arXiv 2511.226452026-03-13cs.CV

GeoZero：从零开始激励地理空间场景的推理

GeoZero: Incentivizing Reasoning from Scratch on Geospatial Scenes

Wuhan University（武汉大学）；Zhongguancun Academy（中关村学院）；Nanyang Technological University（南洋理工大学）；Shanghai AI Laboratory（上海人工智能实验室）

Di Wang, Shunyu Liu, Wentao Jiang, Fengxiang Wang, Yi Liu, Xiaolei Qin, Zhiming Luo, Chaoyang Zhou, Haonan Guo, Jing Zhang, Bo Du, Dacheng Tao, Liangpei Zhang

备注：: Code, data, and models are available at https://github.com/MiliLab/GeoZero

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

GeoZero通过构建两个数据集和A$^2$GRPO方法，使大语言模型在无预定义推理链监督下实现地理空间推理，提升了遥感任务的推理多样性和准确性。

摘要：

多模态大语言模型（MLLMs）在推进地理空间场景理解方面取得了快速发展。近期研究试图通过冷启动训练和精心编排的推理链（CoT）数据来增强遥感MLLMs的推理能力。然而，这种方法不仅导致显著的标注成本，还引入了可能限制模型推理多样性的偏见。为了解决这些挑战，我们提出了GeoZero，一个使MLLM能够在没有预定义CoT监督的情况下进行地理空间推理的框架。具体而言，我们构建了两个数据集，GeoZero-Instruct和GeoZero-Hard。GeoZero-Instruct允许模型通过监督微调获取初步的地理空间知识，而GeoZero-Hard在随后的强化学习阶段刺激深度推理。此外，我们引入了Answer-Anchored Group Relative Policy Optimization（A$^2$GRPO），通过模型自身的答案对推理过程进行正则化，鼓励多样化且准确的思考。在多个遥感视觉-语言基准测试中，GeoZero不仅超越了现有最先进的方法，还促进了跨多样地理空间任务的通用涌现推理能力。代码、数据和模型可在https://github.com/MiliLab/GeoZero获取。

英文摘要：

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 are available at https://github.com/MiliLab/GeoZero.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。