Clever App Uses Smartphone Camera to Identify Plant Species


You've probably heard that a picture's worth a thousand words, but one new app is turning a picture into one single phrase — the name of a plant.

The free app PlantNet, easily described as Shazam (which identifies songs) for plants, can pinpoint the identity of flowers and plants using just a smartphone camera. The collaborative concept was driven by scientists, and amateur and professional botanists, and developed by digital science research firm Inria — with the support of France's agronomy-focused, open-science institute Agropolis Fondation.

The app (whose name is stylized as Pl@ntNet) is designed to use algorithms to identify and track rare, threatened, exotic or invasive plants, as well as satisfy the curiosity of people who would like to know what type of plant they've encountered. Launched in 2014 in France and rolling out across the world, PlantNet also has dozens of other useful applications in government and private industry. For instance, customs officers who might need to quickly identify invasive or rare plants can use the app, suggests Alexis Joly, a researcher at Inria who helped develop the app, in the video above. And farmers and forestry workers could use PlantNet to identify plant species before mistakenly eradicating endangered species.

To activate the PlantNet network, users snap a picture of an isolated leaf, select the proper category and upload it, then add additional photos of other parts of the same plant, such as a flower, fruit or bark. There are thousands of plants in the app's existing database.

"I think that users are quite proud to contribute to such a new collective knowledge," Joly says, adding that the more images the database contains, the more precise and fast the identification will become.

What makes the app so unique is that it uses a huge social network to collect — and continually update — field data that researchers and botanists can use to study the distribution and migration of plants.

Users validate and filter the crowdsourced images by voting on them, and images that receive at least three out of five stars are integrated into the app's permanent data collection. The collaborative validation continues with a second tool, called "IdentiPlante," that focuses not on the image quality, but on taxonomic details. For instance, it gives users the opportunity to confirm or disagree with the plant's common and scientific names.

Now that's citizen science at its finest, and with hundreds of thousands of plant species out there,  those pictures are going to be worth plenty of words.