Pl@ntNet had the pleasure of hosting for the past two weeks Étienne Laliberté, a researcher at MILA (Montréal Institute for Learning Algorithms). Currently visiting Montpellier, he met with the Pl@ntNet team to explore potential collaborations between our two structures. This visit takes place within a particularly dynamic scientific context, supported by his recent work on drone-based hyperspatial imaging applied to tropical canopies.
The MILA team has developed a remarkably efficient method for acquiring aerial images of forest canopies. One research site plays a central role in this setup: Barro Colorado Island. The drones used are capable of covering several square kilometers in a single flight while producing very high-resolution images, reaching up to 4 cm per pixel, enabling highly precise automated segmentation of tree crowns. This technology, which follows on from approaches developed within the XPRIZE Rainforest competition, opens new perspectives for large-scale mapping of tropical trees.
However, these advances face a major challenge: data annotation and access to the botanical expertise required to process such large volumes of images. Traditional field-based approaches do not allow sampling enough individuals for the majority of species, which greatly limits our ability to understand their responses to climate change and other stress factors. It is in this context that MILA and Pl@ntNet have begun discussions to assess to what extent the plant community identification tools developed for the Pl@ntNet platform could meet the needs of Étienne Laliberté’s team and help address the limitations surrounding the significant potential these technologies offer for studying tropical canopies.
On Tuesday, November 4, Étienne Laliberté also presented his work in detail during a one-and-a-half-hour conference that brought together more than fifty francophone researchers. He presented the foundations of this drone imaging solution, its possible applications in science, conservation and restoration of tropical forests, as well as the scientific and technical challenges that must still be addressed for this approach to fully deliver on its promises. The video recording of the conference is also available on YouTube at the following link: https://www.youtube.com/watch?v=KFLmtKB2EsQ
This visit directly echoes the mission carried out by Alexis Joly and Jean-Christophe Lombardo from INRIA at MILA last October and forms part of the ongoing exchanges initiated at that time.
