Speaker: Sarah Lang, Bernhard Liebl and Manuel Burghardt
Affiliation: 1, Department Centre for Information Modelling (ZIM), University of Graz; 2, Computational Humanities Research Group, University of Leipzig
Title: Toward a Computational Historiography of Alchemy: Challenges and Obstacles of Object Detection for Historical Illustrations of Mining, Metallurgy and Distillation in 16th-17th Century Print
Abstract: This study explores the use of modern computer vision methods for object detection in historical images extracted from 16th–17th century printed books containing illustrations of distillation, mining, metallurgy, and alchemical apparatus. We found that the transfer of knowledge from contemporary photographic data to historical etchings proves less effective than anticipated, revealing limitations in current methods like visual feature descriptors, pixel segmentation, representation learning, and object detection with YOLOv8. These findings highlight the stylistic disparities between modern images and early print illustrations, suggesting new research directions for historical image analysis. %This study discusses the utilization of modern computer vision methods, specifically YOLOv8 amongst others, for object detection in historical images extracted from 16th–17th century printed books containing illustrations of distillation, mining, metallurgy, and alchemical objects. %The paper highlights the intricacies of annotating data for supervised learning in this unique context. An extensive analysis reveals significant challenges encountered during the annotation process that may have negatively affected the training process. % %The investigation concludes with an unexpected finding: the knowledge transfer from contemporary photographic data to historical etchings seems to be less effective than anticipated despite the use of current state-of-the-art methods such as YOLOv8, visual feature descriptors, pixel segmentation, representation learning and OWL-ViT. These less good than expected results may have arisen due to the stylistic discrepancies between modern images and early print illustrations. This discovery underscores the limitations of state-of-the-art object detection models when applied to historical image analysis, presenting new avenues for future research in the field. %The paper hence invites discourse on the ‘alchemy of annotation’ in the context of historical imagery, advocating for the development of more sensitive methods capable of ‘distilling knowledge’ from the rich source material of early printed works.