In collaboration with the DHLAB of EPFL (Lausanne), the Time Machine Unit research unit at EPFL, the Master's degree program in Digital Humanities (École des chartes–PSL), and the AHLoMA research team (EHESS/CRH), we aim to launch a series of experiments in computer vision with this project. The goal is to provide the community with an iconographic search engine for the collections available on digital libraries, to promote the discovery of images and contribute to the methodological renewal of iconographic studies. How can we identify illuminations of interest in a digital library containing more than 100,000 manuscripts? This preliminary documentary work, necessary before analysis, is a barrier to image studies that we aim to overcome. We are initiating a cycle of experiments with the ambition to design a processing chain to (1) detect illuminations available in digital libraries, (2) automatically index them, and (3) generate interpretative hypotheses.
