Hardly to detect MRZ text

[MkMehdi]

Hello;

First thank you so much for this brilliant library 👍

Well my issue is when I want try detect MRZ text gives me a wrong result .. is there any configuration can do for get better result !

Thank you :))

[pchx-zz]

Would you mind clarifying what you mean by MRZ text?

Note that the Text Recognition API only supports languages that use Latin characters. See our site for details:

https://developers.google.com/vision/text-overview

[MkMehdi]

Thank you for you reply 👍
Well to be clear text like that :
I<UTOSTEVENSON<<PETER<<<<<<<<<<<<<<<
D231458907UTO3407127M9507122<<<<<<<2
the alphabets detect fine but the difficult with "<" sometimes change it to "s" but not all characters...
Is there any configuration can do to get the result more clear ?

Regards;

[bdhobare]

Hello.Does this library read MRZ lines? I am also interested in reading MRZ lines

[pchx-zz]

The Text API isn't specifically optimized for reading MRZ text, but is certainly capable of reading any Latin text (though it may be less accurate in those cases).

[parulsngh]

Can anyone suggest a better option to read MRZ, text api is not helping out with MRZ.

[ajinkya933]

Try OCRB.traineddata.
Usage: tesseract input.jpg output.txt -l OCRB

[parulsngh]

Have tried the tess-two, its working fine but some time it suddenly stops reading text from image and again starts reading by its own. Didn't figured out the reason for same.

[ajinkya933]

Tesseract needs proper input image try changing the image input by applying morphological transforms/ text localization algorithms and then run it through tesseract

[ajinkya933]

close #110