AI Paper Atlas

Research map / 3D vision / Human pose estimation

DualPoseNet: Category-Level 6D Object Pose and Size Estimation Using Dual Pose Network With Refined Learning of Pose Consistency

Jiehong Lin, Zewei Wei, Zhihao Li, Songcen Xu, Kui Jia, Yuanqing Li

ICCV 2021 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 172 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 implicit pose decoder as auxiliary supervision (ablation)DualPoseNet · REAL275mAP IoU50 10deg 10%: 37.3 -> 43.1Improved
add implicit pose decoder as auxiliary supervision (ablation)DualPoseNet · REAL275mAP IoU75 5deg 10%: 19.7 -> 23.8Improved
add refined learning of pose consistency at test time (ablation)DualPoseNet · REAL275mAP IoU75: 60.1 -> 62.2Improved
add refined learning of pose consistency at test time (ablation)DualPoseNet · REAL275mAP 10deg 5cm: 64.2 -> 66.8Improved
use spherical fusion encoder instead of SCNN early fusion (ablation)DualPoseNet · REAL275mAP IoU50 10deg 10%: 35.5 -> 43.1Improved
use spherical fusion encoder instead of SCNN late fusion (ablation)DualPoseNet · REAL275mAP IoU50 10deg 10%: 41.7 -> 43.1Improved
use spherical fusion encoder instead of Densefusion (ablation)DualPoseNet · REAL275mAP IoU50 10deg 10%: 17.4 -> 43.1Improved
add implicit decoder (ablation)DualPoseNet · YCB-VideoADD-S AUC: 90.5 -> 91.2Improved
add refined learning of pose consistency (ablation)DualPoseNet · YCB-VideoADD-S AUC: 91.2 -> 93.3Improved
add iterative residual pose refinement (ablation)DualPoseNet · YCB-VideoADD-S AUC: 91.2 -> 95.0Improved
use a larger fine-tuning learning rate in pose consistency refinement (ablation)DualPoseNet · REAL275mAP IoU50 10deg 20%Mixed

Later papers that compared with this method

7 later accepted papers report DualPoseNet as a baseline in their results tables; 5 of them report a better number on most of the results they share with it.

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