DNA-PAINT Super-Resolution

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
🥇	DiffPAINT DiffPAINT Gao et al., NeurIPS 2024 39.7 dB SSIM 0.958 Checkpoint unavailable	0.891	39.7	0.958	✓ Certified	Gao et al., NeurIPS 2024
🥈	PhysSTORM PhysSTORM Chen et al., Nat. Commun. 2024 38.1 dB SSIM 0.946 Checkpoint unavailable	0.858	38.1	0.946	✓ Certified	Chen et al., Nat. Commun. 2024
🥉	SwinSTORM SwinSTORM Wang et al., Bioinformatics 2023 36.8 dB SSIM 0.934 Checkpoint unavailable	0.830	36.8	0.934	✓ Certified	Wang et al., Bioinformatics 2023
4	TransPAINT TransPAINT Li et al., Nat. Methods 2022 35.2 dB SSIM 0.918 Checkpoint unavailable	0.796	35.2	0.918	✓ Certified	Li et al., Nat. Methods 2022
5	DECODE DECODE Speiser et al., Nat. Methods 2021 32.6 dB SSIM 0.878 Checkpoint unavailable	0.732	32.6	0.878	✓ Certified	Speiser et al., Nat. Methods 2021
6	DeepSTORM DeepSTORM Nehme et al., Optica 2018 29.1 dB SSIM 0.831 Checkpoint unavailable	0.650	29.1	0.831	✓ Certified	Nehme et al., Optica 2018
7	DAOSTORM DAOSTORM Holden et al., Nat. Methods 2011 25.4 dB SSIM 0.762 Try in SpecLab →	0.554	25.4	0.762	✓ Certified	Holden et al., Nat. Methods 2011
8	PALM PALM Betzig et al., Science 2006 22.8 dB SSIM 0.718 Try in SpecLab →	0.489	22.8	0.718	✓ Certified	Betzig et al., Science 2006
9	STORM-2D STORM-2D Rust et al., Nat. Methods 2006 21.3 dB SSIM 0.695 Try in SpecLab →	0.453	21.3	0.695	✓ Certified	Rust et al., Nat. Methods 2006

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
🥇	DiffPAINT + gradient DiffPAINT + gradient Gao et al., NeurIPS 2024 Score 0.794 Correct & Reconstruct →	0.794	0.863 38.14 dB / 0.983	0.782 32.48 dB / 0.948	0.738 31.0 dB / 0.931	✓ Certified	Gao et al., NeurIPS 2024
🥈	PhysSTORM + gradient PhysSTORM + gradient Chen et al., Nat. Commun. 2024 Score 0.777 Correct & Reconstruct →	0.777	0.825 36.32 dB / 0.975	0.778 32.91 dB / 0.952	0.728 30.45 dB / 0.923	✓ Certified	Chen et al., Nat. Commun. 2024
🥉	SwinSTORM + gradient SwinSTORM + gradient Wang et al., Bioinformatics 2023 Score 0.766 Correct & Reconstruct →	0.766	0.806 33.96 dB / 0.961	0.773 32.46 dB / 0.947	0.720 29.84 dB / 0.914	✓ Certified	Wang et al., Bioinformatics 2023
4	TransPAINT + gradient TransPAINT + gradient Li et al., Nat. Methods 2022 Score 0.700 Correct & Reconstruct →	0.700	0.787 32.82 dB / 0.951	0.695 28.06 dB / 0.882	0.618 24.51 dB / 0.786	✓ Certified	Li et al., Nat. Methods 2022
5	DECODE + gradient DECODE + gradient Speiser et al., Nat. Methods 2021 Score 0.599 Correct & Reconstruct →	0.599	0.751 30.67 dB / 0.926	0.580 22.3 dB / 0.703	0.465 18.63 dB / 0.531	✓ Certified	Speiser et al., Nat. Methods 2021
6	DAOSTORM + gradient DAOSTORM + gradient Holden et al., Nat. Methods 2011 Score 0.578 Correct & Reconstruct →	0.578	0.606 23.29 dB / 0.742	0.574 21.94 dB / 0.687	0.554 21.78 dB / 0.680	✓ Certified	Holden et al., Nat. Methods 2011
7	DeepSTORM + gradient DeepSTORM + gradient Nehme et al., Optica 2018 Score 0.534 Correct & Reconstruct →	0.534	0.710 27.46 dB / 0.869	0.475 19.17 dB / 0.558	0.416 16.73 dB / 0.437	✓ Certified	Nehme et al., Optica 2018
8	PALM + gradient PALM + gradient Betzig et al., Science 2006 Score 0.525 Correct & Reconstruct →	0.525	0.534 20.72 dB / 0.633	0.526 20.23 dB / 0.610	0.516 20.58 dB / 0.626	✓ Certified	Betzig et al., Science 2006
9	STORM-2D + gradient STORM-2D + gradient Rust et al., Nat. Methods 2006 Score 0.441 Correct & Reconstruct →	0.441	0.492 19.19 dB / 0.559	0.437 17.84 dB / 0.492	0.395 16.64 dB / 0.432	✓ Certified	Rust et al., Nat. Methods 2006

