Automated Detection of Prehistoric Tumuli in the Sahara: A Deep Learning Approach to Satellite Imagery
Loading...
Date
2025
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
The Eurographics Association
Abstract
This study focuses on the automated satellite-based identification, classification, and spatial interpretation of prehistoric tumuli in the Central Western Sahara. These dry-stone structures lie in environmentally and geopolitically challenging areas for fieldwork and are therefore largely understudied. They are commonly interpreted as funerary monuments, ritual installations, or territorial markers. Regardless of individual function, they often represent the only evidence of Mid- and Late Holocene nomadic Saharan pastoralists (reflecting identities, subsistence strategies, and mobility patterns) and frequently appear as recurrent visitation points near environmentally attractive areas such as pathways, pastures, rivers, and water ponds. Among the various types of Saharan dry-stone buildings, this study selects four specific archaeological classes, previously documented through fieldwork and remote sensing surveys, that are highly visible and recognizable in satellite imagery. The research adopts an innovative methodological framework combining High- and Very High-Resolution satellite datasets (optical, multispectral, and SAR-derived DEM), image enhancement techniques (spectral indices), a Deep Convolutional Neural Network, clustering methods, spatial analyses, and Least Cost Path modelling to systematically detect, classify, and interpret these features at a regional scale.
Description
@inproceedings{10.2312:dh.20253386,
booktitle = {Digital Heritage},
editor = {Campana, Stefano and Ferdani, Daniele and Graf, Holger and Guidi, Gabriele and Hegarty, Zackary and Pescarin, Sofia and Remondino, Fabio},
title = {{Automated Detection of Prehistoric Tumuli in the Sahara: A Deep Learning Approach to Satellite Imagery}},
author = {Brucato, Alessia},
year = {2025},
publisher = {The Eurographics Association},
ISBN = {978-3-03868-277-6},
DOI = {10.2312/dh.20253386}
}