Claudine Gravel-Miguel
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Claudine Gravel-Miguel has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
Curtis W. Marean
3 shared publicationsColin D. Wren
Univ Colorado, Dept Anthropol, 114 Centennial Hall,1420 Austin Bluffs Pkwy, Colorado Springs, CO 80918 USA
2 shared publicationsGrant Snitker
New Mexico Consortium
2 shared publications- John K. Murray
Institute of Human Origins, School of Human Evolution and Social Change, Arizona State University, Tempe, AZ, USA
2 shared publications - Julien Riel-Salvatore
Département d'Anthropologie, Université de Montréal, Canada,
2 shared publications - Katherine Peck
Department of Anthropology University of New Mexico Albuquerque New Mexico USA
2 shared publications
Latest Publications
- Open Access
- 27 April 2026
- Published by Cambridge University Press (CUP) in Advances in Archaeological Practice
- Vol. 14 (2) , 179-197
Using Simulated Training Data to Locate Archaeological Sites with Machine Learning
Archaeologists have demonstrated the value of deep learning models for detecting archaeological objects in lidar data. As landscape-level projects become the norm, archaeological data derived from deep learning predictions can be integrated into these initiatives through coupled natural-cultural landscapes planning. However, the paucity of archaeological training datasets limits the application of deep learning models to relatively common and well-documented object classes. Using procedurally generated training datasets may be one approach to overcome this bottleneck. To test the efficacy of procedural generation for developing deep learning training data, we trained models to detect a novel object class (hypothesized historic tar kilns) in the Kisatchie National Forest in Louisiana. We developed two procedural generation approaches to embed simulated archaeological objects in a lidar-derived DEM and used these datasets to train deep learning (Mask R-CNN) models. We then evaluated model predictions within lidar-derived visualizations and during field survey. Our trained models detected targets with high recall but low precision. Field investigation suggested that the objects were not tar kilns but a different historic feature class. This study suggests that models trained on simulated objects are a useful addition to lidar analysis tool kits and can be directly integrated into archaeological field investigation workflows.
- 27 April 2026
- Published by Cambridge University Press (CUP) in Advances in Archaeological Practice
- Vol. 14 (2) , 179-197
Open AccessUsing Simulated Training Data to Locate Archaeological Sites with Machine Learning
Archaeologists have demonstrated the value of deep learning models for detecting archaeological objects in lidar data. As landscape-level projects become the norm, archaeological data derived from deep learning predictions can be integrated into these initiatives through coupled natural-cultural landscapes planning. However, the paucity of archaeological training datasets limits the application of deep learning models to relatively common and well-documented object classes. Using procedurally generated training datasets may be one approach to overcome this bottleneck. To test the efficacy of procedural generation for developing deep learning training data, we trained models to detect a novel object class (hypothesized historic tar kilns) in the Kisatchie National Forest in Louisiana. We developed two procedural generation approaches to embed simulated archaeological objects in a lidar-derived DEM and used these datasets to train deep learning (Mask R-CNN) models. We then evaluated model predictions within lidar-derived visualizations and during field survey. Our trained models detected targets with high recall but low precision. Field investigation suggested that the objects were not tar kilns but a different historic feature class. This study suggests that models trained on simulated objects are a useful addition to lidar analysis tool kits and can be directly integrated into archaeological field investigation workflows.
