Film and photography, art, and creative communication frequently include visual motifs. That is, identifiable compositions and patterns that image-makers use to express things visually. The goal of this project is to automatically learn and identify those visual motifs, enhancing the capacity of image analysis with the complex comparative models of iconography.

What visual motifs offer, compared with existing technologies using computer vision strategies, is a more nuanced and refined interpretation of images, based not only on standard recognition of geometrical or semantic data but on the meaningful aesthetic and ideological choices of previous creators through art and media history. Because instead of basing the technology on the imitation of previous artistic movements or authors (like AI based approaches that are trained to create an image “in the style” of a painter), visual motifs serve to contrast and relate multiple image types and authors, comparatively.

Therefore, our first goal is to automate the recognition of actions and patterns occurring in a particular image, video, or film. So that technology not only identifies which visual motif is being used but also traces a sample of previous canonic examples (from key paintings, films and photographs) that become available for users to create new images in an informed way.

In a first stage (6 months) a research assistante hired by the program will adapt and work on the datset extension that will be used to train machine learning algorithms and computer vision technologies.