Test data supporting the paper 'Interactive, shallow machine-learning based semantic segmentation of 2D and 3D geophysical data from archaeological sites'

This repository comprises three geophyiscal test datasets supporting the paper 'Interactive, shallow machine-learning based semantic segmentation of 2D and 3D geophysical data from archaeological sites' by Lieven Verdonck, Michel Dabas and Marc Bui, Remote Sensing, 2025, 17, 3092, https://www.mdpi.com/2072-4292/17/17/3092. Technological developments in archaeological geophysics have led to growing data volumes, so that an important bottleneck is now at the stage of data interpretation. Manual delineation and classification of the significant anomalies are time-consuming. In the abovementioned paper, we describe how shallow machine learning (random forests) can be used to interpret near-surface geophysical data from archaeological sites. We show their potential by applying them to the three datasets in this Zenodo repository, and discuss the limitations and possible further improvements. Please see the instructions in each zip-folder for more information on how to use the data. For information on the provenance of the data and on the methodology of data acquisition and processing, see the paper and the supplementary data.For the software used to perform the machine-learning based semantic segmentation, see also https://github.com/lrverdon/Shallow-Machine-Learning-for-Archaeological-Geophysics.