Résumé
Pl@ntNet is a large-scale participatory platform dedicated to the collection of botanical observations thanks to crowdsourcing approaches and machine learning tools [a]. This initiative, supported since 2009, has allowed developing a computational infrastructure able to propose among others, a mobile plant identification service based on automated image analysis . This service, freely available on iPhone (https://itunes.apple.com/fr/app/plantnet/id600547573?mt=8) Android (https://play.google.com/store/apps/details?id=org.plantnet) and the web (http://identify.plantnet-project.org/), was initially set up for a fraction of the European flora (800 species at the beginning), and now accounts 6 000 species of the European flora, other tropical regions such as Indian Ocean flora, French Guyana flora, and North African flora. With more than one million downloads in 3 years, and a daily use of more than 6 000 people per day, the infrastructure is now able to produce a huge volume of botanical observations contributed by a wide range of actors. An impact study conducted in 2015 has allowed collecting more than 700 responses to a survey dedicated to characterize contexts of uses, and the most important needs by the community of end-users.Based on these feedbacks, we invested in several directions such as:- the educational perspectives of this framework in order to define specific usage scenarios at school and university levels,- specific functionalities such as the off-line function which represents a considerable evolution given the fact that it allows to use this system in field conditions, where 3G connection is lacking, as its often the case in tropical regions, or on tropical mountain ecosystems.We propose to present this initiative, and the latest realizations tested through this infrastructure, and discuss their potentials impacts in biological conservation, educational perspectives, and biodiversity studies