AI Paper Atlas

Research map / 3D vision / Human pose estimation

MetaFuse: A Pre-trained Fusion Model for Human Pose Estimation

Rongchang Xie, Chunyu Wang, Yizhou Wang

CVPR 2020 · 3D vision › Human pose estimation · approach: Transformer-based multi-view 3D pose · 34 citations (OpenAlex)

Check 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 MetaFuse instead of NaiveFuse with 50 adaptation imagesCross-view pose fusion · PanopticMPJPE: $43.39$mm -> $37.27$mmImproved
Use MetaFuse instead of NaiveFuse with 200 adaptation imagesCross-view pose fusion · PanopticMPJPE: $35.60$mm -> $31.78$mmImproved
Use MetaFuse instead of NaiveFuse with 50 adaptation imagesCross-view pose fusion · PanopticJDR (Avg): 85.1 -> 88.9Improved
Use MetaFuse instead of No-FusionCross-view pose fusion · Total CaptureMPJPE (Mean): 41.2 -> 32.4Improved
Add multi-view fusion with 50 labeled examplesH36M 2D pose estimation · H36MJDRImproved
Increase MetaFuse adaptation data from 50 to 500 imagesH36M 3D pose estimation · H36MMPJPE: 32.7mm -> 29.3mmImproved
Add temporal information through LSTM-AE instead of MetaFuse3D pose estimation · Total Capture Walking2MPJPE: 21.7 -> 13.0Improved
Add Savitzky-Golay smoothing to MetaFuse 3D poses (ablation)MetaFuse · Total Capture WalkingMPJPE: about 5mmImproved
Use regular SGD pretraining instead of meta-learning for the factorized fusion model (ablation)AffineFuse · H36MJDRHurt
Replace geometry-inspired factorization with a NaiveFuse model, while retaining meta-learning (ablation)MetaFuse · H36MJDR (average): 87.7% and 89.3%Hurt
Use NaiveFuse with only 50 target images rather than MetaFuse (ablation)Cross-view pose fusion · H36MJDRHurt
Add L2, L1, or L2,1 regularization to NaiveFuse (ablation)NaiveFuse · H36MJDRNo 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