Public

Photometric Stereo — Public Tier

(5 scenes)

Full-access development tier with all data visible.

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.

Parameter Specifications

True spec visible — use these exact values for Scenario III oracle reconstruction.

Parameter Spec Range True Value Unit
light_direction_error -1.0 – 2.0 0.5 degpersource
light_intensity_calibration 0.96 – 1.08 1.02 -
non_lambertian_surface_fraction -6.0 – 12.0 3.0 -
cast_shadow_fraction -3.0 – 6.0 1.5 ofpixels

InverseNet Baseline Scores

Method: CPU_baseline — Mismatch parameter: nominal

Scenario I (Ideal)

5.60 dB

SSIM 0.4704

Scenario II (Mismatch)

5.48 dB

SSIM 0.2508

Scenario III (Oracle)

19.14 dB

SSIM 0.2964

Per-scene breakdown (4 scenes)
Scene PSNR I SSIM I PSNR II SSIM II PSNR III SSIM III
scene_00 5.53 0.4672 5.49 0.2492 19.07 0.2950
scene_01 5.81 0.4796 5.98 0.2471 19.15 0.2888
scene_02 5.66 0.4660 5.19 0.2536 19.29 0.3009
scene_03 5.39 0.4690 5.28 0.2531 19.03 0.3011

Public Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 PS-Transformer + gradient 0.795 32.82 0.951 0.93 ✓ Certified Ikehata, ICCV 2023
2 CNN-PS + gradient 0.771 31.04 0.931 0.94 ✓ Certified Ikehata, ECCV 2018
3 Robust PCA + gradient 0.679 26.58 0.848 0.87 ✓ Certified Wu et al., ECCV 2010
4 LS Normal Est. + gradient 0.591 22.73 0.72 0.88 ✓ Certified Woodham, Opt. Eng. 1980

Visible Data Fields

y H_ideal spec_ranges x_true true_spec

Dataset

Format: HDF5
Scenes: 5

Scoring Formula

0.4 × PSNR_norm + 0.4 × SSIM + 0.2 × (1 − ‖y − Ĥx̂‖/‖y‖)

PSNR: 40% SSIM: 40% Consistency: 20%
Back to Photometric Stereo