Skip to content

Latest commit

 

History

History
37 lines (37 loc) · 2.97 KB

old_install.md

File metadata and controls

37 lines (37 loc) · 2.97 KB

conda

  1. Create conda environment and install required packages according to their respective install instructions:
    • conda create -n twoppp37 python=3.7
    • conda activate twoppp37
    • conda install jupyter
    • DeepFly3D:
      • WARNING: because of a new version of DeepFly3D with incompatibilities to multiple other packages you have to install an older version as follows:
        • git clone https://github.com/NeLy-EPFL/DeepFly3D
        • cd DeepFly3D
        • git checkout JB_stable
        • This reverts the version back to an older version, which is compatible with this package.
        • pip install -e .
    • deepinterpolation: https://github.com/NeLy-EPFL/deepinterpolation/tree/adapttoR57C10 (install in current environment instead of making a new environment)
      • make sure you install the adapttoR57C10 branch and confirm that you are on it using git status. If not: use git checkout adapttoR57C10.
      • this might throw an error like: "ERROR: pip's dependency resolver does not currently take into account all the packages that are installed." ... for incompatibilities of numpy and scipy version, but this can be ignored.
    • conda install pandas
    • conda install numpy
    • utils2p: https://github.com/NeLy-EPFL/utils2p
    • ofco: https://github.com/NeLy-EPFL/ofco
      • this might again throw a dependency error about numpy and scipy, but it can be ignored.
    • utils_video: https://github.com/NeLy-EPFL/utils_video
    • df3dPostProcessing: https://github.com/NeLy-EPFL/df3dPostProcessing
      • use git clone https://github.com/NeLy-EPFL/df3dPostProcessing and pip install -e df3dPostProcessing if no install instructions are inside the repository
    • pip install behavelet
    • try whether the pandas instruction works:
      • type python -c "import pandas as pd; print(pd.__version__)"
      • If this returns just the version number, your're fine. If this returns an ImportError, this means that some of the installs messed up with the pandas/numpy dependencies. Fix the problem as follows:
      • pip uninstall pandas
      • pip uninstall numpy
      • conda install pandas
    • install fictrac according to the instructions here: https://github.com/rjdmoore/fictrac
      • make sure the executable "fictrac" is installed in this directory: "~/bin/fictrac/bin". This means you'll have to specify "~/bin/" as a basedir for installation.
  2. Finally, we can install the twoppp package:
    • clone repository: git clone https://github.com/NeLy-EPFL/twoppp
    • change directory: cd twoppp
    • install using pip: pip install -e .
      • if the install throws an error like "ImportError: cannot import name '_registerMatType' from 'cv2.cv2'", uninstall opencv and reinstall it. this is due to an opencv version incompatibility: pip uninstall opencv-python and pip install opencv-python