The Bauhaus Archive collection is made up of over one million items, and only very few of them have ever been shown publicly. Sketches, photographs, sculptures, paintings, everyday objects and internal work images testify to the wide-ranging creativity of this avant-garde school. In the Bauhaus Infinity Archive, you can now access over 15,000 pictures of the collection for the first time thanks to machine learning.
Visitors can explore a three-dimensional archive by interacting with the collection. The items are sorted each time you tap the touchpad. If you draw a line, you will be shown pictures where it appears. If you select a colour, pieces of furniture and graphics featuring it will pop up. The system reacts to each of your actions by searching the archive for matches in real time: the images float through the space following your search motions. They are sorted into new clusters over and over again before your very eyes.
The machine learning model groups the archive images based on their visual similarity. It positions them within the space in relation to one another. This means the search results are always displayed in the context of the entire collection. As soon as you immerse yourself in the deluge of images, you will discover surprising and interesting links between the items.
Archives are important historical resources holding a wealth of information. They are often too large and diverse to be viewed without aids. For this reason, we have chosen machine learning to make the vast array of data kept at the Bauhaus Archive accessible in an interactive system. With this technology, you do not search with keywords or metadata, but instead actively draw geometric shapes or select a specific colour – this route deliberately plays with the clear colour and form vocabulary of Bauhaus.
Findings from our QURATOR research project form the basis for the installation. QURATOR deals with the curation of digital content using AI tools. The Bauhaus Infinity Archive was created in cooperation with the Bauhaus Archive in Berlin. It was developed as part of the “dive in. Programme for Digital Interactions” that was set up by Germany’s Federal Cultural Foundation.
Within this project a paper was published about the algorithm been used for this installation:
Image Garden - Curating Collections and Designing Smart Exhibitions with AI-Based Tools from the journal i-com
Commissioned by Art+Com at QURATOR - AI Researcher for Digital Curation Technologies.
