Using photogrammetry for the objective study of ancient bowed instruments: a machine learning approach

The morphology of today's violin differs greatly from that of the first instruments of the late 16th century. Between 1750 and 1850, in order to meet the standards suggested by famous orchestras and conservatories, many ancient violins have been reduced. Given the paucity of written sources, the only way to improve our understanding of the violin family in this period is to study the instruments themselves. Hence, we aim at developing a set of tools able to analyse three-dimensional geometrical shapes acquired by photogrammetry. Our purpose is to differentiate ancient reduced violins from ancient unreduced violins. Using their geometrical representation, we want to quantify their specific characteristics, detect possible anomalies and, if applicable, reconstruct their original morphology. Ultimately, we will classify a corpus of ancient violins according to these aspects, thanks to unsupervised and supervised machine learning techniques