hk-kaden-kim/Small_S2_CloudCover_Seg

Small Size Sentinel-2 Cloud Cover Segmentation Dataset

This dataset has a reduced size of samples from 'Sentinel-2 Cloud Cover Segmentation Dataset'. The original dataset has 11748 samples for training (26 GiB) and 10980 samples for test (24 GiB), which is hard to use for ML experiments under the limited resource and take more time to be loaded.
Therefore, I have randomly selected 5 % by each location for each dataset. As a result, this dataset has 585 sampes for training (1.29 GiB) and 547 samples for test (1.2 GiB).

Original Dataset

Sentinel-2 Cloud Cover Segmentation Dataset

Description

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, 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.

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