Post-deforestation Land Use in the Tropics (PLUTo)

PLUTo is a dataset of post-deforestation land uses across the tropics. The corresponding publication "Mapping land uses following tropical deforestation with location-aware deep learning" is currently under review.

Contents

labels.csv

Columns:- 'driver'    - ground truth label- 'sampleid'    - unique identifier of every sample- 'country'    - which country the sample lies in- 'geometry'    - the coordinates of the center of the sample; each sample is 1x1km- 'split'    - whether sample was used for training or evaluation

tabular_data.csv

- 'sampleid'    - unique identifier of every sample- ['pasture', 'forest plantation', 'soy', 'coffee', 'palm oil', 'mining', 'urban', 'cocoa', 'rubber', 'shifting agriculture', 'other']    - processed and normalized values from the DeDuCE dataset (https://github.com/chandrakant6492/DeDuCE) 

deforestation_masks/- one file for each sample named {sampleid}.tif- location of deforestation within the image patch

images/- one folder for each sample named {sampleid}- contains 4 Sentinel-2 images covering the location