There was already a thread about the tesseract version. How can I find out the tesseract-version I’m using?
Concerning the results it seems to be the old version 3.x.
Tesseract really has made a jumpstart with lstm since than.
Currently I’m checking with OCR if I got the correct Window-Pane, which fails because tesseract recognises “3855” from “855” (if I remember correctly these wrong duplicates were typical 3.x-version errors).
So, how to find out the version that is used?
And if it is old, could it be upgraded … the recommendation in the old thread about EasyOCR is also a good one, by the way
OpenRPA uses Emgu.CV ( 4.1.1.3497 right now, latest seems to be using 4.8.0 that uses opencv version 4.8 )
It’s pretty hard finding packages that still support dot net 4.x but a quick google search for alternatives the newest i could find was a package that uses opencv 4.9 … recoding everything for an upgrade from 4.8 to 4.9 seems overkill. ( but if you are willing to fund it, I’m happy to do it )
I’m not sure how much has changed from 4.1 to 4.8 but will have a look the next time I’m working on OpenRPA.
note: I had a quick look and seems EasyOCR is a python package, not a dotnet package. And using paid closed source solution with OpenRPA kind of defies the whole the purpose of being open source. But if you can get your job done, using python, yes, that seems like a good alternative
Thanks for digging into this. I think you’re right. OpenCV 4.x in itself … there wouldn’t be much improvement, so that would be overkill.
I am a newbie to OpenRPA but would like to check for an integration of a python-package.
Can one access the image retrieved by “Get Image”-Component in “Invoke-Code” or do you have to save the image.
By the way, I really appreciate your effort and time, Allan. Chapeau for handling this forum, answering, developing and all this very fast! You rock.
You will need to save an image somewhere.
You COULD see if you can find the image you want in fs.files collection ( go to files in openflow interfase ) then you can download it using get file under openflow activities.
But might be eaiser to just upload the image your self, and then still use get file. ( saving in the db, to avoid loosing it, if you only save it locally and/or need to use on multiple machines )
