March 14th, 2023 
Daniela Lopez de Luise We would like to invite you to participate in GeoLifeCLEF 2023, a machine learning competition that aims at predicting plant species composition in space and time. The competition is designed to support biodiversity management and conservation and to improve species identification and inventory tools.
The primary objective is to predict the set of plant species present at a given location and time using Sentinel-2 satellite images and Landsat time-series, as well as other rasterized environmental data like land-cover, human footprint, bioclimatic, and soil variables. The challenge presents several difficulties, including multi-label learning from single positive labels, strong class imbalance, multi-modal learning, and large-scale data.
We provide a large-scale training set of approximately 5 million plant occurrences in Europe (single-label data) belonging to 10,000 different species, thus covering a large proportion of the European flora. Unlike in previous years, in which occurrence data was also used for evaluation, we provide presence-absence data, thus multi-label, for 5,000 validation and 20,000 test plots.
The competition will be hosted by Kaggle and a summary of the results will be presented at the FGVC workshop at CVPR 2023 in Vancouver and at ImageCLEF 2023 in Thessaloniki.
Give it a try! More details can be found in:
Best regards,
The GeoLifeCLEF23 organizers
Posted in