PointNet2 in conda environment

[Esteban-25]

Hi, I am trying to replicate the PointNet++ results by setting up an environment in conda with the program versions specified in this repository (TF 1.2, python2, etc). When installing tensorflow1.2 through conda, the corresponding CUDA and cuDNN packages are installed by default. However, I have CUDA 10.1 already installed in my computer. So my questions are:

1. How can I change the CUDA path in the .sh scripts to compile them properly? If I leave the path to my CUDA v10.1, it will not be compatible with TF 1.2.
2. How can I make sure that the CUDA version used for compiling isn't the one from my system but the one that came when I installed TF1.2 with conda?

Hope this makes sense and someone can help me. Cheers!

[irfan798]

If you are still interested i can share my conda environment with you And there are some changes to make it work with that environment #188

[chaudherysharif]

If you are still interested i can share my conda environment with you And there are some changes to make it work with that environment #188

Can you please share with me. I'll appreciate it. Thanks

[apfelbaumx]

Hi, I'm also interested in running this in a conda env. I'll appreciate if you could share it. Thx & stay safe

[tgaher]

Hi, Could you share this with me as well? Thanks a ton. Cheers.

[ElsheikhAhmed]

If you are still interested i can share my conda environment with you And there are some changes to make it work with that environment #188

Hallo, could you please share it with me ? Thanks alot!

[aneeta29]

@irfan798 Please I am also interested. Kindly share. Thank you.