OCR execution with mixed fonts

This is the technical documentation for the Goobi plugin for performing OCR with mixed fonts. Images should be marked with the "OCR Page Selection" plugin.

Introduction
This documentation describes the installation, configuration and use of a plug-in for OCR with mixed fonts. This plugin is only useful in combination with the plugin "OCR Page Selection".
Details
​
Identifier
intranda_step_mixedocr
Source code
​https://github.com/intranda/goobi-plugin-step-mixedocr​
Licence
GPL 2.0 or newer
Compatibility
Goobi workflow 3.0.4 and newer
Documentation date
04.03.2019
Precondition
The precondition for using the plugin is the use of Goobi workflow in version 3.0.4 or higher, the correct installation and configuration of the plugin as well as the correct integration of the plugin into the desired workflow steps. In addition, the plugin for the manual selection of the pages is required .
Installation and Configuration
The following files must be installed to use the plugin:
1
/opt/digiverso/goobi/plugins/step/plugin_intranda_step_mixedocr.jar
2
/opt/digiverso/goobi/config/plugin_intranda_step_mixedocr.xml
Copied!
The first file contains the actual plugin. The second file is the configuration file of the plugin.
The content of the configuration file plugin_intranda_step_mixedocr.xml must be structured as follows:
plugin_intranda_step_mixedocr.xml
1
<config_plugin>
2
<!--
3
        order of configuration is:
4
          1.) project name and step name matches
5
          2.) step name matches and project is *
6
          3.) project name matches and step name is *
7
          4.) project name and step name are *
8
-->
9
    <config>
10
        <!-- which projects to use for (can be more then one, otherwise use *) -->
11
        <project>*</project>
12
        <step>*</step>
13
​
14
        <template>template.xml</template>
15
        <itmUrl>http://localhost:8080/itm/service</itmUrl>
16
        <!--  this must be without a trailing slash -->
17
        <callbackBaseUrl>http://localhost:8080/goobi</callbackBaseUrl>
18
        <useOrigDir>false</useOrigDir>
19
        <serverType>intranda-tesseract</serverType>
20
    </config>
21
​
22
    <config>
23
        <!-- which projects to use for (can be more then one, otherwise use *) -->
24
        <project>My special project</project>
25
        <project>Archive_Project</project>
26
        <step>OCR</step>
27
​
28
        <template>template_pdfonly.xml</template>
29
        <itmUrl>https://itm.mydomain.tld/itm/service</itmUrl>
30
        <callbackBaseUrl>https://goobi.mydomain.tld/goobi</callbackBaseUrl>
31
        <useOrigDir>true</useOrigDir>
32
        <serverType>intranda-tesseract</serverType>
33
    </config>
Copied!
Several configurations for different projects and sections are possible. These are defined by project and step. It is also possible to use the wildcard * for all steps and projects. The actual configuration then takes place within the config elements.
The template is the template that the TaskManager should use, the itmUrl is the URL to the endpoint of the TaskManager that accepts new jobs. The callbackBaseUrl must be a URL that can be reached from the TaskManager and points to the Goobi installation where the plugin is installed. It is needed to close the step after successful OCR. The element useOrigDir determines whether the master images or the derivatives should be used for the OCR. The entry serverType is the value entered in the intranda license server for the server that is to be used for the OCR. This value can be requested from intranda or omitted when using another OCR provider.
In addition to this plugin-specific configuration, the file /opt/digiverso/goobi/config/goobi_rest.xml must be released so that the TaskManager can report the successful processing of the jobs to the plugin:
/opt/digiverso/goobi/config/goobi_rest.xml
1
<endpoint path="/plugins/ocr.*">
2
    <method name="post">
3
        <allow netmask="127.0.0.0/8" token="mytoken"/>
4
    </method>
5
</endpoint>
Copied!
Settings in Goobi
After the plugin has been installed and configured, it must be added to a workflow step in Goobi workflow.
Configuration of the step in Goobi Workflow
The plugin is usually executed automatically, so you should check the box Automatic task. The plugin intranda_step_mixedocr must also be selected under Plugin for step.

Step Plugins - Previous

OCR page selection

Next - Step Plugins

Writing XMP metadata to image files

Last modified 1yr ago

Copy link

Contents

Introduction

Precondition

Installation and Configuration

Settings in Goobi