Automatic typography analysis on figurative content

The Museum of Gestaltung, in Zurich, has one of the biggest poster collections in the World. It has established a design research partnership with the EPFL+ECAL lab about visitors’ engagement with this heritage. The collaboration will investigate how artificial intelligence can open new forms of representation, association and perception.

We hypothesize that showing the construction structure of the posters will strengthen visitor engagement with the collection. To understand poster creation some key elements can be analysed such as the grid, the lines of force and the typography.

In the last years many algorithms have been developed in order to automatically recognize the characters present in a text. However, little is known about automatic typographic analysis. It consists in recognizing the font style (e.g. serif, …), the variation (e.g. bold, italic, …), the decoration (e.g. Capitals. underline, …), the height of the letters and the font name (e.g. Helvetica).

The proposed project aims at taking stock on the topic of automatic typographic analysis and at initiating the development of a framework to automatically extract typographic information. The project will start by finding and testing existing algorithms and evaluate their performances then initiating the technological transfer to the physical installation which will reveal the poster collection.  


Mathieu Salzmann, Nicolas Henchoz (co-supervisor)

extract typography information from poster