Arran Rees at the LFFseminar in Denmark. (Photo Niels Løgager Nielsen)
Arran Rees, Doctoral Researcher, University of Leeds presented a paper titled Lacking in context: cataloguing experiments with AI and photography at the well attended seminar THE FUTURE IS NOW January 30 in Odense, Denmark. Arran has been collaborating with the CoSoPho project, and undertook an experiment as part of it, looking at how AI can be used in helping to catalogue photos. Using a sample of 50 photographs collected through the CoSoPho case studies, Arran explored the use of Image recognition tools Clarifai, Amazon Rekognition and Google Vision, to see the difference in the suggested tags compared with the tags assigned through human-led cataloguing. The paper unpacked the findings, arguing that the tools lack the ability to understand context, but could be used as access-widening tools, producing keywords that can help make collections more accessible to online researchers.
The theme of the seminar in Denmark explored the current uses and experiences of using image recognition and machine learning in archives, libraries and museums. It was organised by LFF (National Society to preserve Photographs and Film) which is a national organisation for archives, museums and libraries with photo and film holdings in Denmark. (fotoogfilm.dk)
Arran Rees’ work will be a chapter in the upcoming anthology: Connect to Collect – Approaches on how to collect Social digital Photography in Museums and Archives. which will be launched at the Connect to Collect conference March 19 -20. 2020 in Stockholm. Arran will also be presenting the results of the research at the conference.