2011 | OriginalPaper | Buchkapitel
A Network-Centric Epidemic Approach for Automated Image Label Annotation
verfasst von : Mehdy Bohlool, Ronaldo Menezes, Eraldo Ribeiro
Erschienen in: Complex Networks
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Automatically organizing and searching images by their content in large image datasets are major goals of Web-based search engines. Currently, these goals are accomplished by associating metadata information to each image in the database. In this paper, we investigate the use of network sciences to implement a metadata-propagation mechanism for images. We begin by creating a network of images connected by a part-based appearance-similarity measure, and propose an epidemiology-inspired model for metadata propagation. Our experiments show that organizing images as a network helps us label a large number images in the dataset in an economical way, i.e., with few manual metadata annotations.