Issues while running setup.py about pointnet2_pytorch HOT 2 OPEN

Hi, I had the same problem and changed the line:
_mp_ctx = multiprocessing.get_context("forkserver")
to:
_mp_ctx = multiprocessing.get_context()

I changed it directly in the installed path, in your case:
File "C:\Users\amanja\AppData\Local\Continuum\anaconda3\envs\pytorch\lib\site-packages\etw_pytorch_utils\persistent_dataloader.py", line 20
(this will require admin privelege).
The forkserver is only available for linux, and it should only affect the performance so it should be fine otherwise.

I also had an issue with downloading the data on Windows - I think curl doesn't handle the backslashes very well. I manually downloaded the file in ModelNet40Loader.py ("https://shapenet.cs.stanford.edu/media/modelnet40_ply_hdf5_2048.zip") and unzipped it into pointnet2\data\modelnet40_ply_hdf5_2048\ folder.

One last thing I needed to change for the example to work is the number of points which should be 2048, not 4096 (which is the default when the number of points parameter is not specified).

Hope that helps!

from pointnet2_pytorch.

Hi.

Thanks for the help. I figured the same and did the necessary changes.

from pointnet2_pytorch.