Workshop description: Computer vision has made rapid progress in recent years: computers can now reliably match the same image; find differences in similar images; and classify content within multiple images. Recently (and controversially) computers have also begun to be routinely used to identify people and places. This hands-on session will demonstrate how to install and use state-of-the-art open source software from the University of Oxford’s Visual Geometry Group. Attendees will leave the session with knowledge of how to match, differentiate, classify and annotate content within many kinds of images.
Requirements: a relatively recent laptop (not a tablet or phone), ideally one capable of running Docker (please install in advance). Online demos will be provided for those unable to install software. Demonstration data will be provided, but participants are encouraged to bring along (1) pairs of images exhibiting small differences, such as variant printings, restruck coins or altered photographs and (2) collections of images containing identical features, such as multiple images of buildings, duplicate photographs or other printed images.
Bio: Giles Bergel is a digital humanist and book historian based jointly in the Department of Engineering Sciences at the University of Oxford and the Department of Information Studies at University College London. He has been working on computer vision within the humanities since the creation of the Bodleian Library’s Broadside Ballads Online resource in 2012, which implements a computer vision tool that is among several that he now manages and supports. HIs other interests include bibliography, typography and book design; the histories of copyright and media; text encoding; and digital library architecture.