Speaker
Description
Numismatic die studies are notoriously labor intensive to conduct by hand, and often take years to complete. Recent years have seen an uptick in computational approaches to die studies, taking advantage of computer vision and unsupervised clustering techniques to both improve accuracy and greatly reduce time required. This paper presents recent developments in the Computer-Aided Die Study (CADS), one such project that aims not to replace the numismatists’ role, but instead to aid and empower their efforts. CADS provides a suite of tools that aid in conducting die studies, ranging from initial high-accuracy die clustering to integrating new material into a completed study. CADS is an ongoing project at the American Numismatic Society, and is a continuation of previous work in collaboration with Trinity University.
