Research map / 3D vision / Human pose estimation
NeMo: Learning 3D Neural Motion Fields From Multiple Video Instances of the Same Action
Kuan-Chieh Wang, Zhenzhen Weng, Maria Xenochristou, João Pedro Araújo, Jeffrey Gu, Karen Liu, Serena Yeung
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 |
|---|---|---|---|
| Jointly optimize multiple video instances of the same action with NeMo | VIBE-initialized HMR on the MoCap dataset · MoCap dataset, mean dynamic range | MPJPE (mm): 120.5 mm to 95.5 mm | Improved |
| Use NeMo to jointly recover global root motion | Global HMR on the MoCap dataset · MoCap dataset, mean | Global-MPJPE (mm): 141 mm to 127 mm | Improved |
| Jointly optimize multiple video instances of the same action | HMR on Penn Action · Penn Action, mean | 2D reconstruction error (pixels): 11.2 to 8.1 | Improved |
| Apply SMPLify test-time optimization to VIBE | VIBE on the MoCap dataset · MoCap dataset, mean dynamic range | MPJPE (mm): 120.5 / 132.3 | Hurt |
| Add 3D supervision from initial HMR estimates (ablation) | NeMo on the MoCap dataset · MoCap dataset | MPJPE: 125.22 -> 77.96 | Improved |
| Add 3D supervision from initial HMR estimates (ablation) | NeMo on the MoCap dataset · MoCap dataset, dynamic range | MPJPE: 161.82 -> 95.7 | Improved |
| Add 3D supervision from initial HMR estimates (ablation) | NeMo on the MoCap dataset · MoCap dataset | 2D PCK: 98.4 -> 99.08 | Improved |
| Add instance-specific learnable parameters (ablation) | NeMo on the MoCap dataset · MoCap dataset | MPJPE: 134.85 -> 77.96 | Improved |
| Add instance-specific learnable parameters (ablation) | NeMo on the MoCap dataset · MoCap dataset, dynamic range | MPJPE: 194.6 -> 95.7 | Improved |
| Add instance-specific learnable parameters (ablation) | NeMo on the MoCap dataset · MoCap dataset | 2D PCK: 89.57 -> 99.08 | Improved |
Later papers that compared with this method
1 later accepted paper reports NeMo as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- NOVUM: Neural Object Volumes for Robust Object Classification (ECCV 2024 · better on 3 of 4 shared results)
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.
- VIBE — VIBE: Video Inference for Human Body Pose and Shape Estimation (CVPR 2020) · better on 4 of 4 shared results
- OpenPose — OpenPose: Realtime Multi-Person 2D Pose Estimation Using Part Affinity Fields (TPAMI 2019) · better on 4 of 4 shared results
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
- Mocap-2-to-3: Multi-view Lifting for Monocular Motion Recovery with 2D Pretraining (CVPR 2026 · similarity 0.89)
- Natural Human Motion Recovery by Aligning High-Order Temporal Dynamics from Monocular Videos (CVPR 2026 · similarity 0.88)
- AnyLift: Scaling Motion Reconstruction from Internet Videos via 2D Diffusion (CVPR 2026 · similarity 0.87)
- Uncertainty-Aware Human Mesh Recovery From Video by Learning Part-Based 3D Dynamics (ICCV 2021 · similarity 0.87)
- VideoMDM: Towards 3D Human Motion Generation From 2D Supervision (NeurIPS 2026 · similarity 0.86)
- HardMo: A Large-Scale Hardcase Dataset for Motion Capture (CVPR 2024 · similarity 0.86)
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.
