search

mit-han-lab / pvcnn

[NeurIPS 2019, Spotlight] Point-Voxel CNN for Efficient 3D Deep Learning
https://pvcnn.mit.edu/
MIT License
644 stars 130 forks source link

Can I use just one step of voxelization and devoxelization operator to voxelize and devoxelize the point cloud? #74

Closed OswaldoBornemann closed 1 year ago

OswaldoBornemann commented 1 year ago

Can I use just one step of voxelization and devoxelization operator to voxelize and devoxelize the point cloud? Just like below

voxelize_ops = Voxelization(resolution=32,
                                         normalize=True,
                                         eps=0)
features = x
coords = x[:, :3, :]
voxel_x, voxel_coords = self.voxelize_ops(features, coords)
x = F.trilinear_devoxelize(voxel_x,
                                   voxel_coords,
                                   32,
                                   self.training)
zhijian-liu commented 1 year ago

Yes, you can voxelize and devoxelize the point cloud using the API we provided.

HLJT commented 1 year ago

Can I use just one step of voxelization and devoxelization operator to voxelize and devoxelize the point cloud? Just like below

voxelize_ops = Voxelization(resolution=32,
                                         normalize=True,
                                         eps=0)
features = x
coords = x[:, :3, :]
voxel_x, voxel_coords = self.voxelize_ops(features, coords)
x = F.trilinear_devoxelize(voxel_x,
                                   voxel_coords,
                                   32,
                                   self.training)

Can I ask if you implemented this way and got voxelized features