Following are platform dependent installation instructions
- Create a new conda environment.
- Clone the repository.
conda install --file requirements. txt
Following installation was tested on 4.9.140-tegra aarch64 GNU/Linux.
We use a python package manager to isolate the dependencies required for multisero. conda package manager is not directly supported by above version of Linux. Instead, pip virtual environment works well for package management.
Major steps in setup are:
-
Install virtual environment:
sudo apt install -y python3-venv -
Create an environment:
python3 -m venv ~/python-envs/multisero -
Activate environment:
source ~/python-envs/multisero/bin/activate -
Deactivate environment:
deactivate -
Update pyenv.cfg to use system-wide packages:
cd ~/python-envs/env/multiseroUse
nano pyvenv.cfg tochange:include-system-site-packages = true(allow packages that require lower-level installation to be accessed in the environment)
Some packages need to be installed in local environment and some system-wide.
Use pip install -I to install in the multisero site-packages folder, the environment will access the local pacakages first and then search for global packages.
scikit-image and scipy dependencies:
sudo apt-get install python-dev libfreetype6-dev
sudo apt-get install libfontconfig1-dev libjpeg
sudo apt-get install libblas-dev liblapack-dev libatlas-base-dev gfortran
pip install -I scipy==1.1.0
# compilation of scipy can take >5 min.
pip install -I scikit-image
# compilation of scikit-image takes >5min. May have to try sudo -H pip3 install scikit-image
pip install -I matplotlib
pip install -I pandas==0.24.0
# compilation of pandas can take time.
pip install -I wheel
pip install -I certifi
pip install -I cycler
pip install -I kiwisolver
pip install -I cython
pip install -I numpy==1.18.1
pip install -I python-dateutil
pip install -I pytz
pip install -I six
pip install -I tabulate
pip install -I tornado
pip install -I openpyxl
pip install -I xmltodict
pip install -I xlrd