用卫星影像和大模型检测考古遗址变化，支持月级预警与定位。
WATCH: Wide-Area Archaeological Site Tracking for Change Detection
- 基于时序嵌入距离与自监督重建等三类方法，实现无监督变化检测。
- 在阿富汗1943处遗址上，3个月内误差率超92.5%，部分达55%精确召回。
- 适合文化遗产保护者、遥感研究者使用，可提前发现破坏征兆。
大规模监测考古遗址对保护文化遗产至关重要，但因视觉线索微弱且地面真实数据稀疏，难以精确定位扰动发生时间。我们提出WATCH框架，用于在PlanetScope卫星拼接图（2017–2024年，4.7米/像素）上实现月级变化事件定位，包含三种互补评分方式：(i) 无需训练的时序嵌入距离（TED），通过局部时序参考评估月度偏差；(ii) 自监督变化检测（SSCD），结合重建、预测与潜在异常信号的集成方法；(iii) 基于稀疏事件月份标签的弱监督（WS）时序定位模型。我们在阿富汗1,943个遗址上评估WATCH，使用六种基础模型（CLIP、GeoRSCLIP、SatMAE、Prithvi-EO-2.0、DINOv3、Satlas-Pretrain）的嵌入，并对比手工设计的光谱与纹理基线，同时在叙利亚、土耳其、巴基斯坦和埃及的遗址上测试跨区域泛化能力。无监督方法（TED、SSCD）持续优于弱监督模型。以SatMAE为嵌入的TED在精确月份召回率达55%（m=0），而使用GeoRSCLIP、CLIP或Satlas-Pretrain的TED在3个月容差内达到92.5%（m=3）。手工特征在弱监督下仍具竞争力。方向性边界分析显示系统性时间偏差：与GeoRSCLIP或Prithvi-EO-2.0搭配的SSCD表现出最强的早期预警能力，在记录事件前即检测到异常；而TED更偏向变化发生后的确认型检测。结果表明，结合卫星影像与基础模型嵌入可实现可扩展、决策相关的遗产监测。
原文摘要 · Abstract (English)
Monitoring archaeological sites at scale is vital for protecting cultural heritage, yet pinpointing when disturbances occur remains difficult because visual cues are subtle and ground-truth data are sparse. We introduce WATCH, a framework for month-level change-event localization over PlanetScope satellite mosaics (2017-2024, 4.7 m/px) that supports three complementary scoring approaches: (i) Temporal Embedding Distance (TED), a training-free method that scores month-to-month deviations from a local temporal reference; (ii) Self-Supervised Change Detection (SSCD), an ensemble of reconstruction, forecasting, and latent-novelty signals; and (iii) a Weakly Supervised (WS) temporal localization model trained with sparse event-month labels. We benchmark WATCH on 1,943 archaeological sites in Afghanistan using embeddings from six foundation models (CLIP, GeoRSCLIP, SatMAE, Prithvi-EO-2.0, DINOv3, and Satlas-Pretrain) alongside a handcrafted spectral and texture baseline, and assess cross-regional generalization on sites in Syria, Turkey, Pakistan, and Egypt. The unsupervised approaches (TED, SSCD) consistently outperform the weakly supervised alternative. TED with SatMAE achieves the highest exact-month recall (55% at m=0), while TED with GeoRSCLIP, CLIP, or Satlas-Pretrain reaches 92.5% within a three-month tolerance (m=3). Handcrafted features remain competitive for exact-month detection under weak supervision. Our directional margin analysis reveals systematic temporal biases: SSCD paired with GeoRSCLIP or Prithvi-EO-2.0 exhibits the strongest early-warning profile, detecting anomalies before the recorded event, while TED favors confirmation-oriented detection after a change has materialized. These results show that satellite imagery combined with foundation-model embeddings enables scalable, decision-relevant heritage monitoring. Code: https://github.com/microsoft/WATCH
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。