AI Paper Atlas

Research map / 3D vision / Human pose estimation

Temporal Representation Learning on Monocular Videos for 3D Human Pose Estimation

Sina Honari, Victor Constantin, Helge Rhodin, Mathieu Salzmann, Pascal Fua

TPAMI 2022 · 3D vision › Human pose estimation · approach: Monocular 2D-to-3D pose lifting · 21 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 decoder (image reconstruction) (ablation)DSL-STN · H36M 0.3% labelsN-MPJPE: 236.32 -> 185.75Improved
add latent split into time-variant/invariant (ablation)DSL-STN-Dec · H36M 0.3% labelsN-MPJPE: 185.75 -> 149.32Improved
add latent split (ablation)DSL-STN-Dec · H36M all labelsN-MPJPE: 127.52 -> 95.39Improved
use DSL contrastive loss instead of CSS loss (ablation)CSS-STN-Dec-Split · H36M 0.3% labelsN-MPJPE: 163.61 -> 149.32Improved
add contrastive loss over plain autoencoder (ablation)AE-STN-Dec · H36M all labelsN-MPJPE: 114.73 -> 95.39Improved
add STN to DSL without decoding (ablation)DSL · H36M 0.3% labelsN-MPJPE: 227.24 -> 236.32Hurt
add tracking loss (ablation)Ours · Diving 5% labelsN-MPJPE: 296.22 -> 265.09Improved
use single-view model instead of DrNetOurs · H36M 0.3% labelsN-MPJPE: 158.81 -> 149.32Improved
add 2D pose labels in feature extraction (ablation)Resynthesis · Diving diver-splitN-MPJPE (cm): 15.7 -> 14.1Improved
replace ResNet50 with ResNet101 (ablation)Resynthesis + 2D · 3DPWN-MPJPE (cm): 13.29 -> 10.38Improved
use resynthesis pre-training instead of ImageNet pre-training (ablation)ResNet50 + 2D · 3DPWN-MPJPE (cm): 18.07 -> 13.29Improved
single-view Ours vs multi-view NSDOurs · MPI-INF-3DHP 0.02% labels (MV-data)N-MPJPE: 256.34 -> 261.99Hurt

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