facebookresearch / facebookresearch/pytorch3d
Batched convex hull on GPU
- Dominant language
- Python
- Stars
- 10k
- Forks
- 1.5k
- PR merge metrics
- No merged PRs in 30d
Description
## 🚀 Feature
Batched Convex Hull on GPU
## Motivation
It would be great if I could compute batched point cloud convex hull directly on GPU.
I'm always frustrated when I have to loop through the batch and compute convex hull one-by-one on CPU using [trimesh.convex](https://trimsh.org/trimesh.convex.html) or [scipy.spatial.convexhull](https://docs.scipy.org/doc/scipy/reference/generated/scipy.spatial.ConvexHull.html).
## Pitch
Convex hull could be useful for point cloud segmentation, part features learning and more.
One of the open source method is [ghull](https://www.comp.nus.edu.sg/~tants/gHull.html); hope that it can be implemented in pytorch3d.
Thanks.
Contributor guide
Assessment
This issue has not been assessed yet.