search

jhennies / amst

Alignment to Median Smoothed Template for FIB-SEM data
GNU General Public License v3.0
8 stars 2 forks source link

Installation issue #5

Open haesleinhuepf opened 2 years ago

haesleinhuepf commented 2 years ago

Hi @jhennies ,

I'm having some installation issues on my Windows machine. This command

pip install amst_bin_win-0.1.2-cp36-cp36m-win_amd64

Produces this error message:

ERROR: Could not find a version that satisfies the requirement amst_bin_win-0.1.2-cp36-cp36m-win_amd64 (from versions: none)
ERROR: No matching distribution found for amst_bin_win-0.1.2-cp36-cp36m-win_amd64

I'm running on Windows 10 64-bit with Python 3.6.15 installed in a conda environment.

When upacking the wheel I noticed that inside are not soo many files. But I failed installing them separately using pip because there is no setup.py inside. I was just wondering how you made that wheel? Can you point me to the source code repository and build pipeline for it? I might then be able to install it myself.

Thanks!

Best, Robert

jhennies commented 2 years ago

Hi Robert,

you can try installing it manually with conda instead:

conda create -n amst-env -c conda-forge tifffile scikit-image=0.17.2 vigra silx[full] pyopencl
conda activate amst-env

Pyelastix was a bit of an issue for me, so I ended up installing it afterwards

conda install -c conda-forge pyelastix

If it causes trouble, there is also the pip version

pip install pyelastix

Last but not least, to make the GPU accessible

conda install -c conda-forge ocl-icd-system

Or instead to run SIFT components on the CPU instead

conda install -c conda-forge pocl

I hope this works for you!

If you specifically want the wheel to work, I would have to forward this issue to @josemiserra

Best, Julian

haesleinhuepf commented 2 years ago

Hi @jhennies ,

thanks for the fast feedback. It works like a charm! Just one question: Is there any particular reason why you're using scikit-image=0.17.2? I would like to combine amst with another project which needs scikit-image==0.18.3. Might it amst also work with the recent scikit-image? Or are there any known issues? I would love to hear what you think!

Thanks again, Robert

jhennies commented 2 years ago

Hi Robert,

great that it works now!

I am very sure that there was a reason to use the older version of scikit-image, I think I was using some deprecated function and didn't have the time to change that yet.

I can look into it. I'll keep you posted.

Cheers, Julian