Photography Unbound

Photography Unbound is a collaborative research project I designed at the Getty Research Institute as part of Phototech, a larger initiative focused on photographic archives. Photography Unbound uses data and computational methods to understand how image collections such as albums, archives, and datasets condition the study of 19th-century photography. Through a critical use of computer vision, the project raises historiographical and methodological questions about the technological conditions of photo-historical research.

The dataset for this project consists of approximately 30,000 pre-1925 photographs. This dataset combines a breadth of photo history from the collection of the Getty Museum with deeper dives into Ottoman-era photography and colonial-era photography from Southeast Asia and the Middle East. These latter materials are drawn from a range of international archives and institutions (including the Getty Research Institute), and while they are digitized, they are not currently easy to find or browse. Through investigations of computer vision and person detection algorithms in particular, the project team is both conducting scholarly research into this material as well as exploring and demonstrating how historical photographs can be made accessible in more inclusive and equitable ways. Our research with computer vision and digital collections also extends to these technologies themselves, drawing on the work of Safiya Noble, Tara McPherson, Michelle Caswell, and others to examine their embeddedness in historical and contemporary ideas about race and nation.

Comparison of results of baseline person detection algorithm (left) and custom figure detection algorithm (right) on Francis Frith, 'Temple of Waddy Sabooa in Nubia,' 1859.

As project lead, I have

designed and articulated project themes and outcomes
compiled a dataset approximately 30,000 pre-1925 photographs and associated metadata from a range of international institutions
normalized and manipulated data from JSON-LD and CSV files using OpenRefine
supervised the technical development of a custom person detection algorithm
created a custom datasette instance using SQLite that serves as a research tool and data repository
organized and lead Zoom-based research meetings
coordinated the team’s efforts into presentations for international conferences and events that make our research legible to curators, archivists, and the general public

In December 2022, the Photography Unbound research team presented our work at the National Gallery of Singapore in conjunction with their exhibition Living Pictures: Photography in Southeast Asia.