SCRIBO aims at developing algorithms and free software to extract knowledge from multimedia content (text or image) and also for the semi-automatic annotation and collaborative digital documents. SCRIBO has a budget of 4.3 Meuros and brings together several partners: " AFP, " CEA LIST, " INRIA, " EPITA/LRDE, " Mandriva, " Nuxeo, " Proxem, " Tagmatica and " XWiki.
SCRIBO is part of the latest Olena release and can be installed with the whole platform. See download page.
However, the very latest version (development version) is available through our Git repository. Beware that stability and compilation is not guaranteed!
- 1/ Get Olena sources from our Git repository : See details here
- 2/ Checkout one of the following branches:
- stable/scribo (recommended)
- unstable/scribo (sometimes broken!)
- git checkout -b stable/scribo origin/stable/scribo
- git checkout -b unstable/scribo origin/unstable/scribo
- 3/ Browse through sources in scribo directory or build the provided tools :
- mkdir _build
- cd _build
In order to get access to all functionalities, your system should fulfill the following requirements :
- Tesseract 2.x or 3.x
- Qt 4.3 (minimum)
- Fop 0.95 (minimum)
- Online document binarization (Based on Sauvola's algorithm)
- Online text detection in natural pictures
- Online page segmentation and reconstruction
On the Web
ICDAR 2013 - Competition on Music Scores
The contest (see there) is about removing staff from music score images.
Access the dedicated page that presents our solution.
ICDAR 2011 - Historical Document Layout Analysis Contest
The general objective of this contest is analyze the overall document layout analysis on historical documents. Paragraphs, images, drop characters and separators had to be retrieved and well classified.
Jump to the official 2011 Contest page.
The whole source code of the submitted program is available on our Git repository.
ICDAR 2009 - Handwriting Segmentation Contest
The general objective of this contest is to use well established evaluation practices and procedures in order to record recent advances in off-line handwriting segmentation.
Jump to the official 2009 Contest page.
Access the dedicated module page.