AI Paper Atlas

Research map / 3D vision / Human pose estimation

Lite Pose: Efficient Architecture Design for 2D Human Pose Estimation

Yihan Wang, Muyang Li, Han Cai, Wei-Ming Chen, Song Han

CVPR 2022 · 3D vision › Human pose estimation · approach: Transformer-based multi-view 3D pose · 132 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
Replace EfficientHRNet-H_-1 with LitePose-SBottom-up pose estimation · CrowdPose testAPImproved
Replace Lightweight OpenPose with LitePose-SBottom-up pose estimation · COCO valAP: +14.0APImproved
Shrink the depth of high-resolution branches (ablation)Scaled-HigherHRNet-W16 · Low-computation pose estimationAPImproved
Add the fusion deconv head (ablation)LitePose · CrowdPoseAP: +7.6APImproved
Increase kernel size from 3x3 to 7x7 (ablation)LitePose · CrowdPoseAP: +13.0APImproved
Use a 9x9 kernel instead of 7x7 (ablation)LitePose · CrowdPoseAPHurt
Add one-shot supernet training (ablation)0.5 LitePose · CrowdPoseAP: +1.4APImproved
Add one-shot supernet training (ablation)LitePose-XS · CrowdPoseAP: +2.7APImproved
Apply architecture search with fine-tuning (ablation)0.5 LitePose · CrowdPoseAP: +2.2APImproved
Add teacher-based heatmap distillation (ablation)LitePose-XS · CrowdPoseAP: +1.1APImproved

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