Research map / 3D vision / Human pose estimation
CIGPose: Causal Intervention Graph Neural Network for Whole-Body Pose Estimation
Bohao Li, Zhicheng Cao, Huixian Li, Yangming Guo
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Causal Intervention Module on top of the hierarchical GNN (ablation) | CIGPose-x · COCO-WholeBody val | whole-body AP: 66.8 -> 67.0 | Improved |
| add Causal Intervention Module on top of the hierarchical GNN (ablation) | CIGPose-l · COCO-WholeBody val | whole-body AP: 66.0 -> 66.3 | Improved |
| add hierarchical GNN (Gp and Gh) to RTMPose-x without intervention (ablation) | RTMPose-x · COCO-WholeBody val | whole-body AP: 65.3 -> 66.8 | Improved |
| add only the intra-part graph Gp (ablation) | RTMPose-x · COCO-WholeBody val | whole-body AP: 65.3 -> 66.3 | Improved |
| add only the Causal Intervention Module (no GNN) (ablation) | RTMPose-x · COCO-WholeBody val | whole-body AP: 65.3 -> 66.1 | Improved |
| add the inter-part hypergraph Gh to Gp (ablation) | RTMPose-x · COCO-WholeBody val | whole-body AP: 66.3 -> 66.8 | Improved |
| full CIGPose (CIM + GNN) versus RTMPose baseline | RTMPose-x · COCO-WholeBody val | whole-body AP: 65.3 -> 67.0 | Improved |
| add UBody training data | CIGPose-x · COCO-WholeBody val | whole-body AP: 67.0 -> 67.5 | Improved |
| add UBody training data | CIGPose-l (384x288) · COCO-WholeBody val | whole-body AP: 66.3 -> 66.9 | Improved |
| replace RTMPose-l with CIGPose-l at 384x288 | RTMPose-l · COCO val2017 | AP: 77.3 -> 78.5 | Improved |
| replace RTMPose-m with CIGPose-m | RTMPose-m · COCO val2017 | AR: 80.6 -> 79.3 | Hurt |
| replace RTMPose-m with CIGPose-m | RTMPose-m · CrowdPose test | AP: 70.6 -> 71.4 | Improved |
Closest papers by meaning
- Causal-Inspired Multitask Learning for Video-Based Human Pose Estimation (AAAI 2025 · similarity 0.86)
- Learning Causal Representation for Training Cross-Domain Pose Estimator via Generative Interventions (ICCV 2021 · similarity 0.86)
- Self-Constrained Inference Optimization on Structural Groups for Human Pose Estimation (ECCV 2022 · similarity 0.86)
- CleanPose: Category-Level Object Pose Estimation via Causal Learning and Knowledge Distillation (ICCV 2025 · similarity 0.85)
- Semantic Graph Convolutional Networks for 3D Human Pose Regression (CVPR 2019 · similarity 0.85)
- HiPoser: 3D Human Pose Estimation with Hierarchical Shared Learning at Parts-Level Using Inertial Measurement Units (AAAI 2025 · similarity 0.84)
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.
