Overview of LifeCLEF Plant Identification Task 2019: diving into Data Deficient Tropical Countries

Hervé Goëau (1) , Pierre Bonnet (2) , Alexis Joly (3)

1 Cirad - Centre de Coopération Internationale en Recherche Agronomique pour le Développement
2 UMR AMAP - Botanique et Modélisation de l'Architecture des Plantes et des Végétations
3 ZENITH - Scientific Data Management

Automated identification of plants has improved consider-ably thanks to the recent progress in deep learning and the availabilityof training data. However, this profusion of data only concerns a few tensof thousands of species, while the planet has nearly 369K. The LifeCLEF2019 Plant Identification challenge (or ”PlantCLEF 2019”) was designedto evaluate automated identification on the flora of data deficient regions.It is based on a dataset of 10K species mainly focused on the Guianashield and the Northern Amazon rainforest, an area known to have oneof the greatest diversity of plants and animals in the world. As in theprevious edition, a comparison of the performance of the systems eval-uated with the best tropical flora experts was carried out. This paperpresents the resources and assessments of the challenge, summarizes theapproaches and systems employed by the participating research groups,and provides an analysis of the main outcomes.

Mots clés: Species identification; Guiana Shield leaves; Benchmark; Evaluation; Fine-grained classification; Plant; Expert; LifeCLEF; Amazon rain- forest; Tropical flora; PlantCLEF

Domaines: Systématique, phylogénie et taxonomie; Ecosystèmes; Botanique; Biodiversité et Ecologie

Hervé Goëau, Pierre Bonnet, Alexis Joly. Overview of LifeCLEF Plant Identification Task 2019: diving into Data Deficient Tropical Countries. CLEF 2019 - Conference and Labs of the Evaluation Forum, Linda Cappellato; Nicola Ferro; David E. Losada; Henning Müller, Sep 2019, Lugano, Switzerland. ⟨hal-02283184⟩