AI Paper Atlas

Research map / 3D vision / Human pose estimation

Deep Bingham Networks: Dealing with Uncertainty and Ambiguity in Pose Estimation

Haowen Deng, Mai Bui, Nassir Navab, Leonidas J. Guibas, Slobodan Ilić, Tolga Birdal

IJCV 2022 · 3D vision › Human pose estimation · approach: Learned camera pose estimation · 59 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 Bingham-distribution loss (UBN) instead of L1 lossquaternion regression with PointNet · ModelNet10Chamfer distance (x10^2): 3.605 -> 1.477Improved
replace unimodal UBN by mixture of 10 Bingham components (MBN-10) (ablation)UBN · ModelNet10Chamfer distance (x10^2): 1.477 -> 0.838Improved
increase components from 10 to 25 and 50 (ablation)MBN · ModelNet10Chamfer distance (x10^2): 0.838 -> 0.926 -> 0.973Hurt
add explicit mixture coefficient learning and multi-hypothesis training (MBN-CE) instead of finite mixture MBN-MB (ablation)MBN-MB · Ambiguous relocalizationratio of correct poses (15deg/0.2m): 0.28 -> 0.56Improved
implicit mixture coefficient learning (MBN) versus MBN-CE with 10 hypotheses (ablation)MBN-CE · Ambiguous relocalizationratio of correct poses (15deg/0.2m): 0.56 -> 0.58Improved
use multiple-hypothesis Bingham mixture instead of MC-DropoutMC-Dropout · Synthetic round tablecorrectly detected modes: 24.8% -> 99.1%Improved
use multiple-hypothesis Bingham mixture instead of MC-DropoutMC-Dropout · Synthetic dining tablecorrectly detected modes: 50% -> 96%Improved
choose best branch by l1 distance rather than probability (ablation)MBN training · Ambiguous relocalizationratio of correct posesImproved
vary backbone (ResNet-18, 34, 50, Inception-v3) (ablation)MBN · Ambiguous relocalizationratio of correct poses (15deg/0.2m): 0.53 -> 0.57 -> 0.52 -> 0.50No clear effect
filter hypotheses by predicted uncertainty before registration (ablation)registration · point cloud registrationaverage recall: 0.777 -> 0.730No clear effect
unimodal UBN versus single-pose regressors on non-ambiguous scenesUBN · 7-Scenes Chessmedian rotation error: 4.48 -> 4.97No clear effect

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