Scanning Tunneling Microscopy (STM)

Scanning Tunneling Microscopy (STM)

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
🥇 SPM-Former 0.782 33.25 0.955 ✓ Certified Chen et al., NanoLett 2024
🥈 ScoreSPM 0.737 31.21 0.934 ✓ Certified Wei et al., 2025
🥉 E2E-BTR 0.734 31.8 0.908 ✓ Certified Kossler et al., Sci. Rep. 2022
4 DeepSPM 0.697 30.4 0.880 ✓ Certified Alldritt et al., Commun. Phys. 2020
5 DiffusionSPM 0.686 29.12 0.902 ✓ Certified Zhang et al., 2024
6 U-Net-SPM 0.660 28.13 0.883 ✓ Certified SPM U-Net variant
7 TV-Deconvolution 0.658 28.03 0.881 ✓ Certified TV regularization for SPM
8 Reg-Deconv 0.582 26.8 0.770 ✓ Certified Dongmo et al., 2000
9 MLE Reconstruction 0.529 23.85 0.763 ✓ Certified Classical statistical method
10 BTR 0.452 23.2 0.630 ✓ Certified Villarrubia, JRNIST 1997

Dataset: PWM Benchmark (10 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 0.725
0.782
32.07 dB / 0.943
0.723
28.93 dB / 0.899
0.669
26.39 dB / 0.843
✓ Certified Chen et al., NanoLett 2024
🥈 E2E-BTR + gradient 0.662
0.738
29.7 dB / 0.912
0.641
25.62 dB / 0.821
0.608
23.53 dB / 0.751
✓ Certified Kossler et al., Sci. Rep. 2022
🥉 DeepSPM + gradient 0.647
0.711
28.04 dB / 0.882
0.644
25.24 dB / 0.810
0.586
22.53 dB / 0.712
✓ Certified Alldritt et al., Commun. Phys. 2020
4 ScoreSPM + gradient 0.640
0.750
29.82 dB / 0.914
0.613
24.13 dB / 0.773
0.558
21.56 dB / 0.671
✓ Certified Wei et al., 2025
5 U-Net-SPM + gradient 0.596
0.699
27.09 dB / 0.860
0.583
22.69 dB / 0.719
0.506
20.31 dB / 0.613
✓ Certified SPM U-Net variant
6 Reg-Deconv + gradient 0.589
0.637
24.58 dB / 0.789
0.579
22.33 dB / 0.704
0.552
21.4 dB / 0.664
✓ Certified Dongmo et al., 2000
7 TV-Deconvolution + gradient 0.545
0.662
25.6 dB / 0.821
0.530
21.07 dB / 0.649
0.443
17.89 dB / 0.495
✓ Certified TV regularization for SPM
8 BTR + gradient 0.528
0.575
21.65 dB / 0.675
0.534
20.83 dB / 0.638
0.476
19.19 dB / 0.559
✓ Certified Villarrubia, JRNIST 1997
9 MLE Reconstruction + gradient 0.522
0.558
21.4 dB / 0.664
0.523
20.74 dB / 0.634
0.484
19.48 dB / 0.573
✓ Certified Classical statistical method
10 DiffusionSPM + gradient 0.516
0.711
27.4 dB / 0.868
0.465
18.9 dB / 0.545
0.372
15.37 dB / 0.371
✓ Certified Zhang et al., 2024

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 0.782 32.07 0.943
2 ScoreSPM + gradient 0.750 29.82 0.914
3 E2E-BTR + gradient 0.738 29.7 0.912
4 DeepSPM + gradient 0.711 28.04 0.882
5 DiffusionSPM + gradient 0.711 27.4 0.868
6 U-Net-SPM + gradient 0.699 27.09 0.86
7 TV-Deconvolution + gradient 0.662 25.6 0.821
8 Reg-Deconv + gradient 0.637 24.58 0.789
9 BTR + gradient 0.575 21.65 0.675
10 MLE Reconstruction + gradient 0.558 21.4 0.664
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_electronic_structure -0.15 0.15 -
piezo_creep -1.0 2.0 -
tunneling_barrier_height 4.2 5.1 eV
vibration_amplitude -1.0 2.0 pm
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 0.723 28.93 0.899
2 DeepSPM + gradient 0.644 25.24 0.81
3 E2E-BTR + gradient 0.641 25.62 0.821
4 ScoreSPM + gradient 0.613 24.13 0.773
5 U-Net-SPM + gradient 0.583 22.69 0.719
6 Reg-Deconv + gradient 0.579 22.33 0.704
7 BTR + gradient 0.534 20.83 0.638
8 TV-Deconvolution + gradient 0.530 21.07 0.649
9 MLE Reconstruction + gradient 0.523 20.74 0.634
10 DiffusionSPM + gradient 0.465 18.9 0.545
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_electronic_structure -0.15 0.15 -
piezo_creep -1.2 1.8 -
tunneling_barrier_height 4.14 5.04 eV
vibration_amplitude -1.2 1.8 pm
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 0.669 26.39 0.843
2 E2E-BTR + gradient 0.608 23.53 0.751
3 DeepSPM + gradient 0.586 22.53 0.712
4 ScoreSPM + gradient 0.558 21.56 0.671
5 Reg-Deconv + gradient 0.552 21.4 0.664
6 U-Net-SPM + gradient 0.506 20.31 0.613
7 MLE Reconstruction + gradient 0.484 19.48 0.573
8 BTR + gradient 0.476 19.19 0.559
9 TV-Deconvolution + gradient 0.443 17.89 0.495
10 DiffusionSPM + gradient 0.372 15.37 0.371
Spec Ranges (4 parameters)
Parameter Min Max Unit
tip_electronic_structure -0.15 0.15 -
piezo_creep -0.7 2.3 -
tunneling_barrier_height 4.29 5.19 eV
vibration_amplitude -0.7 2.3 pm
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_e tip_electronic_structure Tip electronic structure (-) 0.0 0.0
p_c piezo_creep Piezo creep (-) 0.0 1.0
t_b tunneling_barrier_height Tunneling barrier height (eV) 4.5 4.8
v_a vibration_amplitude Vibration amplitude (pm) 0.0 1.0

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.