Supporting Cultural Heritage Research in Historic Photography Archives with Machine Learning and Computer Vision

FAIN: HAA-256249-17

Carnegie Mellon University (Pittsburgh, PA 15213-3815)
Golan Levin (Project Director: January 2017 to October 2022)
David Newbury (Co Project Director: July 2017 to October 2022)

The development of a set of prototype image identification tools and techniques to allow enhanced access to large photography archives. The Carnegie Museum of Art’s Charles “Teenie” Harris Archive of African American life in Pittsburgh would serve as the test collection.

We address the challenges faced in the research and annotation of large digital image archives by creating prototype software tools that use machine learning and computer vision. Specifically, we are developing software tools to aid research into the Carnegie Museum of Art’s publicly available Teenie Harris Archive, a major photography collection documenting 20th century African American life in Pittsburgh. Our goal is to create open-source software that uses state-of-the-art techniques to help identify and annotate visually distinctive features across this large (80,000 item) set of digitized photographs, to improve and expedite the Museum's archiving and cataloging process. Through compatibility with International Image Interoperability Framework (IIIF) standards, our project will furthermore provide free tools and reproducible, computer-vision based workflows that other museums, libraries and archives can use to help organize their own digital collections.