- Open Access
- 1 May 2025
- Published by Cambridge University Press (CUP) in Advances in Archaeological Practice
- Vol. 13 (2) , 291-314
Semantic Segmentation of Archaeological Features on Public Lands: Case Study of Historical Cotton Terraces within the Piedmont National Wildlife Refuge, Georgia, USA
The logistics, costs, and capacity needed to complete extensive archaeological pedestrian surveys to inventory cultural resources present challenges to public land managers. To address these issues, we developed a workflow combining lidar-derived imagery and deep learning (DL) models tailored for cultural resource management (CRM) programs on public lands. It combines Python scripts that fine-tune models to recognize archaeological features in lidar-derived imagery with denoising QGIS steps that improve the predictions’ performance and applicability. We present this workflow through an applied case study focused on detecting historic agricultural terraces in the Piedmont National Wildlife Refuge, Georgia, USA. For this project, we fine-tuned pretrained U-Net models to teach them to recognize agricultural terraces in imagery, identified the parameter settings that led to the highest recall for detecting terraces, and used those settings to train models on incremental dataset sizes, which allowed us to identify the minimum training size necessary to obtain satisfying models. Results present effective models that can detect most terraces even when trained on small datasets. This study provides a robust methodology that requires basic proficiencies in Python coding but expands DL applications in federal CRM by advancing methods in lidar and machine learning for archaeological inventorying, monitoring, and preservation.
- 1 May 2025
- Published by Cambridge University Press (CUP) in Advances in Archaeological Practice
- Vol. 13 (2) , 291-314
Open AccessSemantic Segmentation of Archaeological Features on Public Lands: Case Study of Historical Cotton Terraces within the Piedmont National Wildlife Refuge, Georgia, USA
The logistics, costs, and capacity needed to complete extensive archaeological pedestrian surveys to inventory cultural resources present challenges to public land managers. To address these issues, we developed a workflow combining lidar-derived imagery and deep learning (DL) models tailored for cultural resource management (CRM) programs on public lands. It combines Python scripts that fine-tune models to recognize archaeological features in lidar-derived imagery with denoising QGIS steps that improve the predictions’ performance and applicability. We present this workflow through an applied case study focused on detecting historic agricultural terraces in the Piedmont National Wildlife Refuge, Georgia, USA. For this project, we fine-tuned pretrained U-Net models to teach them to recognize agricultural terraces in imagery, identified the parameter settings that led to the highest recall for detecting terraces, and used those settings to train models on incremental dataset sizes, which allowed us to identify the minimum training size necessary to obtain satisfying models. Results present effective models that can detect most terraces even when trained on small datasets. This study provides a robust methodology that requires basic proficiencies in Python coding but expands DL applications in federal CRM by advancing methods in lidar and machine learning for archaeological inventorying, monitoring, and preservation.
- Open Access
- 16 October 2024
- Published by Center for Open Science in OSF Preprints
Semantic segmentation of archaeological features on public lands – Case study of historical cotton terraces within the Piedmont National Wildlife Refuge, GA, USA
The logistics, costs, and capacity needed to complete extensive archaeological pedestrian surveys to inventory cultural resources present challenges to land managers. To address these issues, we developed a workflow combining LiDAR-derived imagery and Deep Learning models tailored for Cultural Resource Management (CRM) programs on public lands. This workflow combines Python scripts that easily fine-tune models to recognize archaeological features in LiDAR-derived imagery with denoising QGIS steps that improve the predictions' performance and applicability. We present this workflow through an applied case study focused on detecting historic agricultural terraces in the Piedmont National Wildlife Refuge, Georgia, USA. We fine-tune pre-trained U-Net models to teach them to recognize agricultural terraces in LiDAR-derived imagery. We test different parameter setting combinations, identify the settings that lead to the highest recall for detecting terraces, and use those settings to train models on incremental dataset sizes to identify the minimum training size necessary to obtain satisfying models. Results present very effective models even when trained on small datasets. This study provides a robust methodology that is sensitive enough to detect most terraces. It also expands Deep Learning applications in federal CRM by advancing methods in LiDAR and machine learning for archeological inventorying, monitoring, and preservation.
