Difference between revisions of "LRDE Document Binarization Dataset (LRDE DBD)"

From TC11
Jump to: navigation, search
(Created page with "Datasets -> Datasets List -> Current Page {| style="width: 100%" |- | align="right" | {| |- | '''Created: '''2010-08-03 |- | {{Last updated}} |} |} =Contact Author=…")
 
Line 7: Line 7:
 
{|  
 
{|  
 
|-
 
|-
| '''Created: '''2010-08-03
+
| '''Created: '''2013-05-30
 
|-
 
|-
 
| {{Last updated}}
 
| {{Last updated}}
Line 34: Line 34:
 
* 125 "scanned documents" based on the "clean documents". They have been printed, scanned and registered to match the "clean documents".
 
* 125 "scanned documents" based on the "clean documents". They have been printed, scanned and registered to match the "clean documents".
  
=Metadata=
 
Text Lines Localization Information has been made available by applying text line localization algorithms. The size category of the text depends on the x-height and is considered with the following rule: 0 < small <= 30 < medium <= 55 < large < +inf
 
  
* 123 large text lines localization (clean)
+
=Ground Truth Data=
* 320 medium text lines localization (clean).
+
* [[Ground Truth for LRDE DBD text line localization]]
* 9551 small text lines localization (clean).
+
* [[Ground Truth for LRDE DBD binarization]]
* 123 large text lines localization (original).
+
* [[Ground Truth for LRDE DBD OCR]]
* 320 medium text lines localization (original).
+
 
* 9551 small text lines localization (original).
+
=Related Tasks=
* 123 large text lines localization (scanned).
+
* [[Document Binarization Evaluation for LRDE DBD]]
* 320 medium text lines localization (scanned).
+
 
* 9551 small text lines localization (scanned).
+
=Software=
 +
* A setup script is provided to download and configure the benchmarking environment. This is the recommanded way to run this benchmark. Note that this script also includes features to update the dataset if a new version is released.
 +
* A Python script is provided to launch the benchmark and compute scores.
 +
* C++ programs (and sources) are provided for performing evaluations and reading ground-truth data.
 +
* 6 binarization algorithms (and their respective C++ sources) are provided and compiled to run this benchmark on their results.
 +
 
 +
Minimum requirements: 5GB of free space, Linux (Ubuntu, Debian, …)
  
 +
Dependencies: Python 2.7, tesseract-ocr, tesseract-ocr-fra, git, libgraphicsmagick++1-dev, graphicsmagick-imagemagick-compat, graphicsmagick-libmagick-dev-compat, build-essential. libtool. automake, autoconf. g++-4.6, libqt4-dev (installed automatically with the setup script on Ubuntu and Debian).
  
 +
=References=
 +
* G. Lazzara, T. Géraud. Efficient Multiscale Sauvola's Binarization. In International Journal of Document Analysis and Recognition 2013 [[http://www.lrde.epita.fr/cgi-bin/twiki/view/Publications/201302-IJDAR]]
  
=Ground Truth Data=
+
=Submitted Files=
The following ground truth data is available: Binarization and OCR ground-truths for the LRDE DBD
+
==Version 1.0==
Image groundtruths have been produced using a semi-automatic process: a global thresholding followed by some manual adjustments.
+
 
 +
----
 +
This page is editable only by [[IAPR-TC11:Reading_Systems#TC11_Officers|TC11 Officers ]].

Revision as of 17:11, 30 May 2013

Datasets -> Datasets List -> Current Page

Created: 2013-05-30
Last updated: 2013-005-30

Contact Author

Thierry Géraud – thierry.geraud@lrde.epita.fr
EPITA Research and Development Laboratory (LRDE)
14-16 rue Voltaire  F-94276 Le Kremlin-Bicetre  France

Current Version

1.0

Keywords

Document binarization, Magazine, Scanned

Description

This dataset is composed of documents images extracted from the same French magazine : Le Nouvel Observateur, issue 2402, November 18th-24th, 2010.

The provided dataset is composed of 375 Full-Document Images (A4 format, 300-dpi resolution)

  • 125 numerical "original documents" extracted from a PDF, with full OCR groundtruth.
  • 125 numerical "clean documents" created from the "original documents" where images have been removed.
  • 125 "scanned documents" based on the "clean documents". They have been printed, scanned and registered to match the "clean documents".


Ground Truth Data

Related Tasks

Software

  • A setup script is provided to download and configure the benchmarking environment. This is the recommanded way to run this benchmark. Note that this script also includes features to update the dataset if a new version is released.
  • A Python script is provided to launch the benchmark and compute scores.
  • C++ programs (and sources) are provided for performing evaluations and reading ground-truth data.
  • 6 binarization algorithms (and their respective C++ sources) are provided and compiled to run this benchmark on their results.

Minimum requirements: 5GB of free space, Linux (Ubuntu, Debian, …)

Dependencies: Python 2.7, tesseract-ocr, tesseract-ocr-fra, git, libgraphicsmagick++1-dev, graphicsmagick-imagemagick-compat, graphicsmagick-libmagick-dev-compat, build-essential. libtool. automake, autoconf. g++-4.6, libqt4-dev (installed automatically with the setup script on Ubuntu and Debian).

References

  • G. Lazzara, T. Géraud. Efficient Multiscale Sauvola's Binarization. In International Journal of Document Analysis and Recognition 2013 [[1]]

Submitted Files

Version 1.0


This page is editable only by TC11 Officers .