
Open Access
Computer vision, a branch of artificial intelligence aimed at understanding digital images, promises to help backlogged archives make their digitized collections more discoverable. However, resonances between this new technology and the history of photography make its application in photographic archives practically and methodologically fraught, especially with images of people. This article describes a critical, pared-back approach to person detection, a common computer vision task, designed in response to the shared physiognomic underpinnings of computer vision and photography. Reframing person detection around the question of presence and absence, it demonstrates how computer vision can enable archival discovery beyond its historically textual and classificatory limits.