Locating Charcoal Production Sites in Sweden Using LiDAR, Hydrological Algorithms, and Deep Learning

Abstract

Over the past several centuries, the iron industry played a central role in the economy of Sweden and much of northern Europe. A crucial component of iron manufacturing was the production of charcoal, which was often created in charcoal piles. These features are visible in LiDAR (light detection and ranging) datasets. These charcoal piles vary in their morphology by region, and training data for some feature types are severely lacking. Here, we investigate the potential for machine automation to aid archaeologists in recording charcoal piles with limited training data availability in a forested region of Jönköping County, Sweden. We first use hydrological depression algorithms to conduct a preliminary assessment of the study region and compile suitable training data for charcoal production sites. Then, we use these datasets to train a series of RetinaNet deep learning models, which are less computationally expensive than many popular deep learning architectures (e.g., R-CNNs), allowing for greater usability. Together, our results demonstrate how charcoal piles can be automatically extracted from LiDAR datasets, which has great implications for improving our understanding of the long-term environmental impact of the iron industry across Northern Europe. Furthermore, our workflow for developing and implementing deep learning models for archaeological research can expand the use of such methods to regions that lack suitable training data.

Automated large‐scale mapping and analysis of relict charcoal hearths in Connecticut (USA) using a Deep Learning YOLOv4 framework

“…Previous research in an effort to automatically detect RCHs has relied on various methods including Template Matching (Schneider et al, 2015; Trier & Pilø, 2012) and Geographic Object‐Based Image Analysis (Witharana et al, 2018), whereas more recently, Machine Learning approaches are being developed and utilized (Anderson, 2019; Bonhage et al, 2021; Carter et al, 2021; Davis & Lundin, 2021; Kazimi et al, 2020, 2019; Oliveira et al, 2021; Suh et al, 2021; Trier et al, 2021, 2018; Verschoof‐van der Vaart et al, 2020). Deep Learning (LeCun et al, 2015), a subfield of Machine Learning, predominantly employs Convolutional Neural Networks (CNNs)—hierarchically structured algorithms that generally consist of a (image) feature extractor and classifier (Guo et al, 2016)—that learn to generalize from a large set of labelled examples, rather than relying on a human operator to set parameters or formulate rules.…”

“…Previous research in an effort to automatically detect RCHs has relied on various methods including Template Matching (Schneider et al, 2015;Trier & Pilø, 2012) and Geographic Object-Based Image Analysis (Witharana et al, 2018), whereas more recently, Machine Learning approaches are being developed and utilized (Anderson, 2019;Bonhage et al, 2021;Carter et al, 2021;Davis & Lundin, 2021;Kazimi et al, 2020Kazimi et al, , 2019Oliveira et al, 2021;Suh et al, 2021;Trier et al, 2021Trier et al, , 2018 (Guo et al, 2016)-that learn to generalize from a large set of labelled examples, rather than relying on a human operator to set parameters or formulate rules. To date, these automated methods are mainly tested in an experimental setting but have yet to be applied in various contexts or on a large (e.g., regional or national) scale (Verschoof-van der Vaart et al, 2020; but see for instance Berganzo-Besga et al, 2021;Orengo et al, 2020), with this being the aim of previous initiatives (Trier et al, 2019).…”

Evaluating Mask R‐CNN models to extract terracing across oceanic high islands: A case study from Sāmoa

“…ArcGIS Pro contains built‐in libraries for deep learning, which have been used successfully for archaeological applications (e.g., Agapiou et al, 2021; Bickler & Jones, 2021; Davis et al, 2021; Davis & Lundin, 2021). To implement deep learning within the ArcGIS Pro environment, input data must be a multiband raster.…”

What Should We Do With These? Challenges related to (semi-)automatically detected sites and features. A note

“…Recent advances in deep learning techniques and improved availability of highresolution aerial laser scanning datasets have brought semi-automatic detection of archaeological features within reach of an increasing number of research groups and institutions (see e.g. Anttiroiko et al 2023;Bonhage et al 2021;Davis and Lundin 2021;Suh et al 2021;Snitker et al 2022;Trier et al 2021;Verschoof-van der Vaart and Lambers 2019). Such techniques make it possible to detect and extract information about very large numbers of archaeologically relevant features over potentially vast areas in a highly efficient manner and, hence, are likely to have a significant positive impact on the amount and quality of data available to heritage management institutions.…”