- 16 October 2024
- Published by Center for Open Science in OSF Preprints
Open AccessSemantic segmentation of archaeological features on public lands – Case study of historical cotton terraces within the Piedmont National Wildlife Refuge, GA, USA
The logistics, costs, and capacity needed to complete extensive archaeological pedestrian surveys to inventory cultural resources present challenges to land managers. To address these issues, we developed a workflow combining LiDAR-derived imagery and Deep Learning models tailored for Cultural Resource Management (CRM) programs on public lands. This workflow combines Python scripts that easily fine-tune models to recognize archaeological features in LiDAR-derived imagery with denoising QGIS steps that improve the predictions' performance and applicability. We present this workflow through an applied case study focused on detecting historic agricultural terraces in the Piedmont National Wildlife Refuge, Georgia, USA. We fine-tune pre-trained U-Net models to teach them to recognize agricultural terraces in LiDAR-derived imagery. We test different parameter setting combinations, identify the settings that lead to the highest recall for detecting terraces, and use those settings to train models on incremental dataset sizes to identify the minimum training size necessary to obtain satisfying models. Results present very effective models even when trained on small datasets. This study provides a robust methodology that is sensitive enough to detect most terraces. It also expands Deep Learning applications in federal CRM by advancing methods in LiDAR and machine learning for archeological inventorying, monitoring, and preservation.
- Open Access
- 19 September 2023
- Published by Elsevier in Quaternary Science Reviews
- Vol. 318
The importance of C3 and C4 grasses and CAM shrubs in the Greater Cape Floristic Region under contemporary and Last Glacial Maximum climates
- 19 September 2023
- Published by Elsevier in Quaternary Science Reviews
- Vol. 318
Open AccessThe importance of C3 and C4 grasses and CAM shrubs in the Greater Cape Floristic Region under contemporary and Last Glacial Maximum climates
- 25 July 2023
- Published by Springer Nature
- p. 61-73
The Impact of Magdalenian Hunter-Gatherers on Their Environment
- 25 July 2023
- Published by Springer Nature
- p. 61-73
The Impact of Magdalenian Hunter-Gatherers on Their Environment
- Open Access
- 30 August 2022
- Published by Springer Nature in Journal of Archaeological Method and Theory
- Vol. 30 (3) , 757-804
The Ornaments of the Arma Veirana Early Mesolithic Infant Burial
Personal ornaments are widely viewed as indicators of social identity and personhood. Ornaments are ubiquitous from the Late Pleistocene to the Holocene, but they are most often found as isolated objects within archaeological assemblages without direct evidence on how they were displayed. This article presents a detailed record of the ornaments found in direct association with an Early Mesolithic buried female infant discovered in 2017 at the site of Arma Veirana (Liguria, Italy). It uses microscopic, 3D, and positional analyses of the ornaments as well as a preliminary perforation experiment to document how they were perforated, used, and what led to their deposit as part of the infant’s grave goods. This study provides important information on the use of beads in the Early Mesolithic, in general, as well as the relationship between beads and young subadults, in particular. The results of the study suggest that the beads were worn by members of the infant’s community for a considerable period before they were sewn onto a sling, possibly used to keep the infant close to the parents while allowing their mobility, as seen in some modern forager groups. The baby was then likely buried in this sling to avoid reusing the beads that had failed to protect her or simply to create a lasting connection between the deceased infant and her community.
- 30 August 2022
- Published by Springer Nature in Journal of Archaeological Method and Theory
- Vol. 30 (3) , 757-804
Open AccessThe Ornaments of the Arma Veirana Early Mesolithic Infant Burial
Personal ornaments are widely viewed as indicators of social identity and personhood. Ornaments are ubiquitous from the Late Pleistocene to the Holocene, but they are most often found as isolated objects within archaeological assemblages without direct evidence on how they were displayed. This article presents a detailed record of the ornaments found in direct association with an Early Mesolithic buried female infant discovered in 2017 at the site of Arma Veirana (Liguria, Italy). It uses microscopic, 3D, and positional analyses of the ornaments as well as a preliminary perforation experiment to document how they were perforated, used, and what led to their deposit as part of the infant’s grave goods. This study provides important information on the use of beads in the Early Mesolithic, in general, as well as the relationship between beads and young subadults, in particular. The results of the study suggest that the beads were worn by members of the infant’s community for a considerable period before they were sewn onto a sling, possibly used to keep the infant close to the parents while allowing their mobility, as seen in some modern forager groups. The baby was then likely buried in this sling to avoid reusing the beads that had failed to protect her or simply to create a lasting connection between the deceased infant and her community.
