The Challenges of Automatic Art Captioning

A virtual presentation in collaboration with the Hasso Plattner Institute

Recorded Webinar Available

Winslow Homer. Eight Bells (1886). Oil on canvas, 64.1 × 76.5 cm (25.2 in × 30.1 in). Addison Gallery of American Art.

Digital libraries caption their online digital images to improve the accessibility of their resources. Although recent machine learning models do a good job captioning photographs of common objects, they fail to thoroughly describe digitized artworks. Alejandro Sierra Múnera from the Hasso Plattner Institute will delve into the results of an academic project designed to better understand these challenges, and showcase the strengths and weaknesses of a state-of-the-art captioning model. 

Alejandro Sierra Múnera will be joined by Dana Anderson, the Digital Assets Manager at WPI. The conversation will be moderated by Executive Director Elizabeth Gorayeb.

Alejandro Sierra Múnera is a PhD student at the Hasso Plattner Institute in Potsdam, Germany, under the supervision of Prof. Dr. Felix Naumann and Prof. Dr. Ralf Krestel. His research is focused on the analysis of cultural heritage text documents, and their connection to visual information and knowledge graphs.

This event was recorded on March 12, 2024.

Scroll to Top