Univesité Rennes 2
DATE 10-02-2023 DURÉE 00:31:20 GENRE Conférence PUBLIC Tous publics DISCIPLINE Architecture et art du paysage, Arts visuels et plastiques, Histoire de l'art, Cinéma, Danse, Musique, Théâtre, Informatique appliquée Producteur Université Rennes 2

Résumé

AI Toolkits for the Social Sciences and Humanities: A Closer Look at ModOAP, BaOIA, EyCon, and PictorIA

Julien Schuh, Université Paris Nanterre (France)

The exploration of vast digital archives, now increasingly accessible due to the digitization policies of numerous heritage institutions, poses a unique challenge. This communication offers an insight into a series of projects deploying AI toolkits to transform the management, analysis, and valorization of cultural heritage. Bridging the gap between historians, linguists, engineers, and conservators, these initiatives offer methods for complex image segmentation, pattern recognition, and automated text analysis. These projects utilize accessible data via APIs from various institutions for model training and fine-tuning, focusing on creating specialized corpora and refining AI methodologies. The collective outcomes of these projects have led to the inception of a national consortium project focused on exploring extensive image corpora within heritage collections and research archives.

Julien Schuh (Senior Lecturer in French Literature and Digital Humanities, Université Paris Nanterre, Deputy Director of the Maison des Sciences de l’Homme Mondes); main coordinator of the pictorIA consortium, which aims to promote interdisciplinary research on automatic pattern recognition in the social sciences.