Dataset: A remote sensing-based solution for national scale tree counting

Outputs (results) of tree counts in Rwanda, Tanzania, Burundi, Kenya, and Uganda, using Deep Learning and satellite images: Planet, Sentinel-1, and Sentinel-2. The dataset is published as part of the study "Mugabowindekwe,  Gominski, D., Reiner, F., Tong, X., Ciais, P., Nyandwi, E., Gasangwa, I., Hategekimana, Y., Elikana, J., Main, R., Laurin, G. V., Fensholt, R. &  Brandt, M. A remote sensing-based solution for national scale tree counting. Journal of Remote Sensing. (2026). Data description: 1. The tree count files are 100m resolution GeoTIFF files with a geographic coordinate reference system in EPSG: 4326 containing the tree count within a ha (one pixel is 100 m by 100 m). The dataset was predicted using 20-band image composites composed of 4 bands from Planet, 10 bands from Sentinel-2, and 6 layers of Sentinel-1 (i.e. 25th, 50th, and 7th percentiles of VV and VH polarisations) collected in 2019. The 20-band images were resampled to a common 3-m resolution (consistent with Planet bands). The predicted tree count ranges from 0, which indicates no trees or no data.Notes:1. The data is provided per country, where you find zipped files for Rwanda, Tanzania, Burundi, Kenya, and Uganda. 1. To visualise the data, first download the zipped file for a country of interest to a local directory, extract the contained GeoTIFF files, then load them into a GIS or image processing software package. For queries related to the dataset, please email Maurice Mugabowindekwe at mmu@ign.ku.dk.