Need help with Installing Monailabel

[jankh0]

Hi Everyone,

I a researcher trying to use monailabel to segment the features of the neck including vasculature, tissue, trachea and muscle. I'm having significant problems with installation and have tried number of permutations including the below.

I am installing it on a windows 11 machine using terminal on VSCode.

1. https://projectweek.na-mic.org/PW37_2022_Virtual/Projects/MONAILabelLung/MONAILabel_Installation.html
2. https://www.youtube.com/watch?v=8y1OBQs2wis
3. https://discourse.slicer.org/t/how-to-start-with-monailabel-for-new-models/21063/65?page=4 - instructions.

I seem to keep running into dll errors, path errors and CUDA cores not detected on Torch despite updated drivers

I've tried Conda and when I am in a Conda environment, the files appear to be missing also.

Are there installation instructions elsewhere that may potentially help?

Kind Regards

Jan

[SachidanandAlle]

This is more of an environment/system issue for cuda rather than monailabel.

If you have cuda is correctly installed and you should be able to run nvidia-smi command successfully.

You can refer PyTorch installation guide to install torch with cuda support
https://pytorch.org/get-started/locally/

Cond/PIP or anything, if you can make python -c "import torch; print(torch.cuda.is_available())" working, you can install monailabel.

[jankh0]

Hi SachidanandAlle,

Thank you for taking the time to reply me.

I've done the below.

   pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu113
  
   python -c "import torch; print(torch.cuda.is_available())" 
  FALSE
  
   nvidia-smi 

    +-----------------------------------------------------------------------------------------+
    | NVIDIA-SMI 566.14                 Driver Version: 566.14         CUDA Version: 12.7     |
    |-----------------------------------------+------------------------+----------------------+
    | GPU  Name                  Driver-Model | Bus-Id          Disp.A | Volatile Uncorr. ECC |
    | Fan  Temp   Perf          Pwr:Usage/Cap |           Memory-Usage | GPU-Util  Compute M. |
    |                                         |                        |               MIG M. |
    |=========================================+========================+======================|
    |   0  NVIDIA GeForce RTX 4090 ...  WDDM  |   00000000:01:00.0  On |                  N/A |
    | N/A   45C    P8              8W /   81W |    2103MiB /  16376MiB |      4%      Default |
    |                                         |                        |                  N/A |
    +-----------------------------------------+------------------------+----------------------+

Perhaps I may not have installed the appropriate pytorch.

  pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

  Requirements satisfied. 

I'm not sure what else needs to be done for cuda availabe to return True.

Thanks again for your help.

[diazandr3s]

Hi @jankh0,

Thanks for sharing these details.
You may find this helpful: https://stackoverflow.com/questions/60987997/why-torch-cuda-is-available-returns-false-even-after-installing-pytorch-with

Taken from this post:

The system requirements to use PyTorch with CUDA are as follows:

1/ Your graphics card must support the required version of CUDA
2/ Your graphics card driver must support the required version of CUDA
3/ The PyTorch binaries must be built with support for the compute capability of your graphics card

[jankh0]

Hi @diazandr3s and @SachidanandAlle,

Thank you for the replies. I've managed to get monailabel started after some troubleshooting.

I'm not sure if I'm speaking out of turn but installation was not as straightforward and ChatGPT managed to get me past the problems.

Kind Regards

Jan