AI Paper Atlas

Research map / 3D vision / Human pose estimation

A Graph-Based Approach for Category-Agnostic Pose Estimation

Or Hirschorn, Shai Avidan

ECCV 2024 · 3D vision › Human pose estimation · approach: Category-level object pose estimation · 31 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
replace decoder MLP FFN with graph-based FFN using the skeleton graphCapeFormer-T · MP-100 1-shotPCK@0.2: 86.25 -> 87.23Improved
replace decoder MLP FFN with graph-based FFN using the skeleton graphCapeFormer-T · MP-100 5-shotPCK@0.2: 90.90 -> 91.16Improved
swap ResNet-50 for SwinV2-T, remove keypoint positional encoding, lower-variance Gaussian keypoint mask (ablation)CapeFormer · MP-100 1-shotPCK@0.2: 85.31 -> 86.25Improved
use keypoint positional encoding (Support Keypoint Identifier) on original keypoint order (ablation)CapeFormer · MP-100PCK: 85.24 -> 89.26Improved
shuffle keypoint order at test time with keypoint positional encoding (ablation)CapeFormer · MP-100PCK: 89.26 -> 64.20Hurt
add graph FFN in place of positional encoding (ablation)CapeFormer without PE · MP-100PCK: 85.24 -> 87.71Improved
use a fully connected or edge-less graph instead of the pose graph (ablation)GraphCape · MP-100 1-shot Split 1PCK@0.2: 91.19 -> 89.35Hurt
use random graphs per instance instead of the pose graph (ablation)GraphCape · MP-100 1-shot Split 1PCK@0.2: 91.19 -> 81.49Hurt
use graph decoder when support or query images are partially masked (ablation)GraphCape · MP-100 split 1 maskingPCKImproved

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