Computer Vision for History
The VoH Working Group in cooperation with Research IT presents a series of lectures titled "Overcoming Obstacles, Learning from Experiences: A Transdisciplinary Conversation about Computer Vision, 3D Models, Preservation, and Outreach in Digital Humanities projects,” running from May–July 2024. The series features speakers from multiple disciplines in the Humanities – History of Science, History, Art History, and Archaeology – who will focus on methods that can be utilized in the systematic DH-related analysis of objects. Topics covered include databases, their development, preservation, and dissemination, computer vision and its components, such as classification, annotation, and vectorization, as well as 3-D modeling.
For a full description of the series, please click here.
This lecture series is open to the public. We welcome both internal and external guests. To register, please click here and choose which event you would like to attend. You can register for multiple events but must do so separately.
For questions on registration please contact event_dept3@mpiwg-berlin.mpg.de and for further information about the series please contact rbrentjes@mpiwg-berlin.mpg.de
In this presentation, I will discuss different ways Computer Vision can be used to address concrete historical questions, and present different tools built in my group, with a particular focus on methods that require limited manual annotation of the data. In particular, I will discuss projects that (i) vectorize historical astronomical diagrams in different geometric primitives; (ii) extract illustrations from documents and look for similarities between them; (iii) extract prototypical letter shapes from handwritten documents and leverage them to answer paleographical questions; and (iv) analyze printed ornaments.
Mathieu Aubry is a tenured researcher in Computer Vision at École des Ponts Paris Tech in the LIGM lab (UMR8049). He obtained his PhD at ENS in 2015, co-advised by Josef Sivic (INRIA) and Daniel Cremers (TUM). In 2015, he spent a year working as a postdoc with Alexei Efros in UC Berkeley. He has a leading role in the ANR EnHerit, VHS and EIDA projects and the ERC DISCOVER project, which focuses on interpretable visual structure discovery. He works on various applications of computer vision to historical data (see EnHerit, VHS and EIDA projects).