English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

In this thesis, we study multiple tasks related to document layout analysissuch as the detection of text lines, the splitting into acts or the detectionof the writing support. Thus, we propose two deep neural models following twodifferent approaches. We aim at proposing a model for object detection thatconsiders the difficulties associated with document processing, including thelimited amount of training data available.  In this respect, we propose a pixel-level detection model and a secondobject-level detection model. We first propose a detection model with fewparameters, fast in prediction, and which can obtain accurate prediction masksfrom a reduced number of training data. We implemented a strategy of collectionand uniformization of many datasets, which are used to train a single linedetection model that demonstrates high generalization capabilities toout-of-sample documents.  We also propose a Transformer-based detection model. The design of such amodel required redefining the task of object detection in document images andto study different approaches. Following this study, we propose an objectdetection strategy consisting in sequentially predicting the coordinates of theobjects enclosing rectangles through a pixel classification. This strategyallows obtaining a fast model with only few parameters.  Finally, in an industrial setting, new non-annotated data are oftenavailable. Thus, in the case of a model adaptation to this new data, it isexpected to provide the system as few new annotated samples as possible. Theselection of relevant samples for manual annotation is therefore crucial toenable successful adaptation. For this purpose, we propose confidenceestimators from different approaches for object detection. We show that theseestimators greatly reduce the amount of annotated data while optimizing theperformances.

D\'etection d'Objets dans les documents num\'eris\'es par r\'eseaux de neurones profonds