Tesseract OCR x86 issue

[Kenz Kenza]

Hello;
I am using VB.NET and Tesseract OCR (using CUBE_COMBINED) to extract text from a 300ppp jpeg Images. I am testing almost 45 Image. When I use x86 as target It works untill the Image number 7 and stops with an SEHException. It does not complet.. I tried to change the platform to x64 CPU and It worked with no Issue.
The costumer of my project has only x86 platform and the project should work on his computer.. So any solution for that?? I am really suck..

[zdenko podobny]

I am not sure, but maybe there is problem with VS optimization... See 

https://code.google.com/p/tesseract-ocr/issues/detail?id=1496

Zdenko

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To post to this group, send email to tesser...@googlegroups.com.
Visit this group at http://groups.google.com/group/tesseract-ocr.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/0583430b-e76b-49df-b783-3ec01294ca09%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Rich Taylor]

We are building the Tesseract code with Visual Studio and we've found where some (apparently) compiler optimization bugs are affecting us - definitely with VS2013 Community, but also perhaps with VS2008.  Some of your details (x86 but not X64) sound similar - but this is a very "conditional" problem depending on image data, code and sometimes threading.

I solved it for us by adding some #pragma statements to disable optimization for the problematic function.
I reported details in my latest notes for https://code.google.com/p/tesseract-ocr/issues/detail?id=1496

If you're building with VS then my changes are worth a try.  Let us know what you find!

-= Rich Taylor

[Mark Seidner]

Hi,
   I maintain the latest release of Tesseract 3.04 in a commercial product, please see www.topocr.com
We build a 32-bit binary for Windows using VS2005, we had to do a small amount of hacking to get it to compile, but it runs just fine!

[buyi wen]

if you like tesseract ocr, you may like this free online ocr tool using tesseract ocr 3.02