Community

Flexi Layout v9: Polish alphabet and the pre-recognition feature

Hi,

We have the following situation.

1. We have created a new FlexiLayout project.
2. The Windows 10 operating system - German version and the Flexi Layout Studio v9 build No. 9.0.4.2617Part#53815 were used. Project's settings are on the screen001.
3. We have an invoice with polish chars inside,
The document format was PDF, 300dpi, 2480x3508 pixels.
4. The Pre-recognition process has been always giving back results shown on the picture screen002,

Then we tried to catch polish letters separatelly.

1. We have created a new FlexiLayout project.
2. The Windows 10 operating system - German version and the Flexi Layout Studio v9 were used with the same Flexi Layout project's settings.
4. We have a picture with polish chars inside.
Format was PNG, 300dpi, 2480x3508 pixels.
5. Results of pre-recognition action shown on the screen003 and screen004 screenshots.

However, using for instance character string element, the polish chars won't be recognized properly.

At first, seems to be an internal bug in Flexi Layout Studio v9. How to process polish characters properly?

Thanks in advance for some tips.
0

Comments

4 comments

  • Avatar
    Vladislav Suvorov
    Hello,

    Your issue seems to be that FlexiLayout Studio actually uses fast recognition and can simply skip polish diacritics that are far enough from the main symbol or not capture them correctly (please refer to first attachment).
    To see how those characters will actually look on export I recommend you to inflate found character string element's region vertically and import AFL into FlexiCapture itself (and set up your project to recognize polish alphabet) because FlexiCapture uses different algorithms of recognition (please refer to second attachment).

    Please let me know if you have any other questions.

    Best regards,
    Vladislav
    0
  • Avatar
    Lukasz Lechert
    Hello Vladislav,

    Thank you very much for response. The FLS and FC screenshots topic. There isn't info about used version. Which one did you use?
    The FLC and the FC use different algorithms during recognition process. What did you use in the FC and the FLC v9?

    Regards,
    Lukasz
    0
  • Avatar
    Vladislav Suvorov
    Hello Lukasz,

    I used FC11 but the whole OCR engine used in FC9 and FC11 is pretty much the same. FlexiLayout studio was made mainly for regions extraction meaning it might not precisely extract all the data from documents. Since it is specifically designed to only draw regions for FlexiCapture.
    FlexiCapture itself is designed for data extraction therefore it treats regions more carefully and symbols with diacritics should be captured way more reliably.
    Just to make it absolutely clear: I didn't change any settings. FC and FLS have different approach to document recognition themselves.

    Hope that helps,
    Vladislav
    0
  • Avatar
    Lukasz Lechert
    Hello Vladislav,

    Thanks for your tips.

    All the best!
    Lukasz
    0

Please sign in to leave a comment.