AI Paper Atlas

Research map / 3D vision / Human pose estimation

Focal Length and Object Pose Estimation via Render and Compare

Georgy Ponimatkin, Yann Labbé, Bryan Russell, Mathieu Aubry, Josef Sivic

CVPR 2022 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 20 citations (OpenAlex)

Check 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
Add disentangled reprojection loss to Huber focal-length regression (ablation)FocalPose · Pix3D sofaMedErr_R: 3.28 -> 2.98Improved
Add disentangled reprojection loss to Huber focal-length regression (ablation)FocalPose · Pix3D sofaMedErr_t: 1.42 -> 1.29Improved
Add disentangled reprojection loss to Huber focal-length regression (ablation)FocalPose · Pix3D sofaMedErr_f: 1.45 -> 1.36Improved
Add reprojection-based supervision to Huber focal-length regression (ablation)FocalPose · Pix3D sofaMedErr_R: 6.61 -> 2.98Improved
Add reprojection-based supervision to Huber focal-length regression (ablation)FocalPose · Pix3D sofaMedErr_f: 4.17 -> 1.36Improved
Use a mix of mostly synthetic data and a small amount of real data rather than synthetic-only or real-only data (ablation)FocalPose coarse model · Pix3D sofaMedErr_R: 5.44 / 4.13 -> 2.98Improved
Use a mix of mostly synthetic data and a small amount of real data rather than synthetic-only or real-only data (ablation)FocalPose coarse model · Pix3D sofaMedErr_t: 2.18 / 1.92 -> 1.29Improved
Use a mix of mostly synthetic data and a small amount of real data rather than synthetic-only or real-only data (ablation)FocalPose coarse model · Pix3D sofaMedErr_f: 2.04 / 1.91 -> 1.36Improved
Increase the real-data fraction in mixed training batches from 0.5% to 5% (ablation)FocalPose coarse model · Pix3D sofaMedErr_R: 2.98 -> 3.08Hurt
Use render-and-compare with jointly estimated focal length and poseFocalPose · Pix3D, CompCars, Stanford CarsMedErr_f: Pix3D 1.72 -> 1.53; CompCars 2.95 -> 2.65; Stanford Cars 2.01 -> 1.09Improved
Use render-and-compare with jointly estimated focal length and poseFocalPose · Pix3D, CompCars, Stanford CarsMedErr_t: Pix3D 1.85 -> 1.49; CompCars 2.55 -> 2.35; Stanford Cars 1.92 -> 1.00Improved
Run more refinement iterations at inference (ablation)FocalPose · Pix3D and car datasetsMedErr_tImproved

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