ModuleNotFoundError: No module named 'surface_distance' when running train.py

[Jeevan-Neupane]

Description:

When trying to run train.py, the script fails with a ModuleNotFoundError due to the missing surface_distance package. This issue occurs in the XrayTo3DShape/utils/callbacks.py file.

Steps to Reproduce:

python train.py configs/paths/femur/30k/TotalSegmentor-femur-left-DRR-30k_train+val.csv \
    configs/paths/femur/30k/TotalSegmentor-femur-left-DRR-30k_test.csv \
    --tags model-compare --size 128 --batch_size 4 --accelerator cpu \
    --res 1.0 --model_name MultiScale2DPermuteConcat --epochs 100 \
    --loss DiceLoss --lr 0.002 --steps 4000 --dropout

The script fails with the following traceback:

Traceback (most recent call last):
  File "train.py", line 28, in <module>
    import XrayTo3DShape
  File "/home/jeevan-neupane/xrayto3D-benchmark/XrayTo3DShape/__init__.py", line 7, in <module>
    from .utils.registry import import_all_modules_for_register
  File "/home/jeevan-neupane/xrayto3D-benchmark/XrayTo3DShape/utils/__init__.py", line 11, in <module>
    from .callbacks import *
  File "/home/jeevan-neupane/xrayto3D-benchmark/XrayTo3DShape/utils/callbacks.py", line 22, in <module>
    from surface_distance.metrics import (
ModuleNotFoundError: No module named 'surface_distance'

System Information:

OS:Ubuntu
Python Version:3.8

[Jeevan-Neupane]

Inorder to solve this we have update benchmark-requirements-cpu.yaml to install surface-distance directly from github repo of
https://github.com/google-deepmind/surface-distance

Updated yaml file

name: xrayto3d
channels:
  - conda-forge
  - defaults
dependencies:
  - python=3.8
  - numpy
  - monai
  - torchvision
  - nibabel
  - itk
  - simpleitk
  - scikit-image
  - pillow
  - pandas
  - matplotlib
  - pytorch  # CPU version (no GPU)
  - wandb
  - tqdm
  - einops
  - pytorch-lightning
  - pip
  - git  # Required for cloning repositories
  - pip:
      - git+https://github.com/deepmind/surface-distance.git

[msrepo]

Please close this issue if it has been resolved and PR merged.