I am already using Tesseract via a wrapper. This works fine save for one issue. Despite having various properties for detecting the page/text orientation, none of them work.
If a document is upside down or rotated 90/270 degrees I would like to be able to detect this and correct the orientation. Does IEVision actually detect the orientation correctly? If so, does it have to use the legacy version of Tesseract to do so?
I would try out the demo myself, but it says ImageEn Version 9 is required.
I don't need to upgrade to version 9 of ImageEn, but would consider doing so (along with acquiring IEVision) if you are able to assure me re the abov.
Hi, It has taken me a while to look at the demo (beset by something greatly resembling the virus!). The demo does recognise text angle very efficiently. Looking at the trained data that comes with the demo, I think that the data maybe from the legacy engine and not the new engine (I believe that is the LSTM). I think I read somewhere that the legacy engine is capable of reading the text angle whereas the newer engine is not. However, the newer engine has better and faster OCR capabilities. I would be happy to use the legacy engine for detecting text angle, but would need to use the current engine for OCR. Are you able to advise me on what version(s) of the engine are being used in the dll? Thanks
Search
IEVision OCR
Posted - Apr 15 2020 : 12:16:52
