Hidden

NeRF — Hidden Tier

(5 scenes)

Fully blind server-side evaluation — no data download.

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.

Parameter Specifications

🔒

True spec hidden — blind evaluation, only ranges available.

Parameter Spec Range Unit
camera_pose -0.7 – 2.3 mm/deg
focal_length -3.5 – 11.5 pixels
distortion -0.007 – 0.023
exposure -7.0 – 23.0 %

Hidden Tier Leaderboard

# Method Score PSNR SSIM Consistency Trust Source
1 NeRFactor2 + gradient 0.674 26.62 0.849 0.84 ✓ Certified Barron et al., NeurIPS 2024
2 3D-GS++ + gradient 0.671 27.09 0.86 0.78 ✓ Certified Kerbl et al., SIGGRAPH 2024
3 COLMAP+MVS + gradient 0.613 24.21 0.776 0.8 ✓ Certified Schonberger & Frahm, CVPR 2016
4 GaussianShader + gradient 0.603 23.34 0.744 0.86 ✓ Certified Wang et al., ICCV 2024
5 Instant-NGP + gradient 0.600 23.68 0.757 0.8 ✓ Certified Muller et al., SIGGRAPH 2022
6 2DGS + gradient 0.585 23.18 0.738 0.79 ✓ Certified Huang et al., CVPR 2024
7 NeRF + gradient 0.580 22.37 0.706 0.87 ✓ Certified Mildenhall et al., ECCV 2020
8 3D-GS + gradient 0.563 22.19 0.698 0.81 ✓ Certified Kerbl et al., SIGGRAPH 2023
9 Photogrammetry + gradient 0.562 22.61 0.715 0.75 ✓ Certified Structure-from-Motion baseline
10 Mesh-GS + gradient 0.419 17.55 0.478 0.76 ✓ Certified Li et al., ECCV 2024
11 Mip-NeRF 360 + gradient 0.409 16.63 0.432 0.85 ✓ Certified Barron et al., CVPR 2022

Dataset

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 NeRF