Image recognition in pre-modern, popular print collections (IPOP)

The aim of this project is to use and improve annotation and classification algorithms to perform computer vision/image recognition on images in Dutch penny prints (1700-1900). We make use of software in the Clariah Media Suite and train the available algorithms with a sample of c. 3000 images from the collections of penny prints in Delpher. This cheap print forms an interesting use case for computer vision, because of its specific visual features (8 to 24 crude woodcuts per broadsheet, with short captions), its adaptability, and its longevity. With the computer vision tooling we will be able to observe the practices of re-using, copying or adapting woodcuts in a large corpus of penny prints. On the long run we can apply this tooling for the comparison between penny prints and other genres