With Info Flora’s FlorApp, botany enthusiasts have been able to report their plant observations from the field to the data centre. Now the app also helps to identify plants. The FlorID image recognition tool developed by WSL researchers makes this possible. It is based on the latest generation of artificial neural networks. "We have trained the computer with almost 1.5 million images and information on locations from the Info Flora data centre," says macroecologist Philipp Brun, who programmed the tool. Images of plant species that are rare in Switzerland supplemented the training images. A WSL staff member specifically searched for and photographed almost 260 species.
The result is an app that can recognise over 3000 plant species with a high degree of reliability - including rare species and those that are very similar. "In over 95 percent of cases, the correct species is among the first five suggestions, with the first suggestion being the correct one in about 85 percent of cases," says Brun. The app also owes its high hit rate to the combination of the images with geo-information on where the plants were found, such as vegetation structure, climate or topography. "This increases the accuracy of the results compared to other plant recognition apps," says Brun.
Data for Swiss research
Not only the users benefit from the new identification tool, but also the research community. Brun expects that the extension of FlorApp will increase data quality and data volume, which will benefit the Info Flora data centre. But WSL research will also benefit, because better data will increase the validity of biodiversity analyses. For example, species distribution models can predict more accurately which plant species can still occur where if site conditions change.WSL’s FlorID recognition tool can be easily developed further so that it can also recognise other organisms from images, such as butterflies or dragonflies. "We expect to be able to improve the data basis for various species groups with our tool. This should make the work around biodiversity surveys noticeably easier," says Brun.
Free identification app
Thanks to the FlorApp, anyone who has not yet dealt with plant identification at all will receive a hint about the name of a plant of interest even without prior knowledge. For a perfect identification and thus a valuable contribution to the recording of plant diversity in Switzerland, a dichotomous key and magnifying glass are still needed. The FlorApp is therefore suitable for interested laymen as well as for experienced botanists.The FlorApp is available for Android and iOS operating systems in German, French, Italian and English. It is free of charge and is available in the usual app stores.
Info Flora - My observations
FlorApp
WSL and SLF provide image and sound material free of charge for use in the context of press contributions in connection with this media release. The transfer of this material to image, sound and/or video databases and the sale of the material by third parties are not permitted.