B.inomen (“the app”), at binomen.app, is a free tool that tries to identify plants, animals and fungi from your photographs. It is operated by [operator name] (“we”). By using the app you agree to these terms. They are short because the app does very little with your data, and we would like to keep it that way.
1. What the app does
You choose up to three photographs. The app compares them against its reference set of species and shows you its best guesses, together with how confident it is. When it cannot name a species with confidence it will say so, or name only the genus.
2. Your photographs
Depending on how the app is configured, identification happens in one of two places.
Through our service. Your photograph is sent to our server to be identified there. When that happens it is processed in memory and discarded as soon as the identification is complete. It is not saved to disk, not stored, not backed up, not used to train or improve anything, and not shared with anyone. Before sending, the app reduces the photograph to at most 1024 pixels on its longest side and removes embedded metadata such as location. The one exception is the approximate area described next, and only if you switch it on.
Photo location (optional). Off unless you switch on “Use photo location”. When it is on, the app reads the location stored in your photograph, on your device, and reduces it to a square of the map about 110 km across. Only that area is sent with the photograph, never the exact position. Our service uses it to favour species recorded nearby, and it is discarded with the photograph. Photographs without a stored location are identified by the picture alone.
On your device. When this is enabled and your browser is able, the identification runs entirely inside your browser. Your photograph is not uploaded, not transmitted, and never leaves your device. The first time, your browser downloads the identification model (about 190 MB) and keeps a copy so it does not have to download it again.
3. Accuracy, and what not to rely on
Identifications are made by a machine-learning model and can be wrong, including confidently wrong. In our testing the first suggestion was correct about three times in four, and when the app named a species, the right one was among those shown about nine times in ten. It knows a large set of species, wild and cultivated, but not every species, variety or hybrid, and it cannot tell you that a photograph is not of a living thing at all.
Do not use the app to decide whether something is safe to eat, touch or handle. Never eat a plant, mushroom or berry, or handle an animal, on the strength of an identification from this app. Consult a qualified expert.
Toxicity notes are informational. Some results carry a note that a species is toxic, poisonous or venomous. These notes are compiled from public sources such as Wikipedia and are incomplete: the absence of a note does not mean a species is safe, and a note marked “considered” has not been individually verified. Similar-species lists are computed from visual resemblance in our index and are a hint, not a determination.
4. Usage information we collect
We record a small amount of anonymous information about how the app is used, so we can see whether it works and where it fails. For each identification we record: whether it ran on your device or on our service and why; how long it took; how many photographs were used; how confident the result was; and whether the app named a species, a genus, or nothing. When it named only a genus, we record which genus. We also record whether the model download started, finished or failed, and how long it took.
We do not record: the photograph; its location or area; the species named; your IP address; a user, session or device identifier; your browser’s identification string; or anything that could be used to recognise you or connect one use of the app to another. These records are kept for 90 days and then deleted.
5. Cookies and local storage
The app sets no cookies. It uses your browser’s local storage for three things only: to remember whether you prefer light or dark mode, to remember whether you switched on photo location, and — when on-device identification is enabled — to keep the downloaded model so it need not be fetched again. All three stay on your device, are never sent to us, and can be cleared at any time through your browser’s site-data settings.
When you press Identify, a box asks you to tick “I’m not a robot” — once every ten minutes, not for every photo. Your browser then solves a short calculation that our server set, and sends the answer with your photos so the server can check it. It uses no cookies, involves no other company, and records nothing about you.
6. Third parties
The app is hosted on Amazon Web Services in the United States. The short description shown with a species or genus is the opening of its Wikipedia article, prepared in advance and served with the rest of our species data, so identifying something sends nothing to Wikipedia. Wikipedia’s text is used under the Creative Commons Attribution-ShareAlike licence and is shown with a link to the original article.
If you choose to buy us a coffee, the payment is made on PayPal, under PayPal’s own terms. PayPal shares the name and email address on your PayPal account with us as the recipient; we see no card or bank details, and the app itself records nothing about the payment.
7. Acceptable use
Use the app for its purpose. Do not attempt to overload it, to interfere with it, or to use it to identify, locate or collect protected or endangered species in breach of any law.
8. Availability and changes
The app is provided free, as is, and may be changed, interrupted or withdrawn at any time without notice. We may update these terms; the date at the top will change when we do, and continued use after a change means you accept it.
9. Liability
To the fullest extent the law allows, we are not liable for any loss or harm arising from use of the app or reliance on any identification it gives, including the consequences of eating, touching or handling anything on the basis of a result. Nothing in these terms limits liability that cannot be limited by law.
10. Credits
Identification uses the BioCLIP 2 model by the Imageomics Institute (MIT licence). The reference set is derived from photographs in the iNaturalist Open Dataset, and English species names come from iNaturalist. Species descriptions are from Wikipedia. Species photographs are from Wikimedia Commons and iNaturalist, under the licence and with the author shown beside each one.
11. Governing law and contact
These terms are governed by the laws of [jurisdiction]. Questions about the app or these terms: [contact email].