AI Paper Atlas

Research map / 3D vision / Human pose estimation

CPPF++: Uncertainty-Aware Sim2Real Object Pose Estimation by Vote Aggregation

Yang You, Wenhao He, Jin Liu, Hongkai Xiong, Weiming Wang, Cewu Lu

TPAMI 2024 · 3D vision › Human pose estimation · approach: Category-level object pose estimation · 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
use 5-point tuples instead of point pairs (N=2) (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 28.7 -> 32.3Improved
increase tuple size from 5 to 10 (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 32.3 -> 32.2 (15deg 5cm: 85.2 -> 83.1)No clear effect
add uncertainty modeling of voting targets (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 24.2 -> 32.3Improved
add noisy pair filtering (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 12.6 -> 32.3Improved
add importance point re-weighting (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 31.0 -> 32.3Improved
add online alignment optimization (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 24.5 -> 32.3 (+7.8 AP@5deg 5cm)Improved
use ensemble of geometric and visual models instead of geometric only (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 19.8 -> 32.3Improved
use ensemble of geometric and visual models instead of visual only (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 22.1 -> 32.3Improved
use model-switching ensemble instead of concatenating geometric and visual features (ablation)CPPF++ · NOCS REAL275mAP 5deg 5cm: 18.2 -> 32.3Improved
use GroundedSAM detector instead of DualPoseNet masksCPPF++ · NOCS REAL275mAP 3D25 / 10deg 5cm: 3D25: 82.4 -> 72.7; 10deg 5cm: 65.9 -> 72.0Mixed
add ICP refinementCPPF++ · YCB-VideoADD(-S) AUC: 55.4 -> 70.0Improved
use zero-shot/CLIP instance detector instead of PoseCNN masksCPPF++ · YCB-VideoADD(-S) AUC: 55.4 -> 46.8Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 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