Atomizer: Generalizing to new modalities by breaking satellite images down to a set of scalars

TL;DR

Atomizer introduces a flexible, scalar-based representation for remote sensing images that enables a single model to generalize across diverse satellite data modalities without retraining.

Contribution

The paper presents Atomizer, a novel architecture that encodes satellite images as scalar sets with metadata, allowing cross-modality generalization without interpolation.

Findings

Outperforms standard models in modality-disjoint evaluations

Robust across different resolutions and spatial sizes

Eliminates need for retraining when new modalities are introduced

Abstract

The growing number of Earth observation satellites has led to increasingly diverse remote sensing data, with varying spatial, spectral, and temporal configurations. Most existing models rely on fixed input formats and modality-specific encoders, which require retraining when new configurations are introduced, limiting their ability to generalize across modalities. We introduce Atomizer, a flexible architecture that represents remote sensing images as sets of scalars, each corresponding to a spectral band value of a pixel. Each scalar is enriched with contextual metadata (acquisition time, spatial resolution, wavelength, and bandwidth), producing an atomic representation that allows a single encoder to process arbitrary modalities without interpolation or resampling. Atomizer uses structured tokenization with Fourier features and non-uniform radial basis functions to encode content and context, and maps tokens into a latent space via cross-attention. Under modality-disjoint evaluations, Atomizer outperforms standard models and demonstrates robust performance across varying resolutions and spatial sizes.