提出GeoMoE模型，实现高效跨视角地理定位与多分辨率适配。
One Query, Many Scales: Sparse Mixture-of-Experts for Efficient Hierarchical Cross-View Geo-Localization
- 采用稀疏专家混合架构，分离全局多尺度表征与局部层级搜索。
- 在VIGOR-M上达62.39% R@1，查询开销仅为全量扫描的5.27%。
- 支持跨分辨率迁移，单模型覆盖多分辨率场景，适合大规模地理检索。
跨视角地理定位（CVGL）旨在为地面视角查询检索带有地理标签的卫星图像。现有系统通常对固定分辨率的平坦图库进行穷举搜索，导致大区域下计算成本高，且难以适应卫星分辨率变化。自回归式的粗到精方法虽减少比对次数，但将后续预测绑定于早期决策并依赖预设层级结构。本文提出GeoMoE，一种稀疏专家混合双编码器，将全局多尺度表征学习与局部层级搜索解耦。通过全局多尺度监督和内容自适应路由，地面与卫星图像在不同分辨率间映射至统一可比嵌入空间。推理时，每张图像仅编码一次，概率性束搜索沿父子链接评估候选子集。后级层级复用前级生成的描述符，而非重新生成特征，从而限制特征级误差传播并降低层级耦合。进一步提出VIGOR-M基准，包含四座城市、显式父子卫星层级结构，以及保留的半步图库，用于单分辨率、跨分辨率及层级评估。GeoMoE在Just Zoom In上取得95.78% R@40m，优于此前最佳2.77个百分点；在VIGOR-M上达到62.39% R@1，查询匹配仅需0.885 MMAC/query，仅为完整L3扫描的5.27%，且超越最强全量基线3.12个百分点。一个在L1、L2、L3上训练的单一模型，在全部六个图库上均优于等量密集对照模型，并成功迁移到三个未见分辨率。通过解耦全局嵌入训练与本地层级搜索，GeoMoE同步提升定位精度、搜索效率与跨分辨率泛化能力。
原文摘要 · Abstract (English)
Cross-view geo-localization (CVGL) retrieves geo-tagged satellite imagery for a ground-view query. Most systems exhaustively search a flat, fixed-resolution gallery, incurring high cost over large areas and adapting poorly to satellite resolution changes. Autoregressive coarse-to-fine alternatives reduce comparisons but bind later predictions to earlier decisions and a predefined hierarchy. We introduce GeoMoE, a sparse mixture-of-experts dual encoder that decouples global multi-scale representation learning from local hierarchical search. Global multi-scale supervision and content-adaptive routing map ground and satellite images across resolutions into a globally comparable embedding space. At inference, each image is encoded once, and probabilistic beam search follows parent--child links to score a small candidate subset. Later levels reuse these descriptors rather than features generated by preceding levels, limiting feature-level error propagation and hierarchy coupling. We further introduce VIGOR-M, a four-city benchmark with an explicit parent--child satellite hierarchy and held-out half-step galleries for single-resolution, cross-resolution, and hierarchical evaluation. GeoMoE achieves 95.78% R@40m on Just Zoom In, 2.77 percentage points above the previous best, and 62.39% R@1 on VIGOR-M. The latter requires 0.885 MMAC/query for descriptor matching, 5.27% of an exhaustive L3 scan, while exceeding the strongest exhaustive baseline by 3.12 percentage points in R@1. One model trained on L1, L2, and L3 also outperforms a matched dense control across all six galleries and transfers to three withheld resolutions. By decoupling globally trained embeddings from local hierarchical search, GeoMoE jointly improves localization accuracy, search efficiency, and cross-resolution transfer.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。