TerraMind: Large-Scale Generative Multimodality for Earth Observation

TL;DR

TerraMind is a pioneering large-scale multimodal foundation model for Earth observation that combines token-level and pixel-level data to enable zero-shot, few-shot, and data augmentation applications, achieving state-of-the-art results.

Contribution

It introduces a dual-scale early fusion approach and the novel 'Thinking-in-Modalities' capability for improved EO data analysis.

Findings

Enables zero-shot and few-shot EO applications

Achieves beyond state-of-the-art benchmark performance

Introduces 'Thinking-in-Modalities' for data augmentation

Abstract

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 representations combining both token-level and pixel-level data across modalities. On a token level, TerraMind encodes high-level contextual information to learn cross-modal relationships, while on a pixel level, TerraMind leverages fine-grained representations to capture critical spatial nuances. We pretrained TerraMind on nine geospatial modalities of a global, large-scale dataset. In this paper, we demonstrate that (i) TerraMind's dual-scale early fusion approach unlocks a range of zero-shot and few-shot applications for Earth observation, (ii) TerraMind introduces "Thinking-in-Modalities" (TiM) -- the capability of generating additional artificial data during finetuning and inference to improve the model output -- and (iii) TerraMind achieves beyond state-of-the-art performance in community-standard benchmarks for EO like PANGAEA. The pretraining dataset, the model weights, and our code are open-sourced under a permissive license.