TerraScope: Pixel-Grounded Visual Reasoning for Earth Observation

2026-03-19

AI Summary

Existing vision-language models struggle to perform complex spatial reasoning in remote sensing tasks due to their inability to leverage precise pixel-level representations. To address this limitation, this work proposes TerraScope, the first pixel-level Earth observation vision-language model that supports modality flexibility (optical/SAR), multi-temporal inputs, and adaptive fusion. TerraScope introduces a chain-of-thought reasoning mechanism grounded in pixel-level mask embeddings to enable interpretable spatial inference. Additionally, the authors construct Terra-CoT, a large-scale dataset comprising one million samples, and TerraScope-Bench, the first pixel-level remote sensing reasoning benchmark. Experiments demonstrate that TerraScope significantly outperforms existing methods on tasks such as change analysis, achieving both high answer accuracy and high-quality mask generation, thereby validating its genuine pixel-level reasoning capability.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Security and Privacy: Data transparency and provenance
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Vision-language models (VLMs) have shown promise in earth observation (EO), yet they struggle with tasks that require grounding complex spatial reasoning in precise pixel-level visual representations. To address this problem, we introduce TerraScope, a unified VLM that delivers pixel-grounded geospatial reasoning with two key capabilities: (1) modality-flexible reasoning: it handles single-modality inputs (optical or SAR) and adaptively fuses different modalities into the reasoning process when both are available; (2) multi-temporal reasoning: it integrates temporal sequences for change analysis across multiple time points. In addition, we curate Terra-CoT, a large-scale dataset containing 1 million samples with pixel-level masks embedded in reasoning chains across multiple sources. We also propose TerraScope-Bench, the first benchmark for pixel-grounded geospatial reasoning with six sub-tasks that evaluates both answer accuracy and mask quality to ensure authentic pixel-grounded reasoning. Experiments show that TerraScope significantly outperforms existing VLMs on pixel-grounded geospatial reasoning while providing interpretable visual evidence.

Problem

• pixel-grounded reasoning
• earth observation
• geospatial reasoning
• vision-language models

Innovation

• pixel-grounded reasoning
• vision-language model
• multi-temporal analysis
• modality fusion
• geospatial reasoning