AI Paper Atlas

Research map / 3D vision / Human pose estimation

Dual Networks Based 3D Multi-Person Pose Estimation From Monocular Video

Yu Cheng, Bo Wang, Robby T. Tan

TPAMI 2022 · 3D vision › Human pose estimation · approach: Multi-person 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
add multi-person pose estimator to top-down network (ablation)top-down network · MuPoTS-3DPCK: 81.6 -> 87.5Improved
add combined heatmap to bottom-up network (ablation)bottom-up network · MuPoTS-3DPCK: 76.6 -> 78.0Improved
use learned integration network instead of hard integration (ablation)TD + BU · MuPoTS-3DPCK: 87.5 -> 88.9Improved
use learned integration network instead of linear integration (ablation)TD + BU · MuPoTS-3DPCK: 88.0 -> 88.9Improved
add pose matching to integration (ablation)TD + BU (IN) · MuPoTS-3DPCK: 85.5 -> 88.9Improved
add semi-supervised learning (ablation)TD + BU (IN) · MuPoTS-3DPCK: 88.9 -> 89.3Improved
add test time optimization (ablation)TD + BU (IN) · MuPoTS-3DPCK: 88.9 -> 89.2Improved
add interaction-aware discriminator to reprojection loss (ablation)semi-supervised learning · MuPoTS-3DPCK: 77.2 -> 89.1Improved
check pairs of poses (Dis2) instead of one pose (Dis1) in discriminator (ablation)inter-person discriminator · MuPoTS-3DPCK: 87.3 -> 89.1Improved
check three poses (Dis3) instead of two (Dis2) (ablation)inter-person discriminator · MuPoTS-3DPCK: 89.1 -> 89.0Hurt
add test time optimization regularizers and two-stage optimization (ablation)GCN baseline · Human3.6M (w/o GT 2D)MPJPE: 58.71 -> 49.31Improved
add reprojection regularization in test time optimization (ablation)GCN baseline · Human3.6M (w/o GT 2D)MPJPE: 58.71 -> 54.45Improved

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