apiquet / segmentation_from_satellite_images Goto Github PK

View Code? Open in Web Editor NEW

This repository shows how to get satellite images to build a dataset to train a neural network. It use the MiniFrance land cover dataset, Google-Earth-Engine to download satellite images, and Pytorch to train a neural network.

License: MIT License

Python 3.32% Jupyter Notebook 96.68%

segmentation_from_satellite_images's Introduction

Segmentation from satellite images

Description

Repository that implements a Deep Learning training using satellite images.

The following article explains:

the MiniFrance land cover dataset,
details about satellite data (TIF files, EPSG projections, etc.),
how to visualize satellite data on Google Maps through the QGIS software,
a description of the two satellites used Sentinel 1 and Sentinel 2,
how to get satellite images using Google Earth Engine.

How to use the code

By running the train.ipynb on Colab using the "Open in Colab" button at the top of the readme.

Or by running the train.ipynb or train.py on your own machine.

segmentation_from_satellite_images's People

Contributors

Watchers

segmentation_from_satellite_images's Issues

Pretrained model and minifrance dataset

@Apiquet ,Is there a file where we can use pre-trained weights for the model you developed? If there is no such file, I couldn't decide which files in the minifrance dataset I should download for training?

Recommend Projects