Archaeological Site Detection: Latest Results from a Deep Learning Based Europe Wide Hillfort Search
The increasing availability and resolution of LiDAR data are revolutionizing landscape archaeology, enabling unprecedented large-scale studies. However, the time-intensive nature of manual analysis has posed significant challenges. This research tackles the complexities of large-scale archaeological...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Ubiquity Press
2025-01-01
|
| Series: | Journal of Computer Applications in Archaeology |
| Subjects: | |
| Online Access: | https://account.journal.caa-international.org/index.php/up-j-jcaa/article/view/178 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The increasing availability and resolution of LiDAR data are revolutionizing landscape archaeology, enabling unprecedented large-scale studies. However, the time-intensive nature of manual analysis has posed significant challenges. This research tackles the complexities of large-scale archaeological site detection by integrating Artificial Intelligence (AI) with LiDAR data, focusing on hillforts across diverse European landscapes. A semi-automated workflow employing Convolutional Neural Networks (CNNs) was developed and tested across three regions – England, Hesse (Germany), and Molise (Italy) – covering a total area of 180,000 km2. The methodology utilized the Atlas of Hillforts of Britain and Ireland to train a CNN on LiDAR datasets and tested the model’s transferability to Germany and Italy. Techniques such as pseudo-labelling and fine-tuning addressed the “Model Drift” problem, improving region-specific performance. The AI classifier achieved F1 scores ranging from 34–38%, demonstrating its adaptability to diverse landscapes, including the Mediterranean terrain of Molise and Hesse’s densely forested regions. Case studies identified new potential hillforts in England and promising candidates in Hesse and Molise, underscoring the effectiveness of the approach. While automation significantly reduces manual workload, human verification remains critical for refining AI predictions and addressing false positives. This study also applies different expert validation workflows, emphasizing their efficiency and adaptability to regional differences. By combining automated detection with expert review, the research showcases the potential for scalable, AI-assisted archaeological discovery across diverse landscapes, providing a valuable tool for academic research and heritage management. |
|---|---|
| ISSN: | 2514-8362 |