In March 2025, a magnitude 7.9 earthquake struck Myanmar. Prompt and accurate identification of damaged Bridges is of great significance for emergency rescue and post-disaster reconstruction. Traditional methods' reliance on substantial sample data, coupled with spectral interference in complex scenarios, and complex algorithmic processes in scarce-sample situations, leads to recognition model construction difficulties and insufficient accuracy/efficiency for emergency response requirements. To enhance emergency response efficiency, this study proposes a rapid damage assessment framework for bridges integrating multi-source geospatial data with vision-language model. The framework incorporates a semantic prompt system for bridge damage assessment in complex disaster scenarios, embedding bridge condition features, typical damage patterns, and contextual descriptors. This system enables the vision-language model to autonomously extract structural damage features from imagery, achieving automated damage assessment directly from visual data. The proposed framework was implemented across 733 bridges within a 30km radius of the Myanmar earthquake epicenter. Experimental results demonstrate the method's capability to achieve precise damage localization and identification within critical post-disaster timeframes. For bridges beyond 30m in length, the system attained over 88% overall recognition accuracy with processing durations under four hours. This study represents the first implementation of vision-language models in emergency response protocols for infrastructure assessment without any samples for training, establishing an automated solution for post-disaster bridges (>30m) damage extraction. The framework demonstrates significant potential for practical implementation and scalability in disaster management operations.

In January 2025, a magnitude 6.8 earthquake struck Dingri County, Shigatse, Tibet, causing severe damage. Rapid and precise extraction of damaged buildings is essential for emergency relief and rebuilding efforts. This study proposes an approach integrating YOLO-E (Real-Time Seeing Anything) and the Segment Anything Model 2 (SAM2) to extract damaged buildings with multi-source remote sensing images, including post-earthquake Gaofen-7 imagery (0.80 m), Beijing-3 imagery (0.30 m), and pre-earthquake Google satellite imagery (0.15 m), over the affected region. In this hybrid approach, YOLO-E functions as the preliminary segmentation module for initial segmentation. It leverages its real-time detection and segmentation capability to locate potential damaged building regions and generate coarse segmentation masks rapidly. Subsequently, SAM2 follows as a refinement step, incorporating shapefile information from pre-disaster sources to apply precise, pixel-level segmentation. The dataset used for training contained labeled examples of damaged buildings, and the model optimization was carried out using stochastic gradient descent (SGD), with cross-entropy and mean squared error as the selected loss functions. Upon evaluation, the model reached a precision of 0.840, a recall of 0.855, an F1-score of 0.847, and an IoU of 0.735. It successfully extracted 492 suspected damaged building patches within a radius of 20 km from the earthquake epicenter, clearly showing the distribution characteristics of damaged buildings concentrated in the earthquake fault zone. In summary, this hybrid YOLO-E and SAM2 approach, leveraging multi-source remote sensing imagery, delivers precise and rapid extraction of damaged buildings with a precision of 0.840, recall of 0.855, and IoU of 0.735, effectively supporting targeted earthquake rescue and post-disaster reconstruction efforts in the Dingri County fault zone.

Complex terrain conditions and dense vegetation cover in a vegetation area present significant challenges for point cloud data processing and the accurate extraction of ground points. This work integrates the physical characteristics between ground and non-ground points from the traditional Cloth Simulation Filter (CSF) algorithm and the strong learning capability of the machine learning Random Forest (RF) framework, developing the CSF-RF fusion algorithm for filtering ground points in vegetated areas, which can improve the accuracy of point cloud filtering in complex terrain environments. Both type I and type II errors do not exceed 0.05%, and the total error is maintained within 0.03%. Particularly in areas with dense vegetation and severe terrain undulations, the advantages are evident: the CSF-RF algorithm achieves a total error of only 0.19%, representing a 79.6% relative reduction compared with the 0.93% error of the CSF algorithm, while also reducing cases of ground point omission. Thus, it can be seen that the CSF-RF algorithm can effectively reduce vegetation interference and exhibits good stability, providing effective technical support for the accurate extraction of Digital Elevation Models (DEMs) in vegetated areas.

