land_cover_tracking

This is a Web APP for land cover tracking and simple analysis using satellite data.

Data is downloaded using integrated sentinelhub API. Area of interest, as well as dates range, can be selected using the GUI.

Prediction is made using the model developed in this repository.

Predicted classes:

• shrublands
• urban
• agricultural
• barren
• snow
• water
• dense_forest
• open_forest
• wetlands
• grasslands

Environment

Poetry

1. Install poetry with pip install poetry
2. Create environment with poetry install
3. Enter environment with poetry shell

Docker container

1. Create container with ./docker/setup_docker
2. Enter your container with docker exec -it land_cover_app /bin/bash

Raw environment

1. Install with pip install -r ./docker/requirements.txt

App

1. Setup SentinelHub config file sentinelhub_config.json, according to this guide
2. Download weigths from here and put it to weights folder (more weigths coming)
3. To start, in your environment, please run python app.py.

Demo:

Some land cover comparisons