AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Photo-Realistic Neural Domain Randomization

Sergey Zakharov, Rareș Ambruș, Vitor Guizilini, Wadim Kehl, Adrien Gaidon

ECCV 2022 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Geometric alignment and synthetic-to-real adaptation · 12 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train on PNDR renderings instead of 1088 ray-traced imagesCBOD · HB Scene 10 (new lighting)ADD AUC: 30.57 -> 40.44Improved
train on PNDR renderings instead of real imagesCBOD · HB Scene 10 (new lighting)ADD AUC: 22.24 -> 40.44Improved
use PNDR instead of 1088 ray-traced imagesCBOD · HB Scene 5 (in-domain)ADD AUC: 83.62 -> 88.18Improved
increase ray-traced training images from 1088 to 4352CBOD · HB Scene 10ADD AUC: 30.57 -> 37.35Improved
use PNDR instead of ray-traced data on LM scenesCBOD · LM Scenes 2, 8, 15ADD AUC: 19.71 -> 27.33Improved
add specularity and roughness randomization to albedo (ablation)PNDR · HB5 ablationADD AUC: 85.43 -> 88.18Improved
use dynamic instead of fixed lighting (ablation)PNDR · HB5 ablationADD AUC: 73.04 -> 88.18Improved
add indirect lighting outputs (ablation)PNDR · HB5 ablationADD AUC: 87.09 -> 88.18No clear effect
add indirect lighting outputs (ablation)PNDR · HB5 ablationCorr (mm): 38.42 -> 34.32Improved
train monodepth2 on PNDR instead of ray-tracedMonodepth2 · HB10AbsRel: 0.162 -> 0.154Improved
train PackNet-SfM on PNDR instead of ray-tracedPackNet-SfM · HB5AbsRel: 0.11 -> 0.082Improved
use paired Pix2PixHD translation instead of unpaired CUTCBOD · HB Scene 5ADD AUC: 66.49 -> 89.59Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy