Beyond Zooming: Learning Multi-Tool Visual Reasoning for Ultra-High-Resolution Remote Sensing

2026-07-28

AI Summary

This work addresses the challenge that task-relevant evidence in ultra-high-resolution remote sensing imagery is often sparse, localized, and spatially dispersed, rendering existing single-scale tools inadequate for complex tasks requiring global search or multi-region reasoning. To overcome this limitation, we propose GeoMTVR, the first remote sensing visual reasoning framework that enables collaborative multi-tool coordination. GeoMTVR achieves efficient reasoning through problem decomposition, adaptive tool selection, and cross-tool evidence fusion, and further introduces a reinforcement learning algorithm with focused tool attention to optimize critical decision-making steps. Experimental results demonstrate that our model, GeoLens, significantly outperforms both direct reasoning approaches and single-tool baselines in terms of accuracy, evidence grounding capability, and tool utilization efficiency.

Abstract

Ultra-high-resolution (UHR) remote-sensing (RS) imagery provides fine-grained Earth-observation evidence over city-scale scenes, but poses a fundamental challenge for multimodal large language models (MLLMs): task-relevant evidence is often sparse, local, and spatially dispersed across extremely large visual contexts. A natural solution is to equip MLLMs with zoom-in tools for active local inspection. However, through a pilot study on XLRS-Bench, we find that zoom-in is only partially effective: it resolves easy and medium-level tasks with locally recoverable evidence, but saturates on hard cases requiring global search, multi-region comparison, path planning, or dispersed-evidence reasoning. Motivated by this finding, we move beyond single-tool zoom-in and introduce GeoMTVR, a large-scale Geospatial Multi-Tool Visual Reasoning dataset built from wide-area satellite imagery. GeoMTVR contains 13K UHR VQA samples with interleaved reasoning trajectories, diverse visual tool calls, and returned visual observations, enabling models to learn question decomposition, tool selection, regional inspection, object-level grounding, auxiliary visual reasoning, and cross-tool evidence integration. Beyond supervised fine-tuning, we propose a tool-attention-focused reinforcement learning algorithm that concentrates optimization on critical tool-use decisions, including when to invoke tools, which tool to select, where to apply it, and how to interpret tool outputs. By combining SFT on GeoMTVR with our RL algorithm, we develop GeoLens, a multi-tool visual reasoning MLLM for UHR RS. Experiments show that GeoLens consistently outperforms direct reasoning and single-tool zoom-in baselines, achieving stronger accuracy, better evidence grounding, and more efficient tool-use trajectories.

Problem

• Ultra-High-Resolution Remote Sensing
• Visual Reasoning
• Multimodal Large Language Models
• Sparse Evidence
• Geospatial Analysis

Innovation

• multi-tool visual reasoning
• ultra-high-resolution remote sensing
• tool-attention reinforcement learning
• geospatial VQA
• evidence integration