How new technologies can help extract archaeological information from historical maps

British maps of modern Pakistan (left) and Syria (right) depicting thousands of potential archaeological sites inadvertently, as topographic anomalies; on purpose, using conventional sites or identified using toponymic references. Image credit: Arnau Garcia-Molsosa.

New research using Deep Learning to extract archaeological information from collections of maps produced during the European colonization of South Asia and Levant 

New research published this week in Archaeological Prospection presents the results of how archaeological information can be extracted from historical maps using deep learning. The paper demonstrates the potential of this technique for large-scale analysis and showcases its limitations and challenges. 

Colonial powers undertook massive programmes of systematic mapping of their possessions for military and economic control purposes. Although the historical value of these maps is well known by archaeologists and historians, exhaustive archival work is needed to use the data they contain, including the thousands of potential locations of archaeological sites that can be inferred from those maps.

This research provides a new tool to facilitate this work and render it accessible to the research community. It explores collections of maps produced by the Survey of India during the British dominion over modern India and Pakistan, and by the French authorities during the control of modern Syria. Using deep learning it is now possible to extract information about 1) mound sites, unintentionally represented as topographical anomalies by the surveyors; 2) sites purposely reported and represented as conventional signs; and 3) toponyms that might refer to archaeological sites. 

This technique provides the potential of exploring large series of hundreds of maps in a quick and reliable way. That can be an invaluable tool for the management of cultural heritage and the development of historical and archaeological research. 

The paper is the result of a collaboration between our researchers Dr Arnau Garcia-Molsosa (agarcia@icac.cat) and Dr Hector A. Orengo (horengo@icac.cat) and colleagues from  the University of Cambridge (Dr Cameron Petrie) and the University of Durham (Dr Dan Lawrence, Dr Kristen Hopper and Dr Graham Phillip).

The paper, entitled: ‘Potential of deep learning segmentation for the extraction of archaeological features from historical map series’, is published Open Access and can be accessed here: https://doi.org/10.1002/arp.1807

Funding information: Agència de Gestió d’Ajuts Universitaris i de Recerca, Grant/Award Number: 2018 BP 00208; H2020 European Research Council, Grant/Award Number: GA648609; H2020 Marie Skłodowska‐Curie Actions, Grant/Award Number: GA746446; Leverhulme Trust, Grant/Award Number: F00128/AR; Ministerio de Ciencia e Innovación, Grant/Award Number: RYC‐2016‐19637

Tags: , , , , , , ,