# Sentinel-2 Cloud Cover Segmentation Dataset
In many uses of multispectral satellite imagery, clouds obscure what we really care about - for example,
tracking wildfires, mapping deforestation, or monitoring crop health. Being able to more accurately remove
clouds from satellite images filters out interference, unlocking the potential of a vast range of use cases.
With this goal in mind, this training dataset was generated as part of [crowdsourcing competition](https://www.drivendata.org/competitions/83/cloud-cover/), and later
on was validated using a team of expert annotators. The dataset consists of Sentinel-2 satellite imagery 
and corresponding cloudy labels stored as GeoTiffs. There are 22,728 chips in the training data, 
collected between 2018 and 2020.

## Documentation
* [Link](https://radiantearth.blob.core.windows.net/mlhub/ref_cloud_cover_detection_challenge_v1/documentation.pdf)

## Tutorials
* [How to use deep learning, Pytorch Lightning, and the Planetary Computer to predict cloud cover in satellite imagery.](https://www.drivendata.co/blog/cloud-cover-benchmark/) by [Katie Westone](https://www.drivendata.co/)

## Creator & Contact
* [Radiant Earth Foundation](https://radiant.earth)
* ml@radiant.earth

## License
* [CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)

## Citation & DOI
Radiant Earth Foundation. (2022). Sentinel-2 Cloud Cover Segmentation Dataset (Version 1). Radiant MLHub. [Date Accessed] https://doi.org/10.34911/rdnt.hfq6m7