docker-compose.yaml

services:
  minio:
    image: minio/minio:RELEASE.2024-02-09T21-25-16Z
    container_name: minio
    environment:
      - MINIO_ROOT_USER=adminadmin
      - MINIO_ROOT_PASSWORD=adminadmin
    volumes:
      - ./minio_data:/data
    ports:
      - "31113:9001"
    command: server /data --console-address ":9001"

  mongo:
    image: mongo:4.2.19
    container_name: mongo
    environment:
      - MONGO_INITDB_ROOT_USERNAME=adminadmin
      - MONGO_INITDB_ROOT_PASSWORD=adminadmin
    volumes:
      - ./mongo_data:/data/db
    ports:
      - "31114:27017"

  landcoverpy:
    image: ghcr.io/khaosresearch/demo-landcoverpy:latest
    container_name: landcoverpy
    depends_on:
      - minio
      - mongo
    environment:
      - MINIO_HOST=minio
      - MONGO_HOST=mongo
      - MINIO_ACCESS_KEY=adminadmin
      - MINIO_SECRET_KEY=adminadmin
      - MONGO_USERNAME=adminadmin
      - MONGO_PASSWORD=adminadmin
      - GOOGLE_APPLICATION_CREDENTIALS=/app/data/gcloud-user.json # File can be generated in https://console.cloud.google.com/apis/credentials
      - MINIO_DATA_FOLDER_NAME=demo_landcoverpy # Name of the folder where the data is stored in Minio. Use it like a "project" name.
      - MIN_USEFUL_DATA_PERCENTAGE=30 # Minimum percentage of useful data to be used in the composite (0 - 100). A high number will increase the quality of the composite but will decrease the number of products to be used, which can lead to a worse result. Recommended value: 30
      - MAX_PRODUCTS_COMPOSITE=3 # Number of products to be used in the composite. A high number will increase the processing time and memory usage (>=2). Recommended value: 4
      - TILES_TO_PREDICT=prediction # Expected format `["NNLLL", "NNLLL", ...]` being N a number and L a letter. If you want to predict the tiles of the data provided for training, use `prediction`.
      - SUBCATEGORY_PREDICTION=["forest", "no_forest"] # Expected format `["category1", "category2", ...]` being the subcategories to predict. If you dont want to predict any subcategory, or you didnt provide any subcategory for training, use `[]`.
    volumes:
      - ./app_data:/app/data
    restart: on-failure:5
    command: python main.py

volumes:
  minio_data:
  mongo_data:
  app_data: