Research map / 3D vision / Human pose estimation
PhaseMP: Robust 3D Pose Estimation via Phase-conditioned Human Motion Prior
Mingyi Shi, Sebastian Starke, Yuting Ye, Taku Komura, Jungdam Won
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 phase feature and dynamic test-time interpolation to the prior (ablation) | PhaseMP · i3DB video-to-motion | MPJPE: 30.49 -> 27.43 | Improved |
| use phase condition alone without dynamic interpolation (ablation) | PhaseMP · i3DB video-to-motion | P-MPJPE: 15.44 -> 18.33 | Hurt |
| add dynamic confidence-based interpolation to phase update (ablation) | PhaseMP with phase condition · i3DB video-to-motion | MPJPE: 31.75 -> 27.82 | Improved |
| add SirenMLP decoder (ablation) | PhaseMP · i3DB video-to-motion | MPJPE: 27.82 -> 27.43 | Improved |
| add phase conditioning (ablation) | HuMoR MLP prior · AMASS 5s sampling | FDE: 47.34 -> 35.73 | Improved |
| add phase feature to SirenMLP prior (ablation) | PhaseMP · AMASS 10s sampling | ADE: 35.47 -> 42.12 | Hurt |
| add phase feature with dynamic interpolation for noisy input | PhaseMP · Noisy 3D input, 12cm noise | All-joint error: 34.08 -> 9.42 | Improved |
| add phase prior constraint | PhaseMP · Fitting 3s, joints above 0.9m | Visible-joint error: 0.67 -> 3.94 (vs VPoser-t) | Hurt |
| PhaseMP versus HuMoR with end-effector-only input | PhaseMP · Fitting 3s, end effectors only | All-joint error: 3.83 -> 3.79 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- HuMoR — HuMoR: 3D Human Motion Model for Robust Pose Estimation (ICCV 2021) · better on 8 of 8 shared results
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- HuMoR: 3D Human Motion Model for Robust Pose Estimation (ICCV 2021 · similarity 0.89)
- TEMPO: Efficient Multi-View Pose Estimation, Tracking, and Forecasting (ICCV 2023 · similarity 0.88)
- Capturing Humans in Motion: Temporal-Attentive 3D Human Pose and Shape Estimation From Monocular Video (CVPR 2022 · similarity 0.88)
- POMP: Physics-consistent Motion Generative Model through Phase Manifolds (CVPR 2025 · similarity 0.88)
- Uncertainty-Aware Human Mesh Recovery From Video by Learning Part-Based 3D Dynamics (ICCV 2021 · similarity 0.88)
- MotionMap: Representing Multimodality in Human Pose Forecasting (CVPR 2025 · 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.
