Research map / 3D vision / Human pose estimation
Learning Local-Global Contextual Adaptation for Multi-Person Pose Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| add local-global contextual adaptation | center-offset pose estimation · COCO-val-2017 fully annotated subset | AP: 60.1 -> 70.0 | Helped |
| add OKS loss for learning local KAMs (ablation) | LOGO-CAP · COCO-val-2017 | AP: 60.0 -> 66.1 | Helped |
| add Gaussian reweighing (ablation) | LOGO-CAP · COCO-val-2017 | AP: 60.0 -> 67.6 | Helped |
| use both OKS loss and Gaussian reweighing (ablation) | LOGO-CAP with BatchNorm message passing · COCO-val-2017 | AP: 66.1 -> 69.0 | Helped |
| add Attentive Norm to the model with OKS loss and Gaussian reweighing (ablation) | LOGO-CAP · COCO-val-2017 | AP: 69.0 -> 69.6 | Helped |
| use 11×11 rather than 7×7 local KEMs (ablation) | LOGO-CAP with 3+1 CMP · COCO-val-2017 | AP: 68.4 -> 69.6 | Helped |
| use 11×11 rather than 15×15 or 19×19 local KEMs (ablation) | LOGO-CAP with 3+1 CMP · COCO-val-2017 | AP: 69.3 -> 69.6; 69.0 -> 69.6 | Helped |
| use 3+1 convolutional message passing layers rather than only the output convolution (ablation) | LOGO-CAP with 11×11 local KEMs · COCO-val-2017 | AP: 68.8 -> 69.6 | Helped |
| use KAMs together with KEMs rather than KEMs alone (ablation) | LOGO-CAP contextual adaptation · COCO-val-2017 | AP: 59.4 -> 69.6 | Helped |
| use KEMs together with KAMs rather than KAMs alone (ablation) | LOGO-CAP contextual adaptation · COCO-val-2017 | AP: 65.7 -> 69.6 | Helped |
| use KAMs rather than KEMs alone as the contextual adaptation prior (ablation) | LOGO-CAP · COCO-val-2017 | AP: 59.4 -> 65.7 | Helped |
| increase local KEM size from 11×11 to 15×15 or 19×19 (ablation) | LOGO-CAP · COCO-val-2017 | FPS: 20.7 -> 16.5; 20.7 -> 13.2 | Hurt |
Closest papers by meaning
- Simple Pose: Rethinking and Improving a Bottom-up Approach for Multi-Person Pose Estimation (AAAI 2020 · similarity 0.89)
- Towards Accurate Multi-Person Pose Estimation in the Wild (CVPR 2017 · similarity 0.88)
- FCPose: Fully Convolutional Multi-Person Pose Estimation With Dynamic Instance-Aware Convolutions (CVPR 2021 · similarity 0.88)
- MV-SSM: Multi-View State Space Modeling for 3D Human Pose Estimation (CVPR 2025 · similarity 0.86)
- Optimizing Human Pose Estimation Through Focused Human and Joint Regions (AAAI 2025 · similarity 0.86)
- ProbPose: A Probabilistic Approach to 2D Human Pose Estimation (CVPR 2025 · similarity 0.85)
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.