Flood Model

Flood Model is a neural network used to perform image segmentation to detect flood areas.

Getting started

Tutorial

If you want to try using this model, follow our tutorial to train the model on a dataset.

Development

Setup environment

First, clone the repository

git clone git@gitlab.develop.eoepca.org:sharinghub-test/flood-model.git

Then clone the dataset repository into the flood-model project

cd flood-model
git clone git@gitlab.develop.eoepca.org:sharinghub-test/sen1floods11-dataset.git

Configure dvc, then pull the dataset with dvc

cd sen1floods11-dataset
dvc pull
cd ..

You can setup your python environnment using poetry

poetry install --no-root

Do not forget to setup mlflow authentication Then you can run a train with poetry

poetry run python3 src/train.py

Your can also run an inference with a pre-trained model

poetry run python3 src/inference.py <model_path> <input_path>