- 8 February 2022
- Published by Elsevier in Quaternary International
- Vol. 638-639, 148-158
Were prehistoric coastal sites more intensively occupied than inland sites? Using an agent-based model to understand the intensity of prehistoric coastal occupation, and what it means for studies on the evolution of the coastal adaptation
- 8 February 2022
- Published by Elsevier in Quaternary International
- Vol. 638-639, 148-158
Were prehistoric coastal sites more intensively occupied than inland sites? Using an agent-based model to understand the intensity of prehistoric coastal occupation, and what it means for studies on the evolution of the coastal adaptation
- Open Access
- 18 November 2021
- Published by Center for Open Science in SocArXiv
EVALUATING THE EFFECTS OF RANDOMNESS ON MISSING DATA IN ARCHAEOLOGICAL NETWORKS
Network science shows promise for archaeologists who want to explore past social dynamics using material culture. Yet, archaeological data is subject to important caveats that exist for all datasets. Almost all archaeological datasets are biased, and these biases are often unknown or only partially understood. Prior research has examined the effects of missing nodes on archaeological networks. Here, we instead focus on the impact of missing links on such networks. We used an agent-based model (ArchMatNet) to generate a simulated, unbiased assemblage of artifacts deposited at sites. We link those sites through the similarity of their artifacts to form the complete network. We also include an obsidian dataset from the US Southwest to compare differences between real and simulated data. We explore how random and nonrandom sampling of the two datasets affect the accuracy of the network reconstructed. Our analysis confirms prior research demonstrating that random samples are representative of the original network, even when they are small, but that biased samples of any size are significantly problematic. This research highlights the need to consider bias in archaeological data and demonstrates the utility of agent-based models in testing archaeological methods. Furthermore, this simulated dataset can better inform how archaeologists judge bias and will help us develop new methods to mitigate the effects of biased data.
- 18 November 2021
- Published by Center for Open Science in SocArXiv
Open AccessEVALUATING THE EFFECTS OF RANDOMNESS ON MISSING DATA IN ARCHAEOLOGICAL NETWORKS
Network science shows promise for archaeologists who want to explore past social dynamics using material culture. Yet, archaeological data is subject to important caveats that exist for all datasets. Almost all archaeological datasets are biased, and these biases are often unknown or only partially understood. Prior research has examined the effects of missing nodes on archaeological networks. Here, we instead focus on the impact of missing links on such networks. We used an agent-based model (ArchMatNet) to generate a simulated, unbiased assemblage of artifacts deposited at sites. We link those sites through the similarity of their artifacts to form the complete network. We also include an obsidian dataset from the US Southwest to compare differences between real and simulated data. We explore how random and nonrandom sampling of the two datasets affect the accuracy of the network reconstructed. Our analysis confirms prior research demonstrating that random samples are representative of the original network, even when they are small, but that biased samples of any size are significantly problematic. This research highlights the need to consider bias in archaeological data and demonstrates the utility of agent-based models in testing archaeological methods. Furthermore, this simulated dataset can better inform how archaeologists judge bias and will help us develop new methods to mitigate the effects of biased data.
- 1 June 2021
- Published by Elsevier in Journal of Human Evolution
- Vol. 155, 102981
Exploring variability in lithic armature discard in the archaeological record
- 1 June 2021
- Published by Elsevier in Journal of Human Evolution
- Vol. 155, 102981
Exploring variability in lithic armature discard in the archaeological record