Reflections on connoisseurship and computer vision
This special issue shows the diversity of approaches to connoisseurship throughout history and demonstrates how this similarity, but also the significant differences between human and machine approaches can be understood as productive interventions in the discourse around connoiseurship.
Abstract
This special issue shows the diversity of approaches to connoisseurship throughout history. One recent area of research where questions of connoisseurship have become particularly relevant is digital art history, specifically where it intersects with computer vision and machine learning. Here, connoisseurship is not re-invented but modelled in a way that seems to stay close to the human connoisseur: as learning from examples. While clearly there is no guarantee that a computer will develop strategies of attribution akin to those of the human connoisseur, both tasks and methods seem to stay essentially the same if connoisseurship is operationalized as machine learning. On the following pages we will demonstrate how this similarity, but also the significant differences between human and machine approaches can be understood as productive interventions in the discourse around connoisseurship. Central to this investigation is the question: How do we teach connoisseurship to a new kind of observer — the computer — and what challenges result from this process?