search

facebookresearch / SparseConvNet

Submanifold sparse convolutional networks
https://github.com/facebookresearch/SparseConvNet
Other
2.04k stars 332 forks source link

voxel input #232

Closed HirojiFukuyama closed 2 years ago

HirojiFukuyama commented 2 years ago

Hi! Suppose I have a voxel file and I wanna do semantic segmentation on it. After I load the file and turn it into a 3D array (or tensor), how can I feed the 3D array into the network (let's say the scn.SubmanifoldConvolution module)? When I directly feed it in, the error indicates that Tensor doesn't have a feature attribute, so how can I convert this 3D array to adapt to your API? Thanks for any reply!

btgraham commented 2 years ago

An example of preparing the input it here: https://github.com/facebookresearch/SparseConvNet/blob/main/README.md#hello-world