tesseract::TessBaseAPI::ProcessPages cannot be stopped on demand

[flaviu22]

Environment

• Tesseract Version: Tesseract 4.1.1
• Platform: Win10 64bit, VS2017, MFC C++ application

Current Behavior:

tesseract::TessBaseAPI::ProcessPages cannot be stopped on demand. I didn't discover any way to stop tesseract::TessBaseAPI::ProcessPages when I want, some timeout could be setup, but this not help

Expected Behavior:

tesseract::TessBaseAPI::ProcessPages should have a solution to be stopped on demand

Suggested Fix:

tesseract::TessBaseAPI::ProcessPages should taking account ETEXT_DESC somehow. More details could be found here: https://stackoverflow.com/questions/72719440/stop-tesseracttessbaseapiprocesspages-on-demand

[amitdo]

The timeout does not work the way you think it does.

The whole OCR process is done in a few phases.

• Binarization
• Image finding
• Horizontal and vertical lines finding
• Layout analysis
• Text recognition

The time is checked before each phase starts, but there is no way to stop Tesseract in the middle of a phase (except by manually killing the whole process).

[flaviu22]

So, in this case there no way to stop tesseract::TessBaseAPI::ProcessPages gracefully when I want, so, you can close the issue. If you have some hint here, please tell me. Thanks.

[amitdo]

This timeout feature can be improved by adding a few sub-phases.

I marked this issue as 'feature request', but it will probably have low priority.

[zdenop]

https://groups.google.com/g/tesseract-ocr/c/qkn_YO4SH-k/m/qIRsKdSJAAAJ Not sure if problem is your implementation or expectation.
If you want to change the current implementation - a patch is welcomed. We have no capacity to support 3rd party projects.

[flaviu22]

Yes, that's will be perfect, I wonder if I can imply myself, to solve this as quickly as possible.