ocr program

Tue Dec 12 12:23:39 UTC 2017

Hi Kåre,
yes, I meant that there are no volunteers for a OCR GUI. That's obiously because everyone's time is limited, and volunteers need to choose what to do in their time.

I compiled Kocr with Qt5.6, both for Kubuntu LTS and Windows 7, and it works fine: probably the problem is that on Neon there are Qt5.9 libraries and maybe QPDFWriter works differently (I did not try). I also don't know if there could be any problem with libpoppler-qt5 bindings in Neon (they should be the same as what I'm using, but who knows?).

Basically, what the program does is passing every image to Tesseract, to obtain a single page PDF with searchable text. Then, every PDF produced is read with Poppler and stored into a multipage PDF with QPDFWriter. The problem might be in one of these three steps. Also, there might be a problem with opening and saving the image before passing it to Tesseract: the image is loaded into a QImage and stored as TIFF, so if your system does not have libtiff you might get and invalid image file and Tesseract should produce an invalid PDF.

Luca Tringali

In data martedì 12 dicembre 2017 13:43:26 CET, Kåre Särs <kare.sars at iki.fi> ha scritto:
> Hi Luca,
> 
> Thanks for creating this application and sharing it with us!
> 
> The problem is that there has not been anybody eager enough to start a OCR 
> frontend application so far for Qt5/KF5. As KDE is done on voluntary basis it 
> is hard to "prioritize" anything. Either we get a volunteer or we don't.
> 
> I saw that Kocr is a GUI application that searches for Tesseract and Cuneiform 
> and then runs the selected application on the imported images. I think these 
> ideas can be useful :) 
> 
> Multi-page document scanning to pdf and OCR is quite often requested for 
> Skanlite, but so far nobody has stepped up to do it :(
> 
> Thanks,
>   Kåre
> 
> 
> Ps, I compiled Kocr on KDE Neon and run it. The resulting PDF unfortunately 
> broken. I have not had the time to check why.
> 
> On tisdag 12 december 2017 kl. 11:18:21 EET tringalinvent at libero.it wrote:
> > Hi Albert,
> > I understand, and the fact itself that today KDE does not have an official
> > OCR GUI means that this is not considered a priority. Anyway, if and when
> > someone will be interested in making a GUI like this part of some KDE
> > Graphics program I think my code would be useful. So thanks for your reply,
> > and just let me know if anyone needs help in editing the code of Kocr.
> > 
> > 
> > Luca Tringali
> > 
> > In data lunedì 11 dicembre 2017 23:00:20 CET, Albert Astals Cid 
> <aacid at kde.org> ha scritto:
> > > El dilluns, 11 de desembre de 2017, a les 15:06:05 CET,
> > > 
> > > tringalinvent at libero.it va escriure:
> > > > Hello everybody,
> > > 
> > > Hi!
> > > 
> > > > a few months ago I was looking for a OCR gui that could offer three
> > > > different kinds of output: plain text, formatted HTML, and PDF. Since I
> > > > was
> > > > not able to find one, I wrote a GUI for both Tesseract and Cuneiform
> > > > (which
> > > > is quite better in formatting recognition than Tesseract). The code is
> > > > published on github: https://github.com/zorbaproject/kocr
> > > > 
> > > > It is based on libpoppler and QPDfWriter for reading and writing PDFs.
> > > > It's
> > > > not rich of features, it just has what my girlfriend needed when she
> > > > asked
> > > > me about a GUI for OCR.
> > > > 
> > > > I believe this code could be integrated into Skanlite or Okular, or
> > > > maybe
> > > > become a Kipi plugin for Gwenview. After all, there is no official GUI
> > > > in
> > > > KDE for optical character recognition engines.
> > > 
> > > This sounds interesting but given you're placing it almost as abandonware
> > > i'm not sure we should take it in unless there's someone really really
> > > interested in maintaining it.
> > > 
> > > In my opinion we have far too many "community maintained" software
> > > already.
> > > 
> > > Cheers,
> > > 
> > >   Albert
> > >   
> > > > I've also built packages for Windows and Debian
> > > > (https://github.com/zorbaproject/kocr/releases/tag/v0.3), this could be
> > > > useful to understand how the program is designed to work in particular
> > > > on
> > > > Windows. The README.md file contains a section called "How to use Kocr",
> > > > which should give you an idea of the workflow.
> > > > 
> > > > Just to be clear, I'm not a full time programmer (I'm a journalist), and
> > > > I
> > > > don't have time to keep working on this program. I'm releasing this code
> > > > to
> > > > let other programmers not have to start from scratch while building a
> > > > OCR
> > > > program for KDE. I'll be happy to help making this program part of KDE,
> > > > but
> > > > I don't have a lot of spare time so I don't know how much I'll be able
> > > > to
> > > > help. For sure, I'll try to reply to your questions as soon as I can.
> > > > 
> > > > 
> > > > Luca Tringali
> 
> 
> 