GRE套件：基于微调视觉-语言模型与增强推理链的地理定位推断

GRE Suite: Geo-localization Inference via Fine-Tuned Vision-Language Models and Enhanced Reasoning Chains

May 24, 2025

作者: Chun Wang, Xiaoran Pan, Zihao Pan, Haofan Wang, Yiren Song

cs.AI

摘要

视觉语言模型（VLMs）在视觉推理任务中展现出了卓越的性能。然而，地理定位任务提出了独特的挑战，需要从图像中提取多粒度视觉线索，并将其与外部世界知识相结合进行系统性推理。当前的地理定位方法往往缺乏稳健的推理机制和可解释性，限制了其有效性。为解决这些局限，我们提出了地理推理增强套件（GRE Suite），这是一个新颖的框架，通过结构化推理链增强VLMs，以实现准确且可解释的位置推断。GRE Suite在三个关键维度上系统性地开发：数据集、模型和基准。首先，我们引入了GRE30K，一个高质量的地理定位推理数据集，旨在促进细粒度的视觉和上下文分析。接着，我们提出了GRE模型，该模型采用多阶段推理策略，逐步推断场景属性、局部细节和语义特征，从而以更高的精度缩小潜在的地理区域。最后，我们构建了地理推理评估基准（GREval-Bench），这是一个全面的评估框架，用于评估VLMs在多样化的城市、自然和地标场景中的表现，衡量从粗粒度（如国家、大陆）到细粒度（如城市、街道）的定位性能。实验结果表明，GRE在所有粒度的地理定位任务中均显著优于现有方法，凸显了推理增强型VLMs在复杂地理推断中的有效性。代码和数据将在https://github.com/Thorin215/GRE发布。

English

Recent advances in Visual Language Models (VLMs) have demonstrated exceptional performance in visual reasoning tasks. However, geo-localization presents unique challenges, requiring the extraction of multigranular visual cues from images and their integration with external world knowledge for systematic reasoning. Current approaches to geo-localization tasks often lack robust reasoning mechanisms and explainability, limiting their effectiveness. To address these limitations, we propose the Geo Reason Enhancement (GRE) Suite, a novel framework that augments VLMs with structured reasoning chains for accurate and interpretable location inference. The GRE Suite is systematically developed across three key dimensions: dataset, model, and benchmark. First, we introduce GRE30K, a high-quality geo-localization reasoning dataset designed to facilitate fine-grained visual and contextual analysis. Next, we present the GRE model, which employs a multi-stage reasoning strategy to progressively infer scene attributes, local details, and semantic features, thereby narrowing down potential geographic regions with enhanced precision. Finally, we construct the Geo Reason Evaluation Benchmark (GREval-Bench), a comprehensive evaluation framework that assesses VLMs across diverse urban, natural, and landmark scenes to measure both coarse-grained (e.g., country, continent) and fine-grained (e.g., city, street) localization performance. Experimental results demonstrate that GRE significantly outperforms existing methods across all granularities of geo-localization tasks, underscoring the efficacy of reasoning-augmented VLMs in complex geographic inference. Code and data will be released at https://github.com/Thorin215/GRE.