FoundationPose icon indicating copy to clipboard operation
FoundationPose copied to clipboard

error when building conda environment

Open JZhao12 opened this issue 7 months ago • 8 comments

Hi Bowen,

I'm trying to reproduce your FoundationPose demo, however I have been building the environment for several days and still having errors at the last step:

CMAKE_PREFIX_PATH=$CONDA_PREFIX/lib/python3.9/site-packages/pybind11/share/cmake/pybind11 bash build_all_conda.sh

Could you please help check what caused the errors?I have tried many solutions from the issues related to GCC version, cuda version, eigen3 path, etc, and they did not work.

The full errors are too long, I have put them into a .txt file

error.txt

Many thanks, Jianyu

JZhao12 avatar Apr 10 '25 09:04 JZhao12

I had problem executing this command too, here is my log, thanks in advance. error.txt

Trinitro-Dopamine avatar Apr 10 '25 12:04 Trinitro-Dopamine

Hello guys,

I have finally finished setting up the conda environment, with an infinite number of tries😵If you encounter errors related to "ninja", "eigen", "gridencoder", "cpp", "mycuda", "cmake", etc. Just leave a message here and I may be able to help.

Best, Jianyu

JZhao12 avatar Jun 28 '25 08:06 JZhao12

I had problem executing this command too, here is my log, thanks in advance. error.txt

@Trinitro-Dopamine Hi mate, I think this is because your system cuda and pytorch cuda version mismatch, you may try to install the correct version and see what would happen.

JZhao12 avatar Jun 28 '25 08:06 JZhao12

I had problem executing this command too, here is my log, thanks in advance.This always show "No module named 'torch' ". I had tried many ways to deal with the problem, but it did not work.

error.txt

mrmeng2475 avatar Aug 22 '25 10:08 mrmeng2475

Hello guys,

I have finally finished setting up the conda environment, with an infinite number of tries😵If you encounter errors related to "ninja", "eigen", "gridencoder", "cpp", "mycuda", "cmake", etc. Just leave a message here and I may be able to help.

Best, Jianyu

when i am trying to run ’python bundlesdf/run_nerf.py --ref_view_dir /home/tiger/Downloads/Dataset/ref_views’, the following error occurs: File "/home/tiger/Projects/FoundationPose/bundlesdf/mycuda/torch_ngp_grid_encoder/grid.py", line 14, in import gridencoder ModuleNotFoundError: No module named 'gridencoder'

do you have any ideas about how to solve this?

langlang-02 avatar Nov 10 '25 12:11 langlang-02

已收到邮件!

mrmeng2475 avatar Nov 10 '25 12:11 mrmeng2475

Hello guys, I have finally finished setting up the conda environment, with an infinite number of tries😵If you encounter errors related to "ninja", "eigen", "gridencoder", "cpp", "mycuda", "cmake", etc. Just leave a message here and I may be able to help. Best, Jianyu

when i am trying to run ’python bundlesdf/run_nerf.py --ref_view_dir /home/tiger/Downloads/Dataset/ref_views’, the following error occurs: File "/home/tiger/Projects/FoundationPose/bundlesdf/mycuda/torch_ngp_grid_encoder/grid.py", line 14, in import gridencoder ModuleNotFoundError: No module named 'gridencoder'

do you have any ideas about how to solve this?

see solution here: https://github.com/NVlabs/FoundationPose/issues/98#issuecomment-2149020378

JZhao12 avatar Nov 11 '25 02:11 JZhao12

I had problem executing this command too, here is my log, thanks in advance.This always show "No module named 'torch' ". I had tried many ways to deal with the problem, but it did not work.

error.txt

you may try to use import torch torch.cuda.is_available() to check if torch is installed correctly. Just for reference, my torch and cuda version is 2.0.0+cu118.

JZhao12 avatar Nov 11 '25 02:11 JZhao12