Automatic Die Studies for Ancient Numismatics
Clément Cornet, Héloïse Aumaître, Romaric Besançon, Julien Olivier, Thomas Faucher, Hervé Le Borgne
TL;DR
This work tackles the automation of die studies in ancient numismatics, addressing the prohibitive manual effort required to compare large coin corpora. It introduces a fully automatic pipeline that computes robust cross-coin similarities using XFeat features filtered by MAGSAC++, then clusters coins into dies via Adaptive Graph Label Propagation with an intrinsic silhouette-based threshold. Across two Greek coin datasets, the method significantly outperforms prior CADS-based automation while maintaining practical runtimes, enabling large-scale die studies. By releasing code and models, the approach offers a reproducible, scalable tool for numismatists to derive historical insights from much larger coin collections.
Abstract
Die studies are fundamental to quantifying ancient monetary production, providing insights into the relationship between coinage, politics, and history. The process requires tedious manual work, which limits the size of the corpora that can be studied. Few works have attempted to automate this task, and none have been properly released and evaluated from a computer vision perspective. We propose a fully automatic approach that introduces several innovations compared to previous methods. We rely on fast and robust local descriptors matching that is set automatically. Second, the core of our proposal is a clustering-based approach that uses an intrinsic metric (that does not need the ground truth labels) to determine its critical hyper-parameters. We validate the approach on two corpora of Greek coins, propose an automatic implementation and evaluation of previous baselines, and show that our approach significantly outperforms them.