Research map / 3D vision / Human pose estimation
Learning Dynamics via Graph Neural Networks for Human Pose Estimation and Tracking
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 |
|---|---|---|---|
| replace IOU matching with GNN-predicted pose matching (ablation) | Pose tracking pipeline · PoseTrack 2017 validation | MOTA: 71.8 -> 73.1 | Helped |
| replace OKS matching with GNN-predicted pose matching (ablation) | Pose tracking pipeline · PoseTrack 2017 validation | MOTA: 72.1 -> 73.1 | Helped |
| add pose refinement by merging predicted and estimated pose heatmaps (ablation) | GNN pose matching pipeline · PoseTrack 2017 validation | mAP: 79.9 -> 81.1 | Helped |
| add pose refinement by merging predicted and estimated pose heatmaps (ablation) | GNN pose matching pipeline · PoseTrack 2017 validation | MOTA: 73.1 -> 73.4 | Helped |
| increase pose-history length from two to four frames (ablation) | GNN pose dynamics model · PoseTrack 2017 validation | mAP: 80.6 -> 81.3 | Helped |
| increase pose-history length from two to four frames (ablation) | GNN pose dynamics model · PoseTrack 2017 validation | MOTA: 72.9 -> 73.4 | Helped |
| halve the output dimension of the feature MLPs (ablation) | GNN pose dynamics model · PoseTrack 2017 validation | mAP: 81.1 -> 80.8 | Hurt |
| halve the output dimension of the feature MLPs (ablation) | GNN pose dynamics model · PoseTrack 2017 validation | MOTA: 73.4 -> 73.2 | No clear effect |
| use GNN pose dynamics and prediction alongside single-frame pose estimation | Pose estimation system · PoseTrack 2017 validation, pure pose estimation | mAP: 83.8 -> 84.9 | Helped |
| use GNN pose dynamics and prediction alongside single-frame pose estimation | Pose estimation and tracking system · PoseTrack 2018 validation, pose tracking | MOTA: 68.9 -> 69.2 | Helped |
Closest papers by meaning
- 3D Human Pose Estimation Using Spatio-Temporal Networks with Explicit Occlusion Training (AAAI 2020 · similarity 0.90)
- Deep Dual Consecutive Network for Human Pose Estimation (CVPR 2021 · similarity 0.90)
- Graph and Temporal Convolutional Networks for 3D Multi-person Pose Estimation in Monocular Videos (AAAI 2021 · similarity 0.89)
- PoseTrack: Joint Multi-Person Pose Estimation and Tracking (CVPR 2017 · similarity 0.89)
- GRPose: Learning Graph Relations for Human Image Generation with Pose Priors (AAAI 2025 · similarity 0.88)
- GCNext: Towards the Unity of Graph Convolutions for Human Motion Prediction (AAAI 2024 · similarity 0.87)
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.