AI Paper Atlas

Research map / 3D vision / Human pose estimation

CosyPose: Consistent multi-view multi-object 6D pose estimation

Yann Labbé, Justin Carpentier, Mathieu Aubry, Josef Sivic

ECCV 2020 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 466 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
add color data augmentation (ablation)CosyPose single-view · T-LESS SiSorecall e_vsd<0.3: 37.0 -> 63.8Improved
add symmetric disentangled loss instead of point-matching loss (ablation)CosyPose single-view · T-LESS SiSorecall e_vsd<0.3: 60.1 -> 63.8Improved
use EfficientNet base network instead of FlowNet (ablation)CosyPose single-view · T-LESS SiSorecall e_vsd<0.3: 59.5 -> 63.8Improved
use 6D rotation parametrization instead of quaternions (ablation)CosyPose single-view · T-LESS SiSorecall e_vsd<0.3: 61.0 -> 63.8Improved
pretrain on synthetic data (ablation)CosyPose refinement · YCB-Video single-viewAUC of ADD(-S): approximately 2 pointsImproved
use CosyPose single-view estimator instead of DeepIMsingle-view pose · YCB-VideoAUC of ADD(-S): 81.9 -> 84.5Improved
use CosyPose instead of Pix2Pose on T-LESSsingle-view pose · T-LESS SiSorecall e_vsd<0.3: 29.5 -> 63.8Improved
add 5 views instead of 1 viewCosyPose multi-view · YCB-VideoAUC of ADD-S: 89.8 -> 93.4Improved
add 8 views instead of 1 viewCosyPose multi-view · T-LESS ViVomAP@ADD-S<0.1d: 55.0 -> 69.0Improved
add 4 views instead of 1 viewCosyPose multi-view · T-LESS ViVoe_vsd<0.3 recall: 62.6 -> 67.6Improved
add global object-level scene refinement (ablation)CosyPose multi-view · YCB-VideoADD-S error (mm): 6.40 -> 5.05Improved
add global object-level scene refinement (ablation)CosyPose multi-view · T-LESSADD-S error (mm): 4.43 -> 3.19Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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