Leucanthemum vulgare observed in Poland by longerjwb, CC BY-SA. View on this portal

About the data

What Pl@ntNet publishes on GBIF, how records are selected, and how to reuse and cite them

Pl@ntNet in one paragraph

Pl@ntNet is a participatory platform and information system dedicated to monitoring plant biodiversity through image-based identification. It was founded in 2010 by a consortium of four French research organisations — CIRAD, INRAE, Inria and IRD — and is now open to other members. Anyone can photograph a plant with the mobile app or the web interface, get an identification suggested by a deep-learning model, and share the observation with the community, which can confirm or correct it.

Since 2020, initially as part of the European Cos4Cloud project, Pl@ntNet has been publishing its most reliable observations on GBIF as open data. This portal is a window on that data.

What is published on GBIF

Pl@ntNet is a registered GBIF publisher with two occurrence datasets. Both are refreshed regularly from the Pl@ntNet database.

Dataset What it contains Records
Pl@ntNet observationsdoi:10.15468/gtebaa Observations whose species name is considered valid after collaborative revision by the community, and for which the authors agreed to share the photos under a Creative Commons licence. Every record has at least one image. ~2.6 million
Pl@ntNet automatically identified occurrencesdoi:10.15468/mma2ec Observations identified only by the deep-learning model, kept when the model’s confidence score is high enough to consider the identification valid. No images are shared for these records. ~12 million

Together they represent more than 14 million occurrences of about 21,000 species from 239 countries and territories. Most observations are recent (2018 onwards) and come from Europe, but the community is growing fast in Africa, the Americas and Asia.

What a record looks like

Each occurrence is published in Darwin Core and includes:

  • the scientific name proposed by Pl@ntNet, which GBIF matches against its backbone taxonomy;
  • the date and coordinates of the observation (with the precision the user allowed);
  • the basis of record HUMAN_OBSERVATION;
  • a link back to the observation on identify.plantnet.org;
  • for the observations dataset, one or more photos with the organ shown (flower, leaf, fruit, bark, habit) and the author’s name.

Observations are opportunistic: users photograph plants where and when they choose. Cities, tourist sites and springtime are over-represented. Anyone modelling species distributions from these data should account for that sampling bias — see the news section for how Pl@ntNet researchers do it.

Licence and citation

All Pl@ntNet data on GBIF are released under the Creative Commons Attribution 4.0 licence. Photos in the observations dataset are credited to their authors under CC BY-SA.

The best way to cite the data is to download them from GBIF, which mints a DOI for every download. Cite that DOI in your work. It lets the community track how the data are reused — more than 1,700 publications already do, and you can browse them in the literature section.

If you need to cite the datasets themselves:

Pl@ntNet (2025). Pl@ntNet observations. Occurrence dataset. https://doi.org/10.15468/gtebaa accessed via GBIF.org.

Pl@ntNet (2025). Pl@ntNet automatically identified occurrences. Occurrence dataset. https://doi.org/10.15468/mma2ec accessed via GBIF.org.

Accessing the data programmatically

Everything shown on this portal comes from the public GBIF API. A few starting points:

# Count Pl@ntNet occurrences of a species (here Quercus ilex, taxonKey 2879098)
https://api.gbif.org/v1/occurrence/search?publishingOrg=da86174a-a605-43a4-a5e8-53d484152cd3&taxonKey=2879098&limit=0

# List Pl@ntNet datasets
https://api.gbif.org/v1/dataset/search?publishingOrg=da86174a-a605-43a4-a5e8-53d484152cd3

For anything larger than a few thousand records, use the GBIF download API or the rgbif / pygbif libraries, and filter on publishingOrg = da86174a-a605-43a4-a5e8-53d484152cd3.

If you want to identify plants rather than analyse observations, the Pl@ntNet API is the right tool.

About this portal

This website is a GBIF hosted portal: the search, map and dataset pages are GBIF’s own components, configured to show only Pl@ntNet data. It is maintained by the Pl@ntNet team. Questions, corrections and collaboration ideas are welcome at contact@plantnet.org.