V. Christlein : Automatic Handwriting Analysis – From Scribe to Script Type Classification
Dans le cadre du programme e.philologie PSL (ENC, EPHE, ENS, EHESS) :
Lundi 27 mars 17-19h, salle Quicherat, ENC, rue Richelieu 65, métro Pyramides.
Vincent Christlein (Université de Erlangen)1
Automatic Handwriting Analysis – From Scribe to Script Type Classification.
Digital libraries, such as Gallica, monasterium.net, or the Vatican’s DigiVatLib, are amassing digitizations of medieval manuscripts. However, those digital copies are often produced without any or only little metadata. This is problematic when trying to actually work scientifically with the data. Enriching the metadata for these collections often requires human experts, making it a time-consuming and expensive task. In this context, methods derived from the field of computer vision are explored to automatically compute missing information such as the script type, the date, or the writer. The talk will start explaining a method for writer recognition based on bag-of-(visual) words. While this method works very good for contemporary data, historical data still poses a challenge, where factors such as decorations or background artifacts might influence a correct assignment. Possibilities to deal with historical data will be shown before proceeding with two other tasks: the classification of medieval Latin script types and the dating of medieval charters.
Pour lire les références relatives à ce séminaire, consultez ephilologie PSL
Le séminaire est ouvert à tous, cependant nous vous demandons de vous inscrire en ligne : https://goo.gl/forms/ycoSTZcuxdrQ1cjY2
- Ce séminaire sera filmé et la communication se fera en anglais [↩]
OpenEdition vous propose de citer ce billet de la manière suivante :
Vanessa Bigot Juloux (24 mars 2017). V. Christlein : Automatic Handwriting Analysis – From Scribe to Script Type Classification. EphéNum. Consulté le 10 septembre 2024 à l’adresse https://doi.org/10.58079/ofzc