Interactive, Shallow Machine Learning-Based Semantic Segmentation of 2D and 3D Geophysical Data from Archaeological Sites

Aug 5, 2026

Recent decades have increased data volumes in archaeological geophysics, creating a bottleneck at the interpretation stage. Manual delineation and anomaly classification are time-consuming, prompting (semi-)automatic image segmentation approaches. This work evaluates shallow machine learning, using random forests instead of deep convolutional neural networks, highlighting suitability for low contrast, limited training data, and the ability to process 3D data. Pixel-level classification employs ilastik with a Jupyter Notebook pipeline and compares RF outputs to manual interpretation using mean intersection over union.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages33

Reading Time1 h 23 min

AccessFree to download

Abstract

Introduction

Geophysical techniques for archaeological prospection

Article Interactive, Shallow Machine Learning-Based Semantic Segmentation of 2D and 3D Geophysical Data from Archaeological Sites Lieven Verdonck 1,2,3,*, Michel Dabas 2 and Marc Bui 4 Academic Editor: Timo Balz Received: 15 May 2025 Revised: 21 August 2025 Accepted: 29 August 2025 Published: 4 September 2025 Citation: Verdonck, L.; Dabas, M.; Bui, M. Interactive, Shallow Machine Learning-Based Semantic Segmentation of 2D and 3D Geophysical Data from Archaeological Sites. Remote Sens. 2025, 17, 3092 . [https://doi.org/10.3390/](https://doi.org/10.3390/)[ ](https://doi.org/10.3390/)rs17173092 Copyright: © 2025 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license ([https://creativecommons.org/](https://creativecommons.org/)[ ](https://creativecommons.org/)[licenses/by/4.0/](licenses/by/4.0/)) . 1 Faculty of Classics, University of Cambridge, Cambridge CB3 9DA, UK 2 Archéologie et Philologie d’Orient et d’Occident, UMR 8546 CNRS-ENS-EPHE (PSL), École Normale Supérieure, 75230 Paris CEDEX 05, France; [michel.dabas@ens.psl.eu](michel.dabas@ens.psl.eu) 3 Department of Archaeology, Ghent University, 9000 Ghent, Belgium 4 Archéologie et Philologie d’Orient et d’Occident, UMR 8546 CNRS-ENS-EPHE (PSL), École Pratique des Hautes Études, 75230 Paris CEDEX 05, France; [marc.bui@ephe.psl.eu](marc.bui@ephe.psl.eu) * Correspondence: [lrmv2@cam.ac.uk](lrmv2@cam.ac.uk) Abstract In recent decades, technological developments in archaeological geophysics have led to growing data volumes, so that an important bottleneck is now at the stage of data interpretation. The manual delineation and classification of anomalies are time-consuming, and different methods for (semi-)automatic image segmentation have been proposed, based on explicitly formulated rulesets or deep convolutional neural networks (DCNNs) . So far, these have not been used widely in archaeological geophysics because of the complexity of the segmentation task (due to the low contrast between archaeological structures and background and the low predictability of the targets) . Techniques based on shallow machine learning (e.g., random forests, RFs) have been explored very little in archaeological geophysics, although they are less case-specific than most rule-based methods, do not require large training sets as is the case for DCNNs, and can easily handle 3D data. In this paper, we show their potential for geophysical data analysis. For the classification on the pixel level, we use ilastik, an open-source segmentation tool developed in medical imaging. Algorithms for object classification, manual reclassification, post-processing, vectorisation, and georeferencing were brought together in a Jupyter Notebook, available on GitHub (version 7.3.2) . To assess the accuracy of the RF classification applied to geophysical datasets, we compare it with manual interpretation. A quantitative evaluation using the mean intersection over union metric results in scores of ~60%, which only slightly increases after the manual correction of the RF classification results. Remarkably, a similar score results from the comparison between independent manual interpretations. This observation illustrates that quantitative metrics are not a panacea for evaluating machine-generated geophysical data interpretation in archaeology, which is characterised by a significant degree of uncertainty. It also raises the question of how the semantic segmentation of geophysical data (whether carried out manually or with the aid of machine learning) can best be evaluated. Keywords: geophysics; archaeology; semantic segmentation; shallow machine learning; random forest; geophysical data interpretation; semi-automated interpretation; archaeological prospection; ground-penetrating radar survey; magnetometry 1. Introduction Geophysical methods record variations in

FAQs

Why is semantic segmentation important in archaeological geophysics?

Geophysical surveys produce large datasets, and interpreting anomalies requires delineation and classification. Semantic segmentation reduces time costs by automating parts of this workflow.

What is the main modeling approach evaluated in the paper?

The study focuses on shallow machine learning, particularly random forests, instead of deep convolutional neural networks, emphasizing performance under low contrast and limited training data.

How is segmentation accuracy assessed?

Accuracy is quantified using mean intersection over union and compared to manual interpretation, including checks against independent manual assessments.

Aug 5, 2026

Upload by Evangeline