7 papers
Now We Know? A Systematic Comparison of TerraMind and THOR
Frederick Schindlegger, Kenzo Bounegta, Eva Gmelich Meijling +5
Benchmarks for Geospatial Foundation Models (GFMs) increasingly rank models by aggregate score, but such rankings obscure why models differ: how much of the gap is architecture, ho…
TerraMind: Large-Scale Generative Multimodality for Earth Observation
Johannes Jakubik, Felix Yang, Benedikt Blumenstiel +13
We present TerraMind, the first any-to-any generative, multimodal foundation model for Earth observation (EO). Unlike other multimodal models, TerraMind is pretrained on dual-scale…
COP-GEN: Latent Diffusion Transformer for Copernicus Earth Observation Data
Miguel Espinosa, Eva Gmelich Meijling, Valerio Marsocci +2
Earth observation applications increasingly rely on data from multiple sensors, including optical, radar, elevation, and land-cover. Relationships between modalities are fundamenta…
NeighborMAE: Exploiting Spatial Dependencies between Neighboring Earth Observation Images in Masked Autoencoders Pretraining
Liang Zeng, Valerio Marsocci, Wufan Zhao +2
Masked Image Modeling has been one of the most popular self-supervised learning paradigms to learn representations from large-scale, unlabeled Earth Observation images. While incor…
THOR: A Versatile Foundation Model for Earth Observation Climate and Society Applications
Theodor Forgaard, Jarle H. Reksten, Anders U. Waldeland +4
Current Earth observation foundation models are architecturally rigid, struggle with heterogeneous sensors and are constrained to fixed patch sizes. This limits their deployment in…
Scaling Laws for Geospatial Foundation Models: A case study on PhilEO Bench
Nikolaos Dionelis, Riccardo Musto, Jente Bosmans +7
Foundation Models (FMs) have achieved state-of-the-art performance across domains by leveraging large-scale pretraining. In Earth Observation (EO), the availability of petabyte-sca…