DuoMo: Dual Motion Diffusion for World-Space Human Reconstruction
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866917310573314048 |
|---|---|
| author | Wang, Yufu Ng, Evonne Shin, Soyong Khirodkar, Rawal Dong, Yuan Su, Zhaoen Park, Jinhyung Kitani, Kris Richard, Alexander Prada, Fabian Zollhofer, Michael |
| author_facet | Wang, Yufu Ng, Evonne Shin, Soyong Khirodkar, Rawal Dong, Yuan Su, Zhaoen Park, Jinhyung Kitani, Kris Richard, Alexander Prada, Fabian Zollhofer, Michael |
| contents | We present DuoMo, a generative method that recovers human motion in world-space coordinates from unconstrained videos with noisy or incomplete observations. Reconstructing such motion requires solving a fundamental trade-off: generalizing from diverse and noisy video inputs while maintaining global motion consistency. Our approach addresses this problem by factorizing motion learning into two diffusion models. The camera-space model first estimates motion from videos in camera coordinates. The world-space model then lifts this initial estimate into world coordinates and refines it to be globally consistent. Together, the two models can reconstruct motion across diverse scenes and trajectories, even from highly noisy or incomplete observations. Moreover, our formulation is general, generating the motion of mesh vertices directly and bypassing parametric models. DuoMo achieves state-of-the-art performance. On EMDB, our method obtains a 16% reduction in world-space reconstruction error while maintaining low foot skating. On RICH, it obtains a 30% reduction in world-space error. Project page: https://yufu-wang.github.io/duomo/ |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_03265 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | DuoMo: Dual Motion Diffusion for World-Space Human Reconstruction Wang, Yufu Ng, Evonne Shin, Soyong Khirodkar, Rawal Dong, Yuan Su, Zhaoen Park, Jinhyung Kitani, Kris Richard, Alexander Prada, Fabian Zollhofer, Michael Computer Vision and Pattern Recognition We present DuoMo, a generative method that recovers human motion in world-space coordinates from unconstrained videos with noisy or incomplete observations. Reconstructing such motion requires solving a fundamental trade-off: generalizing from diverse and noisy video inputs while maintaining global motion consistency. Our approach addresses this problem by factorizing motion learning into two diffusion models. The camera-space model first estimates motion from videos in camera coordinates. The world-space model then lifts this initial estimate into world coordinates and refines it to be globally consistent. Together, the two models can reconstruct motion across diverse scenes and trajectories, even from highly noisy or incomplete observations. Moreover, our formulation is general, generating the motion of mesh vertices directly and bypassing parametric models. DuoMo achieves state-of-the-art performance. On EMDB, our method obtains a 16% reduction in world-space reconstruction error while maintaining low foot skating. On RICH, it obtains a 30% reduction in world-space error. Project page: https://yufu-wang.github.io/duomo/ |
| title | DuoMo: Dual Motion Diffusion for World-Space Human Reconstruction |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2603.03265 |