Train a Supercomputer to Analyze Oil Paintings

Ambitious AI project looks for volunteers.

The Saint George on a Bike project has announced a new crowdsourcing effort to assist with their effort to use supercomputers for describing and classifying paintings from the 12th through 18th centuries. The project builds on previous research aimed at training computers to provide text descriptions of images. According to the announcement, “no Artificial Intelligence system has been built and trained to help in the description of cultural heritage images while factoring in the time-period and scene composition rules for sacred iconography...”

Millions of old paintings exist in the museums of the world. The idea is that, if a computer can write thorough and complete descriptions of each painting, integrating visual data with background information on history and artistic practices, researchers will see connections and understand composition choices in ways that are not possible now.

The crowdsourcing project is looking for volunteers to write captions for paintings, which will then be used to train the AI systems. See the press announcement and the Saint George on a Bike website.