Research map / 3D vision / Human pose estimation
Recurrent 3D Pose Sequence Machines
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 |
|---|---|---|---|
| use recurrent multi-stage pose refinement | RPSM · Human3.6M | 3D pose error (mm), Avg.: 93.15 -> 73.10 | Helped |
| use RPSM instead of the strongest listed baseline | 3D pose estimation · HumanEva-I Walking | 3D pose error (mm), Avg.: 36.6 -> 28.4 | Helped |
| use RPSM instead of the strongest listed baseline | 3D pose estimation · HumanEva-I Jogging | 3D pose error (mm), Avg.: 38.9 -> 33.2 | Helped |
| use RPSM instead of Tekin et al. | 3D pose estimation · HumanEva-I Boxing | 3D pose error (mm), Avg.: 56.6 -> 52.8 | Helped |
| add a second refinement stage (ablation) | RPSM · Human3.6M | 3D pose error (mm), Avg.: 80.15 -> 75.55 | Helped |
| add a third refinement stage (ablation) | RPSM · Human3.6M | 3D pose error (mm), Avg.: 75.55 -> 73.10 | Helped |
| add three-stage refinement while omitting temporal modeling (ablation) | RPSM · Human3.6M | 3D pose error (mm), Avg.: 90.83 -> 81.12 | Helped |
| add temporal modeling with five-frame clips (ablation) | single-stage RPSM · Human3.6M | 3D pose error (mm), Avg.: 90.83 -> 80.15 | Helped |
| use ten-frame rather than five-frame clips (ablation) | single-stage RPSM · Human3.6M | 3D pose error (mm), Avg.: 80.15 -> 82.88 | Hurt |
| remove MPII pretraining of the 2D pose module (ablation) | RPSM-3-stage · Human3.6M | 3D pose error (mm), Avg.: 73.10 -> 111.3 | Hurt |
| share all 2D pose module layers across stages (ablation) | RPSM-3-stage · Human3.6M | 3D pose error (mm), Avg.: 73.10 -> 71.44 | Helped |
Closest papers by meaning
- Direct Prediction of 3D Body Poses From Motion Compensated Sequences (CVPR 2016 · similarity 0.89)
- Coarse-To-Fine Volumetric Prediction for Single-Image 3D Human Pose (CVPR 2017 · similarity 0.89)
- PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation (CVPR 2023 · similarity 0.88)
- Learning 3D Human Dynamics From Video (CVPR 2019 · similarity 0.88)
- PoseMamba: Monocular 3D Human Pose Estimation with Bidirectional Global-Local Spatio-Temporal State Space Model (AAAI 2025 · similarity 0.87)
- ManiPose: Manifold-Constrained Multi-Hypothesis 3D Human Pose Estimation (NeurIPS 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.