Pl@ntNet: the community corrects the machine
A plant identification app that learns from its users and gives back to research what it receives. A good example of what MAUB sets out to highlight.
Projects /
Pl@ntNet
Identify a plant from a photo, and move botany forward along the way.
A citizen science app: you photograph a leaf, a flower or a fruit, and the app suggests the most likely species. Every observation that is shared and validated by the community adds to the botanical data made available to researchers.
And where does AI come in?
Image recognition through deep learning. The model learns from observations validated by users and botanists, so the community corrects the tool as much as the tool helps the community.
The AI label
- Model or technique
- Deep neural network (InceptionV3 according to the API FAQ), retrained roughly every two months.
- Model published
- Not filled in
- Where it runs
- On the project's servers, through its online API.
- Data
- Users' photos; validated observations are shared on GBIF under CC BY-SA.
- Role of people
- Users confirm or correct the suggested species, and their votes feed the training.
- Frugality
- Energy and memory use are among the checks run on every new model version.
“Not filled in”: MAUB could not check it with the project. Know more? Write to us.
Further reading
Something wrong or out of date? Write to contact@subalpine.it.