Explore is an image search tool
created for the Visual Contagions project
at the College of Humanities
of the University of Geneva.
Explore enables the study of image circulation within millions of historical illustrated publications.
The available dataset currently comprises 3,651,742 issues of 80,965 journals that were published between 1797 and 2021 across the world.
These journals are all accessible in open access, in IIIF format, and distributed by international institutions.
The algorithms utilized by Explore operate in several stages.
- They extract the illustrations from each page of the publications;
- They compare the illustrations extracted from different pages to identify similarities between them;
- The images are grouped based on their similarity;
- The illustrations within each group are arranged chronologically;
- With the platform, you can also search in the dataset for the images that are similar to any other image from your computer.
3.6 million images
145834 documents
Corpus contenant des ressources collectées par le projet Visual Contagions. Le corpus est constitué de titres provenant de la Bibliothèque nationale de France, de la Bibliothèque universitaire de Heidelberg, du projet Blue Mountain de Princeton et d'autres. -- Corpus containing resources collected by the Visual Contagions project. The corpus is formed by Titles from the National Library of France, Heidelberg University Library, Princeton Blue Mountain project and others. Le corpus est formé par les images contenues dans 1791 revues dans 304 villes / The corpus is formed by images contained in 1791 journals in 304 cities La liste des revues …
For connected users, additional features are available.
These include:
- access non OA illustrated material;
- compose and organize your own corpus;
- add elements non-native IIIF material;
- extract illustrations from all pages (92% precision 93% recall on CBAD2019);
- select image clusters and organize of these clusters into series;
- comparative spatiotemporal mapping of image dissemination within a given cluster