A modified Mask region‐based convolutional neural network approach for the automated detection of archaeological sites on high‐resolution light detection and ranging‐derived digital elevation models in the North German Lowland

Abstract

Due to complicated backgrounds and unclear target orientation, automated object detection is difficult in the field of archaeology. Most of the current convolutional neural network (CNN) object-oriented detection techniques are based on a faster region-based CNN (R-CNN) and other one-stage detectors that often lack adequate processing speeds and detection accuracies. Recently, the two-stage detector Mask R-CNN technique achieved impressive results in object detection and instance segmentation problems and was successfully applied in the analysis of archaeological airborne laser scanning (ALS) data. In this study, we outline a modified Mask R-CNN technique that reliably and efficiently detects relict charcoal hearth (RCH) sites on light detection and ranging (LiDAR) data-based digital elevation models (DEMs).Using image augmentation and image preprocessing steps combined with the deep learning-based adaptive gradient method with a dynamic bound on the learning rate (AdaBound) optimization technique, we could improve the model's accuracy and significantly reduce its training time. We use DEMs based on high-resolution LiDAR data and the visualization for archaeological topography (VAT) technique that give images with a very strong contrast of the terrain and the outline of the sites of interest in the North German Lowland. Therefore, the model can identify RCH sites with an average recall of 83% and an average precision of 87%. Techniques such as the modified Mask R-CNN method outlined here will help to greatly improve our knowledge about archaeological site densities in the realm of historical charcoal production and past human-landscape interactions. This method provides an accurate, time-efficient and bias-free large-scale site mapping option not only for the North German Lowland but potentially for other landscapes as well.

Machine Learning Arrives in Archaeology

“…The increasing availability of large-scale lidar, satellite, and aerial imagery on local, regional, and national scales, however, is transforming archaeology around the globe-particularly the searching and mapping of archaeological sites (Figure 2). ML algorithms can be used to process the geospatial data in the search for sites in diverse environments (Bonhage et al 2021;Caspari and Crespo 2019;Davis 2019;Davis, DiNapoli, et al 2020;Davis, Seeber, et al 2020;Evans and Hofer 2019;Guyot et al 2018Guyot et al , 2021Orengo et al 2020;Soroush et al 2020;Thabeng et al 2019;Trier et al 2018Trier et al , 2019 Verschoof-van der Vaart and Lambers 2019; Verschoof-van der Vaart et al 2020).…”

Uni-Temporal Multispectral Imagery for Burned Area Mapping with Deep Learning

“…Semantic segmentation architectures with convolutional neural networks (CNNs) have unique characteristics to extract the contextual information in multiple scales and then label each pixel of an image [37]. It is now playing a significant role in many image analysis tasks ranging from autonomous vehicles, human-computer interaction, robotics, medical research, precision farming, and so on [37][38][39][40][41][42][43]. For remote sensing, semantic segmentation algorithms have recently been applied to 2D satellite images and even 3D scenes [44,45].…”

“…A similar U-Net architecture with residual units was employed for road area extraction with relatively high accuracy [50]. A Mask Region-Based CNN(R-CNN) model was applied to automatically mapping applications such as ice-wedge polygons [41,42] and archaeological sites [43] with high-resolution or VHR remote sensing imagery.…”

When Computers Dream of Charcoal

“…Primary among these are RCHs, but others include the remains of roads and colliers' huts. Importantly, RCHs have been recognized using derivatives of high-resolution lidar scanning (e.g., Bonhage et al 2020Bonhage et al , 2021Carter 2019a;Donovan et al 2021;Hirsch et al 2017;Johnson andOuimet 2018, 2021;Kazimi et al 2019;Raab et al 2015;Risbøl et al 2013 et al 2016;Schneider et al 2015). To simplify, lidar is a technology that uses plane-mounted lasers to measure the altitude of the earth's surface and create a point cloud of the surface (for a detailed description, see Fernandez-Diaz et al 2014;Opitz 2013; see also Opitz and Hermann [2018] for a review of remote sensing in archaeology).…”

“…We selected Waleed Abdulla's Mask R-CNN as a deep-learning platform for object detection (see also Bonhage et al 2021). Mask R-CNN is suitable for identifying objects that appear only in a small part of an image (Brownlee 2019a(Brownlee , 2019bHe et al 2020;Rosebrock 2019), such as RCHs in images of much larger landscapes.…”