Difference between revisions of "Upgrade Tesseract OCR"

From LogicalDOC Community Wiki
Jump to navigationJump to search
(See also)
Line 42: Line 42:
  
 
=== See also ===
 
=== See also ===
 +
[https://tesseract-ocr.github.io/tessdoc/ Tesseract documentation]
 
[https://qiita.com/aki_abekawa/items/418e069038fbdb77c59e Re-learn Japanese with Tesseract 4.1 using LSTM]<br>
 
[https://qiita.com/aki_abekawa/items/418e069038fbdb77c59e Re-learn Japanese with Tesseract 4.1 using LSTM]<br>

Revision as of 09:28, 27 April 2020

Starting from LogicalDOC 8.3.4 tests were carried out on a new version of the integrated OCR Tesseract.

More precisely, tests were conducted on version 4.1 of Tesseract.

This new version is much more precise in text recognition and also faster, so with a very simple action you will get 2 important benefits: a faster OCR recognition that uses less system resources and above all better quality in character recognition.

Note: the version of Tesseract 4.1 that we propose to install is perfectly compatible with LogicalDOC starting from LD 6.8.4

Starting from LogicalDOC 8.4.1 this is the version that is distributed by default, so if you have installed your system in version 8.4.1 or 8.4.2 you don't need to upgrade

Windows systems

The change is very simple, let's talk about a simple replacement

  1. Rename the tesseract folder present in our installation to tesseractOLD
    eg: C:\LogicalDOC\tesseract will become C:\LogicalDOC\tesseractOLD
  2. download the file to the following address https://service.logicaldoc.com/tesseract/tesseract41-win.zip
  3. extract the contents of the archive into the folder where you installed LD eg: C:\LogicalDOC
  4. This package already includes optimized dictionaries for English, French, German, Italian and Spanish languages.
    If you need other languages you can download them from the following address https://tesseract-ocr.github.io/tessdoc/Data-Files#updated-data-files-for-version-400-september-15-2017
    download the files and put them in the \tessdata folder

Linux systems

  1. remove tesseract if previously installed
  2. check the availability of tesseract 4.1 and install it
    Note: in the latest versions of Ubuntu this is already available also for many versions of Linux Debian and CentOS it is possible to use packages already available.
    For more information: https://github.com/tesseract-ocr/tesseract/wiki
  3. check the configuration of Tesseract OCR in LogicalDOC by verifying that it points to the path of tesseract command
  4. don't forget to install the language dictionary packages. These may already be available as installable packages
    eg: the following will install the japanese dictionary on Debian Buster
    apt-get install tesseract-ocr-jpn (installs japanese dictionary)
    

    or perhaps you will have to download the language files in the /tessdata folder

    Get dictionaries packages from https://tesseract-ocr.github.io/tessdoc/Data-Files#special-data-files


See also

Tesseract documentation Re-learn Japanese with Tesseract 4.1 using LSTM