Artistic and Artifical Seeing. Computer Vision and Art History in Methodical and Practical Cooperation

A dialogue about artistic recognition and similarity as well as case studies in a wide range of artistic production should combine the two fields in the context of Computational Humanities

The WIN project is devoted to issues that are shared by computer vision and art history. The common focus is the image understood as an epistemological and practical problem. In particular, the relationship between formal and semantic similarity is of great scientific interest as the semantic gap between visual and thematic similarities is a major challenge for both subjects.

We like to implement a technical and methodological model to develop a scientific image search and automated analysis. The applications and the basic research of the group as well as their critical reflection of theory are used to make digital image databases more image-­‐oriented and therefore more efficient. Computer vision can access directly the image information, make descriptions, and show connections between works of art in a quantity that humans cannot survey.

In return computer vision is using art historical description tools and problem-­‐oriented images of selected corpora to develop an automatic image understanding. The immediate analysis and feedback through art history helps computer vision to enhance its existing mostly formal access and develop a content-­‐based approach. As a result, a deeper understanding of the cultural viewing habits and the ability to identify objects by the formal language of art is expected.

WIN-­Kolleg is funded by the Heidelberg Academy of Sciences and Humanities (HAW).

Link to project website.

Name and contact of project responsible(s):

Prof. B. Ommer (Interdisciplinary Center for Scientific Computing, Heidelberg University)

Additionally involved scientists and partners

Dr. P. Bell (Heidelberg Collaboratory for Image Processing and Heidelberg Academy of Sciences and Humanities)


Masato Takami, Peter Bell, and Björn Ommer,
Offline Learning of Prototypical Negatives for Efficient Online Exemplar SVM,
in: Proceedings of the IEEE Winter Conference on Applications of Computer Vision, IEEE, 2014.

Antonio Monroy, Peter Bell, and Björn Ommer,
Morphological analysis for investigating artistic images,
Image and Vision Computing 32(6):414-­‐423, Elsevier, 2014.