DisasterTD: Disaster Toponym Disambiguation Using Multimodal LLMs and Cross-View Geolocalization
DisasterTD: Disaster Toponym Disambiguation Using Multimodal LLMs and Cross-View Geolocalization
DisasterTD：利用多模态大模型与跨视角地理定位进行灾害地名消歧
Abstract: Social media imagery (SMI) provides timely and fine-grained ground perspectives that are valuable for situational awareness and emergency response. Unlike satellite or aerial imagery, SMI can capture disaster impacts and ground-level conditions in a timely manner. However, geographic references in SMI are often vague or ambiguous, making accurate geolocalization challenging.
摘要： 社交媒体图像（SMI）提供了及时且细粒度的地面视角，对于态势感知和应急响应具有重要价值。与卫星或航空影像不同，SMI 能够及时捕捉灾害影响和地面状况。然而，SMI 中的地理参考往往模糊不清，这使得精确的地理定位极具挑战性。
To address this issue, we propose DisasterTD, a disaster toponym disambiguation framework that integrates multimodal large language model (MLLMs)-based semantic reasoning with cross-view geolocalization. First, MLLMs extract toponyms and generate candidate geolocations from noisy textual inputs. Then, cross-view matching between SMI, remote sensing imagery (RSI), and optionally street-view imagery (SVI) is used to verify and refine these candidate results.
为了解决这一问题，我们提出了 DisasterTD，这是一个集成了基于多模态大模型（MLLM）的语义推理与跨视角地理定位的灾害地名消歧框架。首先，MLLM 从嘈杂的文本输入中提取地名并生成候选地理位置。随后，通过 SMI、遥感影像（RSI）以及可选的街景影像（SVI）之间的跨视角匹配，对这些候选结果进行验证和精炼。
We evaluate DisasterTD on the Hurricane Harvey dataset, where SMI is augmented with collected RSI and SVI to construct a cross-view benchmark for disaster geolocalization. The dataset is divided into four categories based on toponym clarity and ambiguity, allowing a fine-grained performance analysis across scenarios.
我们在“哈维飓风”（Hurricane Harvey）数据集上评估了 DisasterTD，该数据集通过收集的 RSI 和 SVI 对 SMI 进行了增强，构建了一个用于灾害地理定位的跨视角基准。根据地名的清晰度和歧义程度，该数据集被分为四个类别，从而能够对不同场景下的性能进行细粒度分析。
Results show that DisasterTD consistently outperforms MLLM-only and cross-view-only baselines without disambiguation, achieving geolocalization accuracies of 71.62% within 1000 m, 62.36% within 500 m, 57.99% within 250 m, 52.09% within 100 m, and 47.01% within 50 m, while reducing the mean and median errors to 11.33 km and 0.68 km, respectively.
结果表明，DisasterTD 在性能上始终优于仅使用 MLLM 或仅使用跨视角匹配且未进行消歧的基准模型。其地理定位准确率在 1000 米范围内达到 71.62%，500 米内为 62.36%，250 米内为 57.99%，100 米内为 52.09%，50 米内为 47.01%；同时，平均误差和中位误差分别降低至 11.33 公里和 0.68 公里。
The largest improvements appear in ambiguous toponyms, where semantic reasoning with cross-view evidence reduces candidate dispersion and errors. These findings demonstrate the effectiveness of integrating MLLM-based candidate generation with cross-view verification for fine-grained disaster geolocalization.
在处理歧义地名时，性能提升最为显著，因为结合跨视角证据的语义推理有效减少了候选位置的离散度和误差。这些发现证明了将基于 MLLM 的候选生成与跨视角验证相结合，对于细粒度灾害地理定位的有效性。