Hybrid Workshop - Automated Text Recognition of Historical Sources: From convolutional neural networks to vision language models

Friday 12 Jun 2026, 09:00 → 18:00 Europe/Zurich

3rd floor, seminary rooms 2 and 3 (Austrian Academy of Sciences)

Automated or Handwritten Text Recognition has been making continuous progress, with new tools, models, and technologies emerging every year. This rapidly evolving field is becoming a "must-have" for many digital endeavors, particularly in the study of historical sources. There are, however, still multiple challenges, e.g., complicated page layouts, lack of models for some languages and scripts, Latin abbreviations, and inconsistencies in the training data. 

Used images. Top: ÖNB, Cod. 15298 HAN MAG, fol. 3r; SBzB Ms Or fol 248;ÖNB, 1175, 4r; ÖNB, Cod. 387 HAN MAG, fol. 117r; BNB, Pelliot tibétain 999

Bottom: The Walters Art Museum, W.537.2V; ÖNB, Cod. 3891, fol. 1r; CFMM 00567; ÖNB, G 16715 PAP MAG 

 

This hybrid workshop at the Austrian Academy of Sciences will showcase a variety of case studies and projects from researchers based in Vienna, Graz, and Innsbruck, along with a guest presentation by David Smith from Northeastern University.

The event will conclude with a round table discussion on the present and future of Automated Text Recognition, featuring (in alphabetical order):

• Anna Dolganov
• Günter Mühlberger
• Jan Odstrčilík
• David Smith
• Georg Vogeler

We invite you to join us to explore the latest developments in ATR.

The event is organised by:

• Digital Lab, Institute for Medieval Research (IMAFO), Austrian Academy of Sciences
• Austrian Archaeological Institute (OeAI), Austrian Academy of Sciences
• Faculty of Historical and Cultural Studies at the University of Vienna

Supported by:

• Machine Learning Topical Platform (MLA2S), Austrian Academy of Sciences

Organisational team:

• Jan Odstrčilík, Helmut Reimitz (IMAFO)
• Anna Dolganov (ÖAI)
• Thomas Wallnig (University of Vienna)

 

 

Session 1

1 Auditing ATR Datasets: From CER to CERatorsaurus

Speakers: Jan Odstrčilík (IMAFO, ÖAW), Martin Roček (IMAFO, ÖAW; Charles University)

2 Building HTR Models for an Underresourced Language: The Case of Medieval Czech

Speaker: Anna Michalcová (IMAFO, ÖAW and Czech Language Institute CAS)

3 Reading the Stars with AI: Handwritten Text Recognition (HTR) for Astronomical Text

Speaker: Doris Vickers (Universität Wien)

10:30 Coffee break

Session 2

4 Open Source Layout Analysis for Sanskrit and Tibetan Manuscripts: First Experiences With Kraken on MUSICA

Speaker: Patrick McAllister (IKGA, ÖAW)

5 HTR for Modelling Syriac Colophonology: Opportunities and Challenges

Speaker: Ephrem Aboud Ishac (Institute for Medieval Research, Division of Byzantine Research, Austrian Academy of Sciences - Vienna)

6 Building a Digital Corpus of Armenian Manuscript Colophons: Training and Applying HTR Models for Classical Armenian

Speaker: Lewis Read (IMAFO, ÖAW)

12:30 Lunch break (not provided)

Session 3

7 A Taste of Honey’: ATR and Early Medieval Glossed Manuscripts

Speaker: Bernhard Bauer (Universität Graz)

8 It’s Philology All the Way Down: Htr and the Division of Labor in Textual Editing

Speaker: David Smith (Northeastern University)

9 Deciphering Historical Documents With the Help of LLMs: New Perspectives in Greek Papyrology

Speaker: Anna Dolganov (ÖAI)

15:30 Coffee break

10 Round table: The Present and Future of Automated Text Recognition: Research Topic and Research Tool

Speakers: Anna Dolganov (ÖAI), David Smith (Northeastern University), Georg Vogeler (Karl-Franzens-Universität Graz, Institut für Digitale Geisteswissenschaften), Günter Mühlberger (Universität Innsbruck), Jan Odstrčilík (Institut für Mittelalterforschung, ÖAW)