AI Paper Atlas

Research map / 3D vision / Human pose estimation

RoboTAG: End-to-end Robot Pose Estimation via Topological Alignment Graph

Yifan Liu, Fangneng Zhan, Wanhua Li, Haowen Sun, Katerina Fragkiadaki, Hanspeter Pfister

CVPR 2026 · 3D vision › Human pose estimation · approach: Monocular 2D-to-3D pose lifting

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 3D backbone via 2D-3D feature fusion (ablation)2D-only baseline · Panda 3C-AK, ORB, Photo, Kuka PhotoADD AUC (%): +0.45% averageImproved
add full TAG alignment with closed loops over feature fusion (ablation)2D-3D fusion · Panda 3C-AK, ORB, Photo, Kuka PhotoADD AUC (%): +1.6% averageImproved
add 2D/3D keypoint closed loops (ablation)RoboTAG without loops · Panda PhotoADD AUC (%): 82.0 -> 83.8Improved
add joint-angle closed loops (ablation)RoboTAG with keypoint loops · Panda ORBADD AUC (%): 76.0 -> 77.1Improved
add point-cloud closed loops (ablation)RoboTAG with keypoint and joint loops · Panda PhotoADD AUC (%): 84.4 -> 84.3No clear effect
use RoboTAG vs supervised-only methods on in-distribution DRRoboPEPP · Panda DRADD AUC (%): 83.0 -> 82.5Hurt
use RoboTAG on last joint J6RoboPEPP · Panda DRjoint angle deviation (deg): 5.4 -> 6.2Hurt
use RoboTAG vs Holistic Pose on averageHolistic Pose · DREAM averageADD AUC (%): 75.7 -> 76.9Improved

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