Detect Text (OCR) (Image Processing)


This operator detects text on images.


This operator uses the tesseract library to detect text in images. It returns the result as a table.

Tesseract provides a large range of language specific models, which you can choose with the languages or all_languages parameter. The chosen model is downloaded and then used to detect text on the image which is given on the img port.

The downloaded model is cached on your computer within the .RapidMiner folder. If the model was already downloaded in a prior run of the operator the cached version is used.

You can define the granularity of the detection using the split_into setting. it allows the user to specify if the results are delivered on block, paragraph, line, sentence or individual character level.

Tesseract allows the user to specify more than 300 parameters. You can specify them using the 'additional parameters' setting list.

On Mac OS and Linux users need to install the tesseract library manually for example using brew or apt-get. The library is already bundled for Windows Users



      Selects the language to detect. This is only a short list of common languages. If set to true the user can select more than just the short list of languages Selects a language from the full list of supported languages. Since this list is quite long this parameter is only available if "show all languages" is set to true. time out for the download of the model. tesseract provides speed and performance optimized models. If set to true the speed optimized are used, if set to false the performance optimized are used. Defines the granuality of text detections. Can be one of Blocks, Paragraphs, Lines, Words or Characters. Allows to set additional tesseract parameters like blacklisting certain characters. Possible parameters are for example available at:

      Tutorial Processes

      OCR on Jule Verne Cover