发表机构
University of Waterloo（滑铁卢大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对建筑损坏分类的空间上下文误用问题，提出灾害条件核正则化图注意力方法，结合残差去相关损失，在xBD数据集上提升宏F1并降低残差空间自相关，实现可靠的跨事件迁移。
AI 中文摘要
灾害损坏具有空间特性：建筑物很少孤立受损。然而，利用空间上下文进行损坏分类的研究却明显不足，许多流程即使在空间结构不确定性占主导时，仍主要依赖单栋建筑的外观线索。更为复杂的是，不同灾害事件对应的合适邻域并不相同：洪水、飓风和野火会呈现截然不同的聚类行为，这使得空间推理极具价值但也易被误用——朴素的上下文聚合可能提升视觉一致性，却会过度平滑边界或传播结构化误差。我们在xBD（xView2挑战赛使用的数据集）上，采用受控的后定位仅分类设置研究这一矛盾：每栋建筑由从提供的多边形中裁剪出的前后组合（PPC）图像块表示，空间上下文通过GPS生成的建筑图建模。我们的方法通过保留灾害损坏模式中的强空间关系，让局部证据“亲近”；同时通过灾害类型条件图模型，仅让合适的邻居“更亲近”，该模型将可学习的多尺度空间核先验注入注意力，使有效邻域尺度能适配不同灾害类型，而非作为单一全局平滑规则学习。为抑制“靠平滑实现一致性”的问题，我们加入残差去相关损失，惩罚预测残差中的正Moran's I。我们在xBD上采用留一事件（LOEO）协议，在事件和数据集偏移下评估方法，并开展从xBD到Ida-BD的跨数据集迁移。该模型在零样本事件偏移下提升了宏F1值，大幅降低了残差空间自相关，表明其能更好地利用空间上下文而非朴素平滑，还能实现对已知灾害类型内未见事件的更可靠迁移。
英文摘要
Disaster damage is spatial: buildings rarely fail in isolation. Yet using spatial context for damage classification remains surprisingly underexplored, and many pipelines still rely primarily on per-building appearance cues even when the dominant uncertainty is spatially structured. Complicating matters, the right neighbourhood is not the same across events. Floods, hurricanes, and wildfires can exhibit very different clustering behaviour, making spatial reasoning valuable but easy to misuse - naive context aggregation can improve visual coherence while oversmoothing boundaries or propagating structured errors. We study this tension on xBD (the dataset used in the xView2 challenge) in a controlled post-localization, classification-only setup: each building is represented by a pre/post combined (PPC) patch cropped from the provided polygons, and spatial context is modelled with GPS-derived building graphs. Our approach keeps local evidence "close" by preserving strong spatial relationships in disaster damage patterns, while bringing only the right neighbours "closer" through a disaster-type-conditioned graph model that injects a learnable multi-scale spatial kernel prior into attention, allowing the effective neighbourhood scale to adapt across disaster types rather than being learned as a single global smoothing rule. To discourage coherence-by-smoothing, we add a residual de-correlation loss that penalizes positive Moran's~I in prediction residuals. We evaluate the method under event and dataset shift with a leave-one-event-out (LOEO) protocol on xBD and cross-dataset transfer from xBD to Ida-BD. The model improves macro-F1 and substantially reduces residual spatial autocorrelation under zero-shot event shift, indicating better use of spatial context rather than naive smoothing and enabling more reliable transfer to unseen events within known disaster types.
CommentsAccepted in ECCV 2026
