Global Initiative

Tree species diversity,
from above.

Building the first global database for tree species annotation using high-resolution aerial images and deep learning.

Open Map Contribute Dataset
100+ Datasets
250K Annotations
5 Continents
Tree species annotations from aerial imagery

Catalogue

Our products, ready to explore.

Tree species

Broadleaf basal-area share across the Swiss forest area Conifer basal-area share across the same area

Tree species composition

A 30 m national map resolves continuous basal-area composition of main tree species in Switzerland.
Same view with the annotated coverage areas shown

Individual tree annotations

Individual tree crowns annotated by species across Swiss sites, paired with high-resolution aerial imagery.
Per-tree species detections over an Austrian forest stand Per-tree species detections over a neighbouring stand

Austrian tree species

Species-level annotations across the Eisenwurzen forests in Austria, built for training and benchmarking vision models.

Forest health

Detected standing dead trees outlined on the canopy Deadwood share aggregated into a grid

Swiss standing deadwood

Nationwide mapping of standing dead trees, a key indicator of forest health and biodiversity.
Swiss vegetation condition, first year Swiss vegetation condition, second year Swiss vegetation condition, third year

Vegetation condition

Monthly Sentinel-2 indicator showing where Swiss vegetation is doing better or worse than its long-term average.
Swiss species condition, first summer Swiss species condition, second summer Swiss species condition, third summer

Swiss species condition

Tree species distribution crossed with monthly vegetation condition: where each species grows, and how it is doing there.
Dead tree detections in Sihlwald, fourth survey year Dead tree detections, latest survey year Dead tree detections, earliest survey year Dead tree detections, second survey year Dead tree detections, third survey year

Dead trees (2014-2024)

Deep-learning detections of dead trees in Sihlwald across a decade of aerial imagery, filterable by confidence and size.
Partner access

Railway corridor watch

Forested stretches beside railway tracks, ranked by vegetation condition to plan inspections.
Coming soon

Forest fire

What we're building

01

Global Database

A global database for tree species and tree mortality annotation paired with high-resolution aerial and drone imagery from sites worldwide.

02

Computer Vision Models

Robust and generalizable deep learning models for individual tree species identification in sub-meter resolution aerial imagery.

03

Biodiversity Monitoring

Advancing forest mapping and biodiversity monitoring through AI-based solutions that scale across continents and ecosystems.

Data Call

400 species so far.
Add the next one.

Each point is a species already in the database. The nearer the centre, the more datasets record it. The open rim is where new species land.

Join the Data Call Read the data preparation guide
—Species so far
—Genera
—Datasets

Academic & Research Partners

Attribution

How to Cite

If you use the TreeAI dataset or any contributing sub-dataset in your research, please cite the following.

APA

Full record, version history, and file downloads on Zenodo.