folia

Le LIRIS diffuse un premier prototype de reconnaissance de feuilles sur l’AppStore

TheM2DisCo and Imagine teams, which develop in particular research activities on object recognition for the first one, and automatic object recognition for the second one, have published on AppStore their first prototype application for tree leaf recognition.

In the context of ANR ReVeS – Plant Recognition for Smartphone Software, the LIRIS laboratory works in collaboration with geographers (EVS laboratory) and specialists in data fusion (LISTIC laboratory). The consortium also contains a non-academic partner (la société linnéenne de Lyon) and relies on a software development company (Trydéa).

The purpose of the application is to allow a walker in middle of the country, to recognize trees around him by taking photographs of their leaves. 76 species  naturally growing in France are currently listed.

folia_body1.png

An algorithm made by the LIRIS laboratory analyzes the photo. It extracts the object of interest object (the leaf) from the image, which might be made up of a complex background with similar colors. After having extracted the contour of this object and estimated its global shape, it computes descriptors (base, tip, teeth, pits, etc.).

These descriptors are then compared to the ones in the learning base to deduce the most probable species.

In a one-step more version, the GPS localization of the user will also be taken into account to increase or reduce the confidences of responses.

 The ReVeS project aims to be collaborative. It invites Folia users to participate to the research program by sending their data to LIRIS laboratory; allowing increasing the database and outperforming the algorithm but also listing new botanical data on the France scale.

Do not hesitate to freely download the application from AppStore and make comments to the head of the project (laure.tougne@liris.cnrs.fr).

folia_body2.png