AI Paper Atlas

Research map / 3D vision / Human pose estimation

PoseGAM: Robust Unseen Object Pose Estimation via Geometry-Aware Multi-View Reasoning

Jianqi Chen, Biao Zhang, Xiangjun Tang, Peter Wonka

CVPR 2026 · 3D vision › Human pose estimation · approach: Category-level object pose estimation · 0 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 point maps as cross-attention input (ablation)PoseGAM · Ablation subsampled dataAUC@3: 15.07 -> 25.94Improved
add learned geometry feature maps on top of point maps (ablation)PoseGAM · Ablation subsampled dataAUC@3: 25.94 -> 28.18Improved
add depth maps as an extra input (ablation)PoseGAM · Ablation subsampled dataAUC@3: 28.18 -> 28.75No clear effect
use farthest point sampling instead of random sampling for template views (ablation)PoseGAM · Ablation subsampled dataAUC@3: 14.23 -> 28.18Improved
increase template views from 5 to 10 (ablation)PoseGAM · Ablation subsampled dataAUC@3: 11.46 -> 28.18Improved
increase template views from 10 to 20 (ablation)PoseGAM · Ablation subsampled dataAUC@3: 28.18 -> 33.59Improved
initialize from VGGT weights (ablation)PoseGAM · Ablation subsampled dataAUC@3: 5.10 -> 28.18Improved
finetune PTv3 instead of freezing it (ablation)PoseGAM · Ablation subsampled dataAUC@3: 25.00 -> 28.18Improved
project PTv3 geometry features into view maps instead of raw tokens (ablation)PoseGAM · Ablation subsampled dataAUC@3: 23.88 -> 28.18Improved
inject view-map geometry features by cross-attention instead of direct addition (ablation)PoseGAM · Ablation subsampled dataAUC@3: 2.419 -> 28.18Improved
add texture color to VecSet geometry features (ablation)VecSet raw injection · Ablation subsampled dataAUC@3: 19.29 -> 22.85Improved
use PTv3 instead of VecSet geometry features (raw format) (ablation)PoseGAM · Ablation subsampled dataAUC@3: 19.29 -> 23.88Improved

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