Italic info in hocr output

[niksedk]

I cannot find any italic info In Tesseract 4.00.00alpha hocr output.

Tesseract 3.x included this info via the em tag.

It would be very helpful if this could be added again in some way.

[stweil]

That's a missing feature of the new LSTM engine: it does not support attributes like bold, italic and more.

Tesseract 4 still supports the old OCR engine as long as you use traineddata files which include the necessary information. The files from https://github.com/tesseract-ocr/tessdata will work for you.

[niksedk]

@stweil: thx for the info :)
Is it likely this feature will be added to the new LSTM engine at some point?

[stweil]

Who knows? I don't – maybe @theraysmith has plans to enhance the LSTM engine in that direction.

[amitdo]

#1074 (comment)