Atomic Force Microscopy (AFM)

Atomic Force Microscopy (AFM)

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
🥇 DiffusionAFM 0.795 34.5 0.940 ✓ Certified Diffusion for SPM, 2024
🥈 SPM-Former 0.760 33.0 0.920 ✓ Certified Chen et al., Nano Letters 2024
🥉 Self-Sup AFM 0.723 31.5 0.895 ✓ Certified Self-supervised tip deconvolution, 2023
4 DeepAFM 0.685 30.0 0.870 ✓ Certified Somnath et al., NPJ Comput. Mater. 2021
5 PnP-ADMM 0.577 26.5 0.770 ✓ Certified Venkatakrishnan et al., 2013
6 Wiener Deconv 0.458 23.0 0.650 ✓ Certified Klapetek et al., Meas. Sci. Technol. 2011
7 Plane Fit 0.363 20.0 0.560 ✓ Certified Nečas & Klapetek, Open Physics 2012

Dataset: PWM Benchmark (7 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
🥇 SPM-Former + gradient 0.730
0.780
31.79 dB / 0.940
0.721
29.15 dB / 0.903
0.689
27.51 dB / 0.870
✓ Certified Chen et al., Nano Letters 24:3891, 2024
🥈 DiffusionAFM + gradient 0.695
0.776
32.05 dB / 0.943
0.671
26.17 dB / 0.837
0.638
25.42 dB / 0.815
✓ Certified Score-based diffusion for SPM image restoration, 2024
🥉 DeepAFM + gradient 0.608
0.733
28.95 dB / 0.899
0.555
21.52 dB / 0.669
0.537
21.09 dB / 0.650
✓ Certified Somnath et al., NPJ Comput. Mater. 2021
4 Self-Sup AFM + gradient 0.593
0.726
28.79 dB / 0.896
0.583
22.85 dB / 0.725
0.471
19.45 dB / 0.572
✓ Certified Self-supervised tip artifact deconvolution, 2023
5 Wiener Deconv + gradient 0.539
0.576
21.84 dB / 0.683
0.545
21.23 dB / 0.656
0.497
19.84 dB / 0.591
✓ Certified Klapetek et al., Meas. Sci. Technol. 2011
6 PnP-ADMM + gradient 0.525
0.629
24.24 dB / 0.777
0.483
19.04 dB / 0.552
0.463
18.83 dB / 0.541
✓ Certified Venkatakrishnan et al., IEEE GlobalSIP 2013
7 Plane Fit + gradient 0.422
0.450
17.8 dB / 0.490
0.433
17.51 dB / 0.476
0.383
16.14 dB / 0.408
✓ Certified Nečas & Klapetek, Open Physics 2012

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 3 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 SPM-Former + gradient 0.780 31.79 0.94
2 DiffusionAFM + gradient 0.776 32.05 0.943
3 DeepAFM + gradient 0.733 28.95 0.899
4 Self-Sup AFM + gradient 0.726 28.79 0.896
5 PnP-ADMM + gradient 0.629 24.24 0.777
6 Wiener Deconv + gradient 0.576 21.84 0.683
7 Plane Fit + gradient 0.450 17.8 0.49
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_shape_convolution -0.15 0.15 -
piezo_nonlinearity -1.0 2.0 -
thermal_drift -0.2 0.4 nm/s
scanner_hysteresis -2.0 4.0 -
View Details →
Dev 3 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 SPM-Former + gradient 0.721 29.15 0.903
2 DiffusionAFM + gradient 0.671 26.17 0.837
3 Self-Sup AFM + gradient 0.583 22.85 0.725
4 DeepAFM + gradient 0.555 21.52 0.669
5 Wiener Deconv + gradient 0.545 21.23 0.656
6 PnP-ADMM + gradient 0.483 19.04 0.552
7 Plane Fit + gradient 0.433 17.51 0.476
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_shape_convolution -0.15 0.15 -
piezo_nonlinearity -1.2 1.8 -
thermal_drift -0.24 0.36 nm/s
scanner_hysteresis -2.4 3.6 -
Submit Reconstruction View Details →
Hidden 3 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 SPM-Former + gradient 0.689 27.51 0.87
2 DiffusionAFM + gradient 0.638 25.42 0.815
3 DeepAFM + gradient 0.537 21.09 0.65
4 Wiener Deconv + gradient 0.497 19.84 0.591
5 Self-Sup AFM + gradient 0.471 19.45 0.572
6 PnP-ADMM + gradient 0.463 18.83 0.541
7 Plane Fit + gradient 0.383 16.14 0.408
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_shape_convolution -0.15 0.15 -
piezo_nonlinearity -0.7 2.3 -
thermal_drift -0.14 0.46 nm/s
scanner_hysteresis -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

S → D

S Sampling
D Detector

Mismatch Parameters

Symbol Parameter Description Nominal Perturbed
t_s tip_shape_convolution Tip shape convolution (-) 0.0 0.0
p_n piezo_nonlinearity Piezo nonlinearity (-) 0.0 1.0
t_d thermal_drift Thermal drift (nm/s) 0.0 0.2
s_h scanner_hysteresis Scanner hysteresis (-) 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.