Ph.D. Student on domain adaptation and self-supervised learning for Earth Observation at Conservatoire National des Arts et Métiers, as part of the ANR MAGE.
Formerly MSc. SDIA (Centrale Lille) and MSc. ATIAM (IRCAM, Sorbonne University, Telecom Paris)

Research Interests:

Earth Observation, Flow Matching, Diffusion models, Self-supervised Learning, Normalizing Flows, Computational Optimal Transport…

Publications

Detecting Out-Of-Distribution Earth Observation Images with Diffusion Models Georges Le Bellier, Nicolas Audebert. CVPR 2024 Earthvision workshop. [Proceedings][Poster]

Flow Matching for Domain Adaptation in Earth Observation,
Georges Le Bellier, Nicolas Audebert. RFIAP 2024. [Poster]

Projects

PANGAEA: A Global and Inclusive Benchmark for Geospatial Foundation Models. [code]

Introduction to Flow Matching: Annotated “Flow Matching for Generative Modeling” paper. [code]

Discrete Flow Matching: Educational implementation of the Discrete Flow Matching paper. [code]

Jax Flow Matching: A minimal implementation of Flow Matching in Jax. [code]

Diffusion Models for Expressive Musical Performances Generation: violin performances modeling with diffusion models [code]