Context

The construction of this database has been motivated by the research conducted for the eBDtheque project. This project has been initiated by the Laboratory in Computer Science, Image and Interaction (L3I) of La Rochelle (France) in collaboration with the CIBDI. It started in June 2011 and was initially supported by the European Doctorate founds of the University of La Rochelle, European Regional Development Fund, the region Poitou-Charentes (France), the General Council of Charente Maritime (France), the town of La Rochelle (France).

Presentation

Since the last century, comic books have become an important piece of the cultural legacy for a lot of countries. Western Europe, but also USA and Japan respectively with comics and mangas, have produced a large amount of material which is now massively digitized according to preservation and content enhancement concerns. The aim of this project is to create a comic books information retrieval system which will be based upon:

a multi-criterion segmentation of comic books images
a standardized representation of the extracted complex data

The "Cité Internationale de la Bande Dessinée et de l'Image" (CIBDI) has already digitized an important part of their database. Comic books digitization implies a need to new tools to handle this data. While existing tools usually provide a classical keywords navigation (writer, style, year...), the image analysis approach introduces new information retrieval possibilities such as indexation (retrieve similar images) or content based image retrieval (find a panel, a character, an object in an image).

Moreover, regarding state of the art and feedbacks from companies, there is no existing tool providing a satisfying extraction of panels, balloons, characters, etc. The creation of digital comic books, meant to be read on smartphones and tablets, still has to be done manually which is another motivation for this project.

The main objective of the eBDtheque project is the creation of an information retrieval platform powered by a knowledge base gathering heterogeneous data: keyword taxonomy, extraction of regions of interest (panels, characters, balloons, etc.), screenplay's semantic (view angle, shot, role of characters, etc.) and so on.
The second objective is to set up an interactive reading application so a former printed comic book can be read on a portable device in a convenient way.

Current/past participants

Professors and assistant professors

Jean-Christophe Burie, Jean-Marc Ogier, Karell Bertet, Arnaud Revel, Farid Ammar-Boudjelal, Alain Bouju, Ronan Champagnat, Georges Louis

Post-docs

Christophe Rigaud, Nhu Van Nguyen, Muhammad Muzzamil Luqman, Norbert Tsopze, Clément Guérin, Srikanta Pal

Ph.D. students

Thanh Nam Le, Clément Guérin, Christophe Rigaud

Software engineers

Julien Maître, Antoine Mercier

Fundings

- CPER program 10: Image and Interactivity, sub-action 2: Platform for information retrieval in complex data (started in 2007 but founded from 2011-2013).
- PHC Sakura with Osaka Prefecture University (OPU) (2014-2015).
- PIA iiBD with Actialuna startup, Lip6 lab, Ecole des Mines de Paris and TerraLab company (2015-2017).
- CPER NUMERIC theme 2 : e-création in collaboration with the CIBDI and the CEPE of Angoulême (2015-2020).
- MEDIA-DB with the laboratory CLIMAS (EA 4196) and the laboratory MICA (EA 4426) from the university of Bordeaux Montaigne (2017-2020).