English

MAESTRO: Masked AutoEncoders for Multimodal, Multitemporal, and Multispectral Earth Observation Data

Computer Vision and Pattern Recognition 2025-10-10 v2

Abstract

Self-supervised learning holds great promise for remote sensing, but standard self-supervised methods must be adapted to the unique characteristics of Earth observation data. We take a step in this direction by conducting a comprehensive benchmark of fusion strategies and normalization schemes of reconstruction targets for multimodal, multitemporal, and multispectral Earth observation data. Based on our findings, we introduce MAESTRO, a novel adaptation of the Masked Autoencoder with optimized fusion mechanisms and a normalization scheme that incorporates a spectral prior as a self-supervisory signal. Evaluated on four Earth observation datasets in both intra- and cross-dataset settings, MAESTRO achieves state-of-the-art performance on tasks that strongly rely on multitemporal dynamics, while also remaining competitive on others. Code to reproduce all our experiments is available at https://github.com/ignf/maestro.

Keywords

Cite

@article{arxiv.2508.10894,
  title  = {MAESTRO: Masked AutoEncoders for Multimodal, Multitemporal, and Multispectral Earth Observation Data},
  author = {Antoine Labatie and Michael Vaccaro and Nina Lardiere and Anatol Garioud and Nicolas Gonthier},
  journal= {arXiv preprint arXiv:2508.10894},
  year   = {2025}
}
R2 v1 2026-07-01T04:50:26.068Z