Archv

Archv is the python version of the archive-vision project: https://datalab.ucdavis.edu/archv/

Getting Started

Clone this repository

git clone https://github.com/datalab-dev/archv-py.git

Download anaconda

https://docs.anaconda.com/anaconda/install/

Setup

Install appropriate python version

Project requires python 3.8+ (see pyproject.toml)

I recommend using pyenv to manage python versions.

However, if you will use homebrew to install opencv you probably need to use homebrew to install an appropriate python version.

Using homebrew or pyenv to install appropriate version python

homebrew

brew install [email protected]

pyenv

Install pyenv: https://github.com/pyenv/pyenv-installer

Install an appropriate python version:

pyenv install 3.8.10
pyenv local 3.8.10

Install Poetry

from https://python-poetry.org/docs/:

run the install script:
curl -sSL https://raw.githubusercontent.com/python-poetry/poetry/master/get-poetry.py | python -

open shell and test with:

poetry --version

If the above gives you an error then you need to add to the config file to your PATH. Assuming that poetry is located in your $HOME directory, ie /User/bitnguyen/.poetry

export PATH="$HOME/.poetry/bin:$PATH"

Then rerun:

poetry --version

Install dependencies with poetry

Inside archv/ directory run:

poetry install

You should be good to go. You can now drop into the virtualenv by running poetry shell, or run an arbitrary command in the virtualenv without dropping into it by using poetry run <your_command>, for example poetry run python some_script.py

Want to add dependencies?
Simply run poetry add <package_name>. Don't forget to commit the resulting changes to pyproject.toml and poetry.lock!

Build Opencv and include cv2.so in site-packages

First find the location of the site-packages directory for this project.

poetry run python -m site

find the one in the default pypoetry virtualenv location

1. From Source

Follow instructions from this page: https://docs.opencv.org/4.5.2/dd/dd5/tutorial_py_setup_in_fedora.html

Then copy the cv2.so to site-packages directory from above

cp cv2.so <site-packages-dir>

2. With Homebrew

Create an environment in conda with python version 3.9:

conda create -n py39 python=3.9

Start the environment:

conda activate py39

Make sure you are in the environment:

* indicates which environment is active

conda info --envs

Your site packages should be in a similar directory like this:

/Users/bitnguyen/opt/anaconda3/envs/py39/python3.9/site-packages/

Install opencv with brew:

#in case there was a version built before installing brew python
brew uninstall opencv

brew install opencv

Copy the cv2 directory from

/usr/local/Cellar/opencv/4.5.3_2/lib/python3.9/site-packages/cv2

to your python site-packages:

cp /usr/local/Cellar/opencv/4.5.3_2/lib/python3.9/site-packages/cv2/ /Users/arthurkoehl/opt/anaconda3/envs/py39/lib/python3.9/site-packages/

Run

Afterward, run:

poetry build

#change version appropriately
pip3 install dist/archv-0.1.0-py3-none-any.whl