installation

Clone the repository from https://github.com/ProjectAdA/ada-va

Move into the folder ada-va of the repository:

cd ada-va/ada-va

extractors

Build the Dockerfiles for the extractors:

docker-compose -f docker-compose.extractors.yml build

The imagesearch client is built with

docker build -f docker/Dockerfile_client_flask -t ada-va/client_flask:1.0 .

airflow

First create volumes for the cache and logs fro airflow:

mkdir -p ./volumes/ada-va-cache && mkdir -p ./volumes/airflow-logs

Refer to folder structure to see how the volume is structured. Initialize the database:

docker network create ada-rproxy
docker-compose -f docker-compose.yml up init

Start the airflow stack:

docker-compose -f docker-compose.yml up

Configuration

The docker-compose files is configured with three files.

• .env

• secrets.env

• ada-va.cfg

• airflow.env

.env sets the versions of python, the postgres database and airflow. The names of the extractor docker images are also set here.

secrets.env contains the credentials that docker-compose uses to set the users for the services used by ada-va (database, airflow ..). The values have to be filled out by hand and not be pushed onto the repository.

ada-va.cfg contains the default parameters for the extractors as well parameters shared by all extractors.

• adava_cache_dir

• docker_url

• shotdetection sensitivity

• deepfeatures frame_width