Research map / 3D vision / Human pose estimation
Graph and Temporal Convolutional Networks for 3D Multi-person Pose Estimation in Monocular Videos
Read the paperCheck 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 GCNs to the baseline (ablation) | Baseline pose estimator · MuPoTS-3D | PCK_abs: 29.8 -> 35.1 | Helped |
| Add TCNs to the baseline (ablation) | Baseline pose estimator · MuPoTS-3D | PCK_abs: 29.8 -> 38.7 | Helped |
| Use ground-truth bounding boxes instead of detected boxes (ablation) | Baseline pose estimator · MuPoTS-3D | PCK_abs: 29.8 -> 31.2 | Helped |
| Use a directed rather than undirected joint graph (ablation) | Joint GCN · MuPoTS-3D | PCK: 73.1 -> 76.8 | Helped |
| Add the bone GCN (ablation) | Directed joint GCN · MuPoTS-3D | PCK: 76.8 -> 82.6 | Helped |
| Add data augmentation (ablation) | Joint + Bone GCN · MuPoTS-3D | PCK: 82.6 -> 83.2 | Helped |
| Add velocity-TCN to joint-TCN (ablation) | Joint-TCN pose estimator · MuPoTS-3D | PCK: 86.2 -> 87.5 | Helped |
| Combine GCN and TCN modules (ablation) | Baseline pose estimator · MuPoTS-3D | PCK_abs: 29.8 -> 45.7 | Helped |
| Use the full method rather than the strongest reported 3DPW baseline | 3D pose estimation · 3DPW original test set | PA-MPJPE: 51.9 -> 64.2 | Hurt |
| Use the full method rather than the strongest reported 3DPW occlusion-subset baseline | 3D pose estimation · 3DPW occlusion subset | PA-MPJPE: 82.5 -> 85.7 | Hurt |
| Evaluate on the occlusion subset rather than the original test set | Our method · 3DPW | PA-MPJPE: 64.2 -> 85.7 | Hurt |
| Use the proposed method instead of the best reported baseline | 3D pose estimation · Human3.6M | MPJPE: 41.1 -> 40.9 | Helped |
Closest papers by meaning
- 3D Human Pose Estimation Using Spatio-Temporal Networks with Explicit Occlusion Training (AAAI 2020 · similarity 0.92)
- Monocular 3D Multi-Person Pose Estimation by Integrating Top-Down and Bottom-Up Networks (CVPR 2021 · similarity 0.91)
- Deep Dual Consecutive Network for Human Pose Estimation (CVPR 2021 · similarity 0.90)
- Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation From Monocular Video (CVPR 2022 · similarity 0.90)
- Monocular 3D Pose and Shape Estimation of Multiple People in Natural Scenes - The Importance of Multiple Scene Constraints (CVPR 2018 · similarity 0.90)
- CoMotion: Concurrent Multi-person 3D Motion (ICLR 2025 · similarity 0.88)
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.