Talk by Loïc Landrieu

Title

Recent Advances in Geospatial Machine Learning

Abstract

Geospatial machine learning has emerged as a compelling testing ground for modern computer vision and machine learning algorithms. In this talk, I will present recent progress in large-scale forest monitoring, including multi-year, tree-level, country-scale canopy height mapping, efficient biomass estimation from aerial LiDAR, and massively multimodal modeling. I will then discuss our latest work on visual geolocation, spanning both standardized benchmarking and diffusion-based generative models that output calibrated spatial probability maps.

Bio

Loïc Landrieu is a research scientist at ENPC and a Senior Hi! PARIS Fellow, working on machine learning for large-scale geospatial analysis. He received his PhD from École Normale Supérieure (ENS) Paris in 2016. Active in both remote sensing and computer vision, he serves as Area Chair for CVPR, ECCV, 3DV, and IGARSS, and sits on the editorial board of the ISPRS Journal while co-chairing the ISPRS Working Group on Temporal Data Understanding. He was co-program Chair of the 2022 ISPRS Congress, the CVPR EarthVision workshop, and the inaugural REO workshop at NeurIPS 2025.
Find more information about Loïc Landrieu and his research here.

Please sign up if you wish to attend the talk in person or online via Zoom.