ULR - Université de La Rochelle : EA2118 (Technoforum - 23 avenue Albert Einstein - BP 33060 - 17031 La Rochelle - France - France)
Abstract : Speech text in comic books is written in a particular manner by the scriptwriter which raises unusual challenges for text recognition. We first detail these challenges and present different approaches to solve them. We compare the performances of pre-trained OCR and segmentation-free approach for speech text of comic books written in Latin script. We demonstrate that few good quality pre-trained OCR output samples, associated with other unlabeled data with the same writing style, can feed a segmentation-free OCR and improve text recognition. Thanks to the help of the lexi-cality measure that automatically accept or reject the pre-trained OCR output as pseudo ground truth for a subsequent segmentation-free OCR training and recognition.
https://hal.archives-ouvertes.fr/hal-01719619
Contributor : Christophe Rigaud
<>
Submitted on : Friday, March 2, 2018 - 9:52:58 AM
Last modification on : Saturday, March 3, 2018 - 1:01:08 AM
Document(s) archivé(s) le : Thursday, May 31, 2018 - 1:00:59 PM
Christophe Rigaud, Jean-Christophe Burie, Jean-Marc Ogier. Segmentation-free speech text recognition for comic books. 2nd International Workshop on coMics Analysis, Processing, and Understanding (MANPU), Nov 2017, Kyoto, Japan. IEEE, 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), 2017. 〈10.1109/ICDAR.2017.288〉. 〈hal-01719619〉
