AI Paper Atlas

Research map / 3D vision / Human pose estimation

SO-Pose: Exploiting Self-Occlusion for Direct 6D Pose Estimation

Yan Di, Fabian Manhardt, Gu Wang, Xiangyang Ji, Nassir Navab, Federico Tombari

ICCV 2021 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 166 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 the cross-layer 3D consistency loss L_cl-3D (ablation)SO-Pose · LMADD(-S) 0.02d: 41.6 -> 45.9Improved
add the cross-layer 3D consistency loss L_cl-3D (ablation)SO-Pose · LM2deg2cm: 67.4 -> 76.9Improved
add the cross-layer 2D consistency loss L_cl-2D (ablation)SO-Pose · LMADD(-S) 0.02d: 44.7 -> 45.9Improved
add the cross-layer 2D consistency loss L_cl-2D (ablation)SO-Pose · LM2deg2cm: 73.1 -> 76.9Improved
add both cross-layer consistency losses (ablation)SO-Pose · LMADD(-S) 0.02d: 36.6 -> 45.9Improved
add both cross-layer consistency losses (ablation)SO-Pose · LM2deg2cm: 59.1 -> 76.9Improved
add both cross-layer consistency losses (ablation)SO-Pose · LMADD(-S) 0.1d: 94.0 -> 96.0Improved
add the self-occlusion branch and L_cdpn to CDPN*CDPN* · LMADD(-S): 92.82 -> 94.77Improved
add the consistency loss L_cdpn on top of the self-occlusion branch in CDPN* (ablation)CDPN*-ours · LMADD(-S): 93.09 -> 94.77Improved
add the self-occlusion branch to CDPN* without L_cdpn (ablation)CDPN* · LM5deg5cm: 97.03 -> 97.06No clear effect
use ResNeSt50 instead of ResNet34 backbone (ablation)SO-Pose · YCB-VADD(-S): 54.6 -> 56.8Improved
use ResNeSt50 instead of ResNet34 backbone (ablation)SO-Pose · YCB-VAUC of ADD(-S): 82.3 -> 83.9Improved

Later papers that compared with this method

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

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