![[2602.19608] Satellite-Based Detection of Looted Archaeological Sites Using Machine Learning](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.19608] Satellite-Based Detection of Looted Archaeological Sites Using Machine Learning
Summary
This article presents a machine learning approach to detect looted archaeological sites using satellite imagery, demonstrating significant advancements in cultural heritage preservation.
Why It Matters
The study addresses the critical issue of archaeological site looting, which threatens cultural heritage globally. By leveraging machine learning and satellite technology, this research offers a scalable solution for monitoring and protecting these sites, potentially influencing future conservation efforts.
Key Takeaways
- The study uses satellite imagery to identify looted archaeological sites.
- ImageNet-pretrained CNNs significantly outperform traditional machine learning methods.
- Spatial masking enhances detection accuracy in identifying looting signatures.
- The research provides a curated dataset of 1,943 archaeological sites for further studies.
- Results indicate that geospatial foundation model embeddings are competitive with handcrafted features.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.19608 (cs) [Submitted on 23 Feb 2026] Title:Satellite-Based Detection of Looted Archaeological Sites Using Machine Learning Authors:Girmaw Abebe Tadesse, Titien Bartette, Andrew Hassanali, Allen Kim, Jonathan Chemla, Andrew Zolli, Yves Ubelmann, Caleb Robinson, Inbal Becker-Reshef, Juan Lavista Ferres View a PDF of the paper titled Satellite-Based Detection of Looted Archaeological Sites Using Machine Learning, by Girmaw Abebe Tadesse and 9 other authors View PDF HTML (experimental) Abstract:Looting at archaeological sites poses a severe risk to cultural heritage, yet monitoring thousands of remote locations remains operationally difficult. We present a scalable and satellite-based pipeline to detect looted archaeological sites, using PlanetScope monthly mosaics (4.7m/pixel) and a curated dataset of 1,943 archaeological sites in Afghanistan (898 looted, 1,045 preserved) with multi-year imagery (2016--2023) and site-footprint masks. We compare (i) end-to-end CNN classifiers trained on raw RGB patches and (ii) traditional machine learning (ML) trained on handcrafted spectral/texture features and embeddings from recent remote-sensing foundation models. Results indicate that ImageNet-pretrained CNNs combined with spatial masking reach an F1 score of 0.926, clearly surpassing the strongest traditional ML setup, which attains an F1 score of 0.710 using SatCLIP-V+RF+Mean, i.e., location and vision embeddings fe...
