Hello Great Work

[YUBO427]

Could you please tell me the version of scikit-image
I can not use this single_mesh.zip to install scikit-image

[mdfirman]

Thanks for your interest!

The single_mesh.zip requirement is needed to use a specific version of marching cubes which exports only a single mesh from a TSDF rather than a double mesh. You need this, otherwise you will end up with double-walled meshes which are likely to lead to bad planes when plane fitting. If you just want to try something then you could try just installing any standard version of scikit-image, but results from this would be unlikely to match what's in our paper.

Two questions which might help:

• Are you on a unix machine? I don't think we've tried our code on any platform other than Ubuntu.
• Are you following exactly the installation instructions in the readme?

Thanks

[mdfirman]

Are you getting the following error?

          from numpy.distutils.command.config import config as cmd_config
        File "/tmp/pip-build-env-yg_243jq/overlay/lib/python3.9/site-packages/numpy/distutils/command/config.py", line 19, in <module>
          from numpy.distutils.mingw32ccompiler import generate_manifest
        File "/tmp/pip-build-env-yg_243jq/overlay/lib/python3.9/site-packages/numpy/distutils/mingw32ccompiler.py", line 28, in <module>
          from distutils.msvccompiler import get_build_version as get_build_msvc_version
      ModuleNotFoundError: No module named 'distutils.msvccompiler'
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: subprocess-exited-with-error

× Getting requirements to build wheel did not run successfully.
│ exit code: 1
╰─> See above for output.

note: This error originates from a subprocess, and is likely not a problem with pip.
                                                                                                                                failed

CondaEnvException: Pip failed

I think I can reproduce this on Ubuntu.

It looks like numpy has deprecated distutils.

There are several ways this could be fixed, but we probably don't have time to try these ourselves before CVPR. For example we could imagine

• Pinning more of the requirements e.g. numpy and distutils (I tried pinning numpy to be <1.24 but this didn't fix it)
• Bringing the single_mesh.zip scikit-image fork to be up to date with scikit-image main branch (this is probably the right approach here)