Datasets:
The Dataset Viewer has been disabled on this dataset.
Leaves.PH: Metro Manila tree-canopy series
Open per-LGU and per-barangay annual tree-canopy estimates for the 17 LGUs of Metro Manila (2019 to 2026), measured from Sentinel-2 imagery with a human-calibrated classifier. Canopy here is a dense-vegetation proxy, not a pixel-exact tree census. Read each year as a cross-sectional snapshot, not a validated change series.
What is in this repo
per_lgu/per_lgu_canopy_2019_2026.csv,per_barangay/per_barangay_canopy_2019_2026_model.csv: the published canopy series.MODEL_CARD.md,BENCHMARKS.md,RESULTS.md: accuracy, intended use, known biases, and the 656-label evaluation. All metrics live here.canopy_model/: the published classifier, its 656 manual gold labels, and the ablation it was chosen from.validation_v3/*.png: per-LGU visual validation panels.
The CLIP detection artefacts are a separate research track (in optimization) and are not the source of any published figure.
Links
- Code and full methodology: https://github.com/xmpuspus/leaves-ph
- Interactive map: https://leaves.ph
- Model repository: https://huggingface.co/xmpuspus/leaves-ph
- DOI: https://doi.org/10.5281/zenodo.20470306
License
Data: CC-BY-4.0. Code (in the GitHub repository): MIT. Upstream inputs retain their own licenses (Copernicus Sentinel-2, Hansen GFC, ESA WorldCover, Google Dynamic World, Meta Canopy Height, OpenStreetMap, PSA); see the repository LICENSE for the required attribution string.
Citation
@software{puspus_leaves_ph_2026,
title = {Leaves.PH: open tree-canopy measurement for Metro Manila},
author = {Puspus, Xavier},
year = {2026},
version = {0.7.0},
doi = {10.5281/zenodo.20470306},
url = {https://github.com/xmpuspus/leaves-ph}
}
- Downloads last month
- 73