Emblinator backend

Emblinator is an image annotation tool, designed to make it easy to create annotated images for (among others) supervised machine learning tasks.

This repo hosts the backend for Emblinator. You should checkout the frontend to get both the frontend and the backend running.

Running
Setup

Run locally

Running application

export env variables:

. ./export_local_db_variables.sh

start Postgres in case it is stopped:

docker start emblinator-postgres

Run Flask server:

flask run

Run and build docker in development

fill docker.env_default and rename to docker.env
Create network for containers docker network create emblinator
Add postgres to network docker network connect emblinator emblinator-postgres
docker build -t annotator-backend .
Run dev server: docker run -it -p 5000:5000 --env-file docker.env --network emblinator annotator-backend flask run --host 0.0.0.0

Run and build docker in production

TODO

Important commands

Create db schema flask create-db
Create job in db flask insert-file-info

Run tests

Make sure you have postgres database annotator-test (createdb annotator-test)
Run py.test

Setup

Create python environment

Create virtualenvrionment for python. You should use Python 3.7.4.
Install imagemagick brew install freetype imagemagick
Install requirements pip install -r requirements.txt

Create GCP environment

Emblinator uses GCP storage buckets for image storage.

Set up Google storage (guide here)

your images should be sorted into folders like so:

  - folder1
      - image1_1
      - image1_2
      - ...
  - folder 2
      - image2_1
      - image2_2
      - ...

Set up Google Cloud Platform application credentials (guide here)
- you can create a /secrets/-folder for them, that folder is in .gitignore

Create database

export env variables:

. ./export_local_db_variables.sh

start a new postgres in docker:

docker run --name emblinator-postgres -d -e POSTGRES_DB=$POSTGRES_DB_NAME -e POSTGRES_PASSWORD=$POSTGRES_PASSWORD -e POSTGRES_USER=$POSTGRES_USER -p 5432:5432 postgres

setup DB:

flask create-db

Create database entities

Use flask insert-file-info to create the image and category DB entries. Run this before running server!

flask insert-file-info [OPTIONS] PREFIX JOB_NAME CATEGORIES

where

PREFIX is folder where images are located in bucket
JOB_NAME is used for ui listing in jobs
CATEGORIES Comma separated list of categories to be added for this job

For example:

#flask insert-file-info waterplants Waterplants water,plant

Run application:

flask run

josalmi/emblinator-back