Deep Discoveries: A Towards a National Collection Foundation Project Final Report

Deep Discoveries was a Towards a National Collection Foundation Project exploring the application of computer vision (CV) and explainable artificial intelligence (XAI) methods for enhancing the ability of general audiences and specialist researchers to discover visual collections in new and/or more effective ways. The team developed and user-tested a CV-based search platform that allowed users to visually articulate their search task, understand how the CV algorithm found similarity between their input image and the returned image results, and to carry out a ‘visual dialogue’ with the AI to refine their search further. We aimed to design a platform that would allow users to both narrow-in on specific images during a directed search as well as to broaden-out and discover new collection items during an exploratory search.