Integrating Remote Sensing, Machine Learning, and Citizen Science in Dutch Archaeological Prospection

Abstract

Although the history of automated archaeological object detection in remotely sensed data is short, progress and emerging trends are evident. Among them, the shift from rule-based approaches towards machine learning methods is, at the moment, the cause for high expectations, even though basic problems, such as the lack of suitable archaeological training data are only beginning to be addressed. In a case study in the central Netherlands, we are currently developing novel methods for multi-class archaeological object detection in LiDAR data based on convolutional neural networks (CNNs). This research is embedded in a long-term investigation of the prehistoric landscape of our study region. We here present an innovative integrated workflow that combines machine learning approaches to automated object detection in remotely sensed data with a two-tier citizen science project that allows us to generate and validate detections of hitherto unknown archaeological objects, thereby contributing to the creation of reliable, labeled archaeological training datasets. We motivate our methodological choices in the light of current trends in archaeological prospection, remote sensing, machine learning, and citizen science, and present the first results of the implementation of the workflow in our research area.

Searching for Viking Age Fortresses with Automatic Landscape Classification and Feature Detection

“…As further automatic identification of larger areas can produce numerous points of interest, a natural next step for research in surface feature detection is to include citizen science for ground-truthing, as demonstrated by a recent Dutch study [63].…”

“…Automatic feature detection approaches for archaeology are a rapidly and fast developing field in archaeological prospection [60][61][62]. Recent studies have demonstrated machine learning methods for detecting a wide variety of features, including burial mounds, charcoal kilns, buildings, and field systems [10,[63][64][65][66].…”

Mapping and Damage Assessment of “Royal” Burial Mounds in the Siberian Valley of the Kings

“…Since mound-like archaeological features are a wide-spread and relatively homogenous phenomenon, archaeological landscapes with mounds are an ideal case for semi-automated and automated methods of classification and detection. Such methods were specifically applied with LiDAR-derived digital elevation models based on convolutional neural networks for identifying Neolithic mounds [13,14], and object-based approaches to find earthen mounds and shell heaps [15]. Other approaches used high-resolution optical satellite data, like Ikonos-2, and applied a random forests algorithm [16] or developed a training dataset for a neural network based on Google Earth images [17].…”

Global-Scale Archaeological Prospection using CORONA Satellite Imagery: Automated, Crowd-Sourced, and Expert-led Approaches

“…Recent research has sought to replace rule-based automated detection with machine learning-based approaches that instead begin with a large training sample of sites and allow for a more flexible set of criteria defining what constitutes a "site" to be iteratively established (Lambers, Verschoof-van der Vaart, and Bourgeois 2019; Zingman et al 2016). While such approaches have promise, they remain in early stages of development, and are often hamstrung by the complexity and inconsistency of archaeological site databases upon which they are intended to be built (e.g., Lambers, Verschoof-van der Vaart, and Bourgeois 2019;Sadr 2016). Despite the effort being devoted to development of automated and machine-learning approaches to identification of archaeological sites and features in remotely sensed imagery, nearly all of these studies produce a large number of false positives as well as missing many known sites, and cannot yet outperform trained human analysts in terms of accuracy.…”