.529	21.38	0.663
6	Baseline Correction + gradient	0.516	20.71	0.632
7	SG-ALS + gradient	0.512	20.03	0.6
8	DiffusionSpectra + gradient	0.506	20.42	0.619
9	ScoreSpectra + gradient	0.505	19.9	0.594
10	SVD + gradient	0.495	19.49	0.574
11	U-Net-Spectra + gradient	0.453	18.43	0.521

Spec Ranges (3 parameters)

Parameter	Min	Max	Unit
lock_in_phase_error	-1.4	4.6	deg
cross_phase_modulation	-0.7	2.3	-
laser_intensity_noise_(rin)	-151.4	-145.4	dBc/Hz

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	Nominal	Perturbed
l_p	lock_in_phase_error	Lock-in phase error (deg)	0.0	2.0
c_m	cross_phase_modulation	Cross-phase modulation (-)	0.0	1.0
l_i	laser_intensity_noise_(rin)	Laser intensity noise (RIN) (dBc/Hz)	-150.0	-148.0

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.