Computer Vision for Digital Humanists

Project lead: Dr Sarah Lang; Sean M. Winslow, PhD

Institution: Centre for Information Modelling, University of Graz

Project duration: 1.10.2022 – 31.08.2023

Computer Vision has become a very relevant skill for many Digital Humanists, but it is hard to distinguish how we can best budget our time and efforts to engage with this new technology. This workshop will provide a conceptual introduction to the processes involved, coupled with hands-on exercises that focus on the ways that we, as humanists, can curate content and manage metadata to make the best use of new tools available.

Keywords: computer vision, winter school, teaching videos

(intermediary) Outcomes

Project Impact
The project has filled a significant gap in the field of Digital Humanities by providing highly sought after resources for those interested in incorporating Computer Vision in their research. The teaching materials, along with the winter school, provide the necessary springboard for scholars and students to gain expertise in Computer Vision and Distant Viewing, a skill that has become increasingly relevant as a paradigm and visual turn in the Digital Humanities in recent years. The project outcomes have the potential to significantly transform how Digital Humanists approach their work, taking visual sources (which are said to have long been neglected in DH in favour of text-based data) into account and enabling them to tackle more complex projects with a better understanding of the technologies involved.

Winter School and Feedback
An essential part of the project was the winter school (call for Participation) held in Graz from February 7th to 10th, 2023. The organizers had the pleasure of welcoming external expert Germaine Götzelmann among the speakers. Other speakers included Anguelos Nicolaou, Nicolas Renet, Niklas Tscherne, Suzana Sagadin, and Sarah Lang from the Centre for Information Modelling. The event proceeded smoothly, and the feedback received from the participants was overwhelmingly positive, underscoring the need for such teaching materials in DH.

Teaching Materials
A central part of the project was the creation of educational self-learning resources on Computer Vision specifically for Digital Humanities. These materials consist of slide decks, Jupyter Notebooks with practical exercises in Python as well as teaching videos. A whole playlist of comprehensive teaching videos (more than 10 hours of video material in total) was created, covering a range of topics from the basics of computer vision and machine learning to training custom deep learning models for one’s own historical data. Additionally, a number of slide decks and Jupyter notebooks were created, filled with hands-on exercises based on the workshops. These materials, in conjunction with the videos, provide a comprehensive self-learning course in Computer Vision for Digital Humanists. They are now available on relevant platforms such as DARIAH Campus (slide decks and Jupyter Notebooks) and Youtube (videos). We have created much more video material than initially anticipated, totaling around 10 hours of video course material. These videos provide a standalone self-learning class, offering valuable skills for computer vision for Digital Humanists.

  • Video playlist for the self-learning class “Computer Vision for Digital Humanists”:
    • Introduction of the Course “Computer Vision for Digital Humanists” [by Sarah Lang]
    • Machine Learning for Computer Vision: Introduction to terms & concepts [CV for DH, by Sarah Lang]
    • History of Computer Vision: A Timeline [Nicolas Renet, Computer Vision for Digital Humanists]
    • Diving into Data [Practical Exercise, by Suzana Sagadin, Computer Vision for Digital Humanities]
    • Hello World! A mini deep learning model [Practical Exercise, Suzana Sagadin, Computer Vision for DH]
    • Taxonomy of Methods: How is your problem called? [by Angelos Nicolaou, Computer Vision for DH]
    • Performance Evaluation: Epistemological challenges in Distant Seeing [Angelos Nicolaou, CV for DH]
    • Exploring the large model [Practical Exercise, Angelos Nicolaou, CV for DH]
    • Data Management with Tropy [Practical Exercise, CV for DH, by Niklas Tscherne]
    • Negotiating Real-World Projects [by Daniel Luger & Angelos Nicolaou (DiDip ERC), CV for DH]
    • State of the Art? Projects in the Digital Humanities [CV for DH, by Sarah Lang & Suzana Sagadin]
    • What is a medieval charter?
    • Takeaways from a computer vision collaboration with Humanists
    • Things to consider in data criticism for computer vision
    • The engineer’s perspective on machine learning problems
    • What does input shape mean in computer vision?
    • Preprocessing image data for PyTorch
    • Labelling data to create ground truth for supervised learning