Message Boards

WOLFRAM COMMUNITY

10490 Views

4 Replies

5 Total Likes

View groups...

Follow this post

Share this post:

GROUPS:

Data Science Mathematica

TextRecognize options and Tesseract page mode options.

Rodrigo Murta

Rodrigo Murta, Looqbox

Posted 12 years ago

I saw that Mathematica uses Tesseract as OCR function (you can check in /Applications/Mathematica.app/SystemFiles/Converters/Tesseract/tessdata) I recent discovery that Tesseract has some nice options that assumes some text orientations called page mode. You can see these options here in this SE answer. Can I control it with some Mathematica undocumented options? If not, it would be nice to have this connection in future Mathematica version.

POSTED BY: Rodrigo Murta

4 Replies

Sort By:

Ilian Gachevski

Ilian Gachevski, Wolfram Research

Posted 8 years ago

If I run something like the following in version 11.1.0 libs = FileNames[".dylib" \| ".so*", FileNameJoin[{ParentDirectory[DirectoryName[FindFile["TesseractTools`"]]], "LibraryResources", $SystemID}]]; Map[Function[StringCases[ReadString[#], "tesseract " ~~ (DigitCharacter \| ".") ..]], libs] on my Mac and Linux machines, it seems like version 3.04.00 is being used. Not sure about Windows.

If I run something like the following in version 11.1.0

libs = FileNames["*.dylib" | "*.so*", 
 FileNameJoin[{ParentDirectory[DirectoryName[FindFile["TesseractTools`"]]], "LibraryResources", $SystemID}]];
Map[Function[StringCases[ReadString[#], "tesseract " ~~ (DigitCharacter | ".") ..]], libs]

on my Mac and Linux machines, it seems like version 3.04.00 is being used. Not sure about Windows.

POSTED BY: Ilian Gachevski

Alexey Popkov

Posted 8 years ago

Is it possible to find out which version of Tesseract Mathematica uses under the hood?

POSTED BY: Alexey Popkov

Rodrigo Murta

Rodrigo Murta, Looqbox

Posted 12 years ago

Risk accepted. Worked just perfect. tks!

POSTED BY: Rodrigo Murta

Ilian Gachevski

Ilian Gachevski, Wolfram Research

Posted 12 years ago

Undocumented functionality is usually not documented for very good reasons, see all the caveats listed in this SE discussion. That being out of the way, after some spelunking, e.g. inspecting the output of TextRecognize; Information["`TextRecognize*"] it seems an option along the lines of "SegmentationMode" -> 6 might do something (but try it at your own risk).

POSTED BY: Ilian Gachevski

Reply to this discussion

Reply Preview

Attachments

Remove Add a file to this post

Follow this discussion

or Discard

Group Abstract

Feedback