Robust Human Registration with Body Part Segmentation on Noisy Point Clouds
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866909565498425344 |
|---|---|
| author | Lascheit, Kai Barath, Daniel Pollefeys, Marc Guibas, Leonidas Engelmann, Francis |
| author_facet | Lascheit, Kai Barath, Daniel Pollefeys, Marc Guibas, Leonidas Engelmann, Francis |
| contents | Registering human meshes to 3D point clouds is essential for applications such as augmented reality and human-robot interaction but often yields imprecise results due to noise and background clutter in real-world data. We introduce a hybrid approach that incorporates body-part segmentation into the mesh fitting process, enhancing both human pose estimation and segmentation accuracy. Our method first assigns body part labels to individual points, which then guide a two-step SMPL-X fitting: initial pose and orientation estimation using body part centroids, followed by global refinement of the point cloud alignment. Additionally, we demonstrate that the fitted human mesh can refine body part labels, leading to improved segmentation. Evaluations on the cluttered and noisy real-world datasets InterCap, EgoBody, and BEHAVE show that our approach significantly outperforms prior methods in both pose estimation and segmentation accuracy. Code and results are available on our project website: https://segfit.github.io |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2504_03602 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Robust Human Registration with Body Part Segmentation on Noisy Point Clouds Lascheit, Kai Barath, Daniel Pollefeys, Marc Guibas, Leonidas Engelmann, Francis Computer Vision and Pattern Recognition Registering human meshes to 3D point clouds is essential for applications such as augmented reality and human-robot interaction but often yields imprecise results due to noise and background clutter in real-world data. We introduce a hybrid approach that incorporates body-part segmentation into the mesh fitting process, enhancing both human pose estimation and segmentation accuracy. Our method first assigns body part labels to individual points, which then guide a two-step SMPL-X fitting: initial pose and orientation estimation using body part centroids, followed by global refinement of the point cloud alignment. Additionally, we demonstrate that the fitted human mesh can refine body part labels, leading to improved segmentation. Evaluations on the cluttered and noisy real-world datasets InterCap, EgoBody, and BEHAVE show that our approach significantly outperforms prior methods in both pose estimation and segmentation accuracy. Code and results are available on our project website: https://segfit.github.io |
| title | Robust Human Registration with Body Part Segmentation on Noisy Point Clouds |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2504.03602 |