The PlantCLEF 2025 challenge, part of the international LifeCLEF initiative, launched earlier this year with the goal of pushing the boundaries of automated plant identification. Supported by MAMBO and GUARDEN, the challenge focused on one of biodiversity monitoring’s most complex tasks: recognising multiple plant species from a single vegetation plot image.

Participants were invited to develop AI models capable of identifying all plant species present within high-resolution quadrat images. Unlike traditional classification tasks, this required predicting entire plant communities from images representing diverse European habitats. The training resources included over a million labelled images from the Pl@ntNet platform and GBIF, as well as a large collection of unlabelled vegetation images to support self-supervised approaches. Test data featured expertly curated photos from Mediterranean and Pyrenean ecosystems.

This year’s edition encouraged the use of modern machine learning strategies such as transfer learning, self-supervised learning, and the integration of ecological metadata. Final results will be presented at the CLEF 2025 Conference, taking place in Madrid from 9-12 September.

For more on PlantCLEF and other LifeCLEF challenges, see here.