Installation

There are two ways for installing the project:

  • install all the dependencies yourself.

  • use a Docker image to run the code.

The Docker image contains all the C dependencies and python packages, that can be sometimes very time consuming to install depending on your OS. So I recommend using the Docker image. The code is using some of the astromatic softwares that can be difficult to run on Windows. So, it is strongly advised to use the Docker image for Windows users.

Prerequisites

For both ways you will need to get the project first.

The sources for otrain can be downloaded from the Github repo.

git clone git://github.com/dcorre/otrain

Note that you will need to update the project regularly to check for updates. Ideally each time you want to use it, type the following command to search for updates:

git pull origin master

  • Or download:

    Simply download this tarball. Or through the console:

curl  -OL https://github.com/dcorre/otrain/tarball/master

Cloning the project allows to retrieve easily future code updates. If you downloaded the project, you will need to download it again to retrieve future updates.

Installation with Docker

The usage of a Docker allows to build an OS environment on your machine and thus avoid compatibility problems when running the code under Linux, Mac or Windows. If you have not Docker installed on your machine install it first.

sudo groupadd docker
sudo usermod -aG docker $USER

Log out and log back in so that your group membership is re-evaluated. For more information see https://docs.docker.com/engine/install/linux-postinstall/#manage-docker-as-a-non-root-user.

You can test that Docker is installed correctly and can be run without sudo:

docker run hello-world

Download the otrain Docker image

To retrieve the Docker image:

docker pull dcorre/otrain

Check that it appears in the list of images:

docker images

Installation without Docker

I advise to create a virtual environment to avoid messing with different python libraries version that could be already installed on your computer and required for other projects.

Install conda: https://docs.conda.io/en/latest/miniconda.html

You can also install everything with pip if you prefer not to use conda.

Python 3 environment:

conda create -n otrain python=3.8 numpy scipy matplotlib astropy h5py scikit-image

Activate the environment:

conda activate otrain

Install other libraries

Once you have activated the environment, install the packages that are not available with conda using pip:

python3 -m pip install keras tensorflow opencv-python-headless sklearn

Testing that it is working

Run Docker

Run the Docker image in the background:

docker run --name otrain -dit -v /path_to_your_data/:/home/newuser/data/ dcorre/otrain
This means that you run the docker image dcorre/otrain, and give the name otrain to the created container.
-d runs the container in backgound.
-i gives the possibility to enter in the container to run commands interactively in a bash terminal.
-t allocates a pseudo-TTY.
The -v option means that you mount a volume in the Docker pointing to a directory on your computer. This allows to exchange data between the Docker and your machine.
The volume is pointing to the directory containing your images on your machine. You need to edit the path before the :.

Once you have executed this command, you can run any command in the container by typing:

docker exec otrain ls
docker exec otrain pwd

to make a ls or a pwd in the container named otrain, or any other bash commands.

In the following, if you are using a Docker image just prepend the command `docker exec otrain` to run the given commands within the container instead of your machine.

The container is alive as long as you do not shut down your machine. It is important to know that you can not give the same name to two containers. So if for some reasons you need to remove the current container to start a new one, type:

docker rm otrain

You can list the containers, active or not, on your machine with:

docker ps -a

Install otrain

Inside the Docker image

Already installed.

Without Docker image

python3 setup.py develop

Run otrain on a test image

To test if otrain is running normally:

otrain-train -h

It should return you the list of accepted arguments for this executable.