Search for: All records

To date, Deep Learning models for archaeological feature detection have generally been built on the back of off-the-shelf convolutional neural networks (CNNs) and vision Transformer (ViT) models, which are pretrained on a variety of image types, sources, and subjects that are not specific to analyzing high-resolution satellite imagery. Recent advances in transformer-based vision models and self-supervised training approaches make it possible for researchers to generate foundation models that are more finely attuned to specific domains, without huge amounts of human-annotated training data. We discuss the development of two such models employing Meta's transformer-based DINOv2 framework. The first, DeepAndes, is based on the ingestion of a 3 million chip sample from a two million square km area of high-resolution multispectral satellite imagery of the Andean region. This foundation model has broad utility across the social and earth sciences. The second, DeepAndesArch is fine-tuned labeled archaeological training data collected by the GeoPACHA project to create an archaeology-focused version of DeepAndes. We present the processes involved in generating DeepAndes and DeepAndesArch and discuss prospects for foundation models in archaeological research 

The advent of AI-based imagery survey presents the opportunity to explore new kinds of questions about large scale archaeological distributions. Such questions are not only different in degree (scale) but in kind; they require new modes of inquiry, not unlike how “distant reading” of texts en masse is a different mode of textual analysis from traditional textual reading and hermeneutics. Here, we explore distant reading of the archaeological record by first delineating categories of inquiry, such as human ecodynamics and human-environment coupled systems approaches, settlement pattern analysis, and network-based analysis. We present initial results from our large AI-Assisted imagery survey spanning much of the Andean region, which documented in excess of a million features via object detection techniques, and mass characterization of archaeological landscapes via semantic segmentation techniques. These prospects toward continental-scale views of patterns and processes would be impossible in the absence of such continuous coverage beyond the scale of field-based methodologies. We thus advocate for the value of such perspectives as complementary and additive rather to traditional archaeological modes of analysis. 

Abstract Recent archaeological research in the Andes suggests that Indigenous herders carefully managed their environments through the modification of local hydrology and vegetation. However, the limited geographical scale of previous research makes it challenging to assess the range and prevalence of pastoralist land management in the Andes. In this article, the authors utilise large-scale, systematic imagery survey to examine the distribution and environmental contexts of corrals and pastoralist settlements in Huancavelica, Peru. Results indicate that corrals and pastoralist settlements cluster around colonial and present-day settlements and that a statistically significant relationship exists between pastoral infrastructure and perennial vegetation. This highlights the utility of remote survey for the identification of trans-regional patterns in herder-environment relationships that are otherwise difficult to detect. 

Archaeological surveys conducted through the inspection of high-resolution satellite imagery promise to transform how archaeologists conduct large-scale regional and supra-regional research. However, conducting manual surveys of satellite imagery is labour- and time-intensive, and low target prevalence substantially increases the likelihood of miss-errors (false negatives). In this article, the authors compare the results of an imagery survey conducted using artificial intelligence computer vision techniques (Convolutional Neural Networks) to a survey conducted manually by a team of experts through the Geo-PACHA platform (for further details of the project, see Wernkeet al. 2023). Results suggest that future surveys may benefit from a hybrid approach—combining manual and automated methods—to conduct an AI-assisted survey and improve data completeness and robustness. 

The north coast of Peru is among the most extensively surveyed regions in the world, yet variation in research questions, sampling strategies and chronological and geospatial controls among survey projects makes comparison of disparate datasets difficult. To contextualise these issues, the authors present a systematic survey of satellite imagery focusing on hilltop fortifications in the Jequetepeque and Santa Valleys. This digital recontextualisation of pedestrian survey data demonstrates the potential of hybrid methodologies to substantially expand both the identification of archaeological sites within difficult terrain and, consequently, our understanding of the function of defensive sites. 

Archaeology has long faced fundamental issues of sampling and scalar representation. Traditionally, the local-to-regional-scale views of settlement patterns are produced through systematic pedestrian surveys. Recently, systematic manual survey of satellite and aerial imagery has enabled continuous distributional views of archaeological phenomena at interregional scales. However, such ‘brute force’ manual imagery survey methods are both time- and labour-intensive, as well as prone to inter-observer differences in sensitivity and specificity. The development of self-supervised learning methods (e.g. contrastive learning) offers a scalable learning scheme for locating archaeological features using unlabelled satellite and historical aerial images. However, archaeological features are generally only visible in a very small proportion relative to the landscape, while the modern contrastive-supervised learning approach typically yields an inferior performance on highly imbalanced datasets. In this work, we propose a framework to address this long-tail problem. As opposed to the existing contrastive learning approaches that typically treat the labelled and unlabelled data separately, our proposed method reforms the learning paradigm under a semi-supervised setting in order to fully utilize the precious annotated data (<7% in our setting). Specifically, the highly unbalanced nature of the data is employed as the prior knowledge in order to form pseudo negative pairs by ranking the similarities between unannotated image patches and annotated anchor images. In this study, we used 95,358 unlabelled images and 5,830 labelled images in order to solve the issues associated with detecting ancient buildings from a long-tailed satellite image dataset. From the results, our semi-supervised contrastive learning model achieved a promising testing balanced accuracy of 79.0%, which is a 3.8% improvement as compared to other state-of-the-art approaches.