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	DiffPAINT + gradient	0.863	38.14	0.983
2	PhysSTORM + gradient	0.825	36.32	0.975
3	SwinSTORM + gradient	0.806	33.96	0.961
4	TransPAINT + gradient	0.787	32.82	0.951
5	DECODE + gradient	0.751	30.67	0.926
6	DeepSTORM + gradient	0.710	27.46	0.869
7	DAOSTORM + gradient	0.606	23.29	0.742
8	PALM + gradient	0.534	20.72	0.633
9	STORM-2D + gradient	0.492	19.19	0.559

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
binding_on_rate	-0.4	0.8	relative
imager_strand_concentration	2.0	11.0	nM
drift_rate	-0.6	1.2	nm/frame
background_from_non_specific_binding	-2.0	4.0	-

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	DiffPAINT + gradient	0.782	32.48	0.948
2	PhysSTORM + gradient	0.778	32.91	0.952
3	SwinSTORM + gradient	0.773	32.46	0.947
4	TransPAINT + gradient	0.695	28.06	0.882
5	DECODE + gradient	0.580	22.3	0.703
6	DAOSTORM + gradient	0.574	21.94	0.687
7	PALM + gradient	0.526	20.23	0.61
8	DeepSTORM + gradient	0.475	19.17	0.558
9	STORM-2D + gradient	0.437	17.84	0.492

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
binding_on_rate	-0.48	0.72	relative
imager_strand_concentration	1.4	10.4	nM
drift_rate	-0.72	1.08	nm/frame
background_from_non_specific_binding	-2.4	3.6	-

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	DiffPAINT + gradient	0.738	31.0	0.931
2	PhysSTORM + gradient	0.728	30.45	0.923
3	SwinSTORM + gradient	0.720	29.84	0.914
4	TransPAINT + gradient	0.618	24.51	0.786
5	DAOSTORM + gradient	0.554	21.78	0.68
6	PALM + gradient	0.516	20.58	0.626
7	DECODE + gradient	0.465	18.63	0.531
8	DeepSTORM + gradient	0.416	16.73	0.437
9	STORM-2D + gradient	0.395	16.64	0.432

Spec Ranges (4 parameters)

Parameter	Min	Max	Unit
binding_on_rate	-0.28	0.92	relative
imager_strand_concentration	2.9	11.9	nM
drift_rate	-0.42	1.38	nm/frame
background_from_non_specific_binding	-1.4	4.6	-

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 → D

M Modulation

D Detector

Mismatch Parameters

Symbol	Parameter	Description	Nominal	Perturbed
b_o	binding_on_rate	Binding on-rate (relative)	0.0	0.4
i_s	imager_strand_concentration	Imager strand concentration (nM)	5.0	8.0
d_r	drift_rate	Drift rate (nm/frame)	0.0	0.6
b_f	background_from_non_specific_binding	Background from non-specific binding (-)	0.0	2.0

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.