FlorID identifies species based on the images you provide. Once the identification is done, you will see a list of taxons associated with a value representing the level of confidence of the identification. You can improve the identification accuracy by adding more pictures, better pictures (see our recommendations) or using our API and adding information on location and observation date.
FlorID provides automatically generated model predictions that are subject to mistakes. We make no warranty of any kind regarding their accuracy and reliability. As a general guideline, low confidence scores indicate uncertain identifications, but high confidence scores do not guarantee correct identifications.
FlorID uses state-of-the-art artificial intelligence to identify observed vascular plant species in Switzerland based on images, habitat conditions, and observation date, with a focus on species occurring in the wild. The FlorID classifier is trained on millions of quality controlled images from citizen scientists and experts and even more geolocated species observations. With over 2500 taxa it can distinguish a major part of the wild plant species of the Swiss flora, including a large list of alien species that occur abundantly in the wild. Yet, the tool was not designed to identify ornamental plants that do not spread. So, do not expect accurate results with cultivated plants in gardens or at home for example. Currently, plants are identified to the level of species or species aggregates, but not yet to subspecies.
I want to integrate FlorID using this web application
Send a HTTP POST request to the FlorID URL and include the images you want to identify in the request body. Images can be submitted either as base64 encoded (parameter "images") or as URLs (parameter "urls"). The response HTML page showing the identification results can be embedded in your application.
Optionally, you can include two query parameters "redirect_url" and "token". If these parameters are submitted in your request, the identification results link back to the given redirect_url and will include the following query parameters: "token", "selected_taxonid", "probable_taxonids". This information allows you to identify the user and which identification result was selected to return back to your application.
If you integrate the FlorID service in your application, you have to acknowledge accordingly: Identified with the help of FlorID (and link to this page)
We would like to thank for the image contributions of the projects Atlas de la flore vaudoise, Flora des Kantons Zürich, Flora beider Basel, as well as Muriel Bendel, Françoise Alsaker, Andres Klein, Reto Deutschle, Robert Feller, Andreas Huber and all the contributing citizen scientists. We also would like to acknowledge scientific support by Steffen Boch, Christophe Bornand, Jonas Brännhage, Walter Brücker, Patrice Descombes, Günter Gottschlich, Andreas Huber, Philippe Juillerat, Adrian Möhl, Fernando Perrez-Cruz and the Swiss Data Science Center at ETHZ, Nina Richner, Alberto Serres-Hänni, Jan Dirk Wegner and the EcoVision lab at ETHZ, and Damaris Zurell.