Destructive earthquakes frequently sever transportation lifelines, significantly impeding the progress of emergency rescue and post-disaster reconstruction efforts. The automated identification of road damage utilizing high-resolution remote sensing imagery is strictly constrained by the scarcity of post-disaster labeled samples and the morphological complexity of road networks. Consequently, model segmentation results frequently suffer from discontinuities in topological connectivity and confusion between background features and damaged roads. To address these challenges, this study proposes a road damage detection framework that integrates generative artificial intelligence with vector prior knowledge. A data simulation pipeline utilizing a stable diffusion model was constructed, employing topologically constrained masking to generate high-fidelity synthetic damage samples based on the DeepGlobe dataset, thereby mitigating the data deficit. The proposed Vector-Guided Damaged Road Segmentation Network (VRD-U2Net) employs wavelet convolutions (WTConv) to decouple high-frequency noise from low-frequency structural components and utilizes a Multi-Scale Residual Attention (MSRA) module to align visual features with vector priors. Furthermore, a vector-prior-driven dynamic upsampling mechanism is introduced to enforce geometric constraints on model predictions. Experimental results demonstrate that the method achieves an mIoU of 0.884 on the synthetic dataset. In validation using real-world imagery from the 2023 Turkey earthquake, the model attained an F1-score of 65.3% and recall of 72.3% without fine-tuning, exhibiting robust generalization capabilities to support manual damage assessment in data-scarce emergency scenarios.

As critical infrastructure for flood control and disaster mitigation, the completeness of a dam spatial database directly impacts regional emergency disaster response. However, existing dam data in some developing countries suffer from severe gaps and outdated information, particularly concerning small- and medium-sized dams, hindering rapid response during disasters. There is an urgent need to improve the physical dam database and implement dynamic monitoring. Yet, current remote sensing identification methods face limitations, including a lack of diverse dam samples, limited analysis of geographical factors, and low efficiency in full-image processing, making it difficult to efficiently enhance dam databases. To address these issues, this study proposes a dam extraction framework integrating comprehensive geographical factor analysis with deep learning detection, validated in Sindh Province, Pakistan. Firstly, multiple geographical factors were fused using the Random Forest algorithm to generate a dam existence probability map. High-probability candidate areas were delineated using dynamic threshold segmentation (precision: 0.90, recall: 0.76, AUC: 0.86). Subsequently, OpenStreetMap (OSM) water body data excluded non-dam potential areas, further narrowing the candidate areas. Finally, a dam image dataset was constructed to train a dam identification model based on YOLOv11, achieving an mAP50 of 0.85. This trained model was then applied to high-resolution remote sensing imagery of the candidate areas for precise identification. Ultimately, 16 previously unrecorded small and medium-sized dams were identified in Sindh Province, enhancing its dam location database. Experiments demonstrate that this method, through the synergistic optimization of geographical constraints and deep learning, significantly improves the efficiency and reliability of dam identification. It provides high-precision data support for dam disaster emergency response and water resource management, exhibiting strong practical utility and regional scalability.

Accurate mapping of building damage after destructive natural disasters is essential for emergency response and recovery planning. However, annotated post-disaster samples are often scarce, and conventional supervised models usually generalize poorly across scenes and hazard types. These constraints limit the operational use of remote sensing-based damage assessment during emergency response. To address this problem, this study proposes SCTSC-VLM, a training-free framework for damage grading of pre-localized building instances, implemented using a locally deployed Q4_K_M 4-bit quantized version of the Qwen3-VL-30B vision–language model. The framework has three main components. First, an instance-level visual input construction strategy integrates independent assessment of building instances localized by the supplied footprint data without a dedicated instance segmentation model. Second, a hierarchical reasoning mechanism decomposes the assessment into four progressive stages: multi-view structured feature extraction, independent visual description, cross-validation, and rule-constrained classification. This design is intended to reduce hallucination and visually ungrounded reasoning. Third, a multi-sampling ensemble decision mechanism mitigates the randomness of autoregressive generation. The framework was evaluated using post-disaster optical imagery from the 2025 Dingri earthquake and the 2025 Eaton Fire in California. On the full Dingri evaluation dataset comprising Areas A–C (n = 398), SCTSC-VLM achieved a Macro-F1 score of 88.2% and a Kappa coefficient of 0.806. On the separate full California evaluation dataset (n = 386), it achieved a Macro-F1 score of 83.8% and a Kappa coefficient of 0.800. Separate ablation experiments were conducted on explicitly defined analysis populations to examine the contributions of the framework components. In both full-dataset evaluations, the fully damaged and undamaged classes obtained higher F1-scores than the partially damaged class. These results provide case-level evidence that structured reasoning can support training-free damage grading of pre-localized buildings, although broader generalization and large-area operational efficiency remain to be established.