GeoZero: Incentivizing Reasoning from Scratch on Geospatial Scenes

2025-11-27

AI Summary

Existing remote sensing multimodal large language models (MLLMs) rely on manually annotated chain-of-thought (CoT) data for cold-start training, incurring high annotation costs, introducing human bias, and limiting reasoning diversity. To address this, we propose GeoZero: a framework that eliminates the need for manual CoT annotations by leveraging two-stage unsupervised geographic reasoning datasets—GeoZero-Instruct and GeoZero-Hard—to enable zero-shot learning of deep spatial reasoning. We further introduce Answer-Anchored Group Relative Policy Optimization (A²GRPO), a novel algorithm unifying supervised fine-tuning and reinforcement learning to achieve answer-anchored, diverse, and high-accuracy autonomous reasoning. Evaluated across multiple remote sensing vision-language benchmarks, GeoZero significantly outperforms state-of-the-art methods, demonstrating strong generalization and emergent geographic reasoning capabilities. Our work advances geospatial intelligence toward low-label-dependency and high-autonomy understanding.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Multimodal large language models (MLLMs) have undergone rapid development in advancing geospatial scene understanding. Recent studies have sought to enhance the reasoning capabilities of remote sensing MLLMs, typically through cold-start training with elaborately curated chain-of-thought (CoT) data. However, this approach not only incurs substantial annotation costs but also introduces human biases that may limit the diversity of model reasoning. To address these challenges, we propose GeoZero, a framework that enables MLLMs to perform geospatial reasoning without any predefined CoT supervision. Specifically, we construct two datasets, GeoZero-Instruct and GeoZero-Hard. GeoZero-Instruct allows the model to acquire preliminary geospatial knowledge through supervised fine-tuning, while GeoZero-Hard stimulates deep reasoning during the subsequent reinforcement learning stage. Furthermore, we introduce Answer-Anchored Group Relative Policy Optimization (A$^2$GRPO), where the reasoning process is regularized by the model's own answers, encouraging diverse yet accurate thinking. Extensive experiments on multiple remote sensing vision-language benchmarks demonstrate that GeoZero not only surpasses existing state-of-the-art methods but also fosters universal emergent reasoning capabilities across diverse geospatial tasks. Code,data,and models will be publicly available at https://github.com/MiliLab/GeoZero.

Problem

• Enhance geospatial reasoning without predefined chain-of-thought data
• Reduce annotation costs and human bias in remote sensing MLLMs
• Foster diverse and accurate reasoning across geospatial tasks

Innovation

• GeoZero enables geospatial reasoning without predefined chain-of-thought supervision
• It uses supervised fine-tuning and reinforcement learning on two datasets
• A$^2$GRPO regularizes reasoning with model answers to encourage diverse thinking