ClassificationApp back-end service

A server site support for ClassificationApp.

In order to run the back-end service, a Geopedia or equivalent database with tables as defined in schemas.py needs to be created. Get in touch if you are interested in reproducing the results using Geopedia.

A valid Sentinel Hub instance ID is required to retrieve Sentinel imagery. The instance ID need be added to the data/input_sources.json file.

How to run the service

There are multiple ways how to run the service:

Running development server locally (using pip)

(Required: Python >= 3.5)

• Clone git repository, move to the main project folder and install the project as a normal Python package. The following will install it in editable mode:

pip install -e . --upgrade

(On Windows some dependency packages might raise an error during installation. Install them from wheels repo.)

• Start running the service with:

python main.py

Running development server locally (using pipenv)

(Required: Python >= 3.5)

• Clone git repository, move to the main project folder and install packages:

pipenv install

• Enter virtual environment:

pipenv shell

• Start the service:

python main.py

Running production server locally

(Required: docker)

• Clone git repository, move to the main project folder and create a docker image:

docker build . --tag=class-service

• Run the image in a container:

docker run -p 5000:80 class-service

(The service will be available at port 5000)

Development

Service end points

The running service will produce a swagger documentation at /docs (e.g locally that is 'http://127.0.0.1:5000/docs').

For more details check ./classification-service/service.py. There you will also see example calls with curl.