Creating ALTO [enhancement]

[FoxKyong]

Are there any plans for creating ALTO support in tesseract? I was thinking about programming a module for it. I was searching some conversion tool but found nothing working. I need it to be working in linux terminal, I found just conversion tool from hOCR to ALTO but the output is wrong. It would be also better if tesseract would generate ALTO.

[stweil]

Did you try https://github.com/UB-Mannheim/ocr-fileformat?

[FoxKyong]

Yes I have and the generated ALTO isn't valid. It can't be imported to the software I use and also the validator ocr-validate says it's not valid.

[stweil]

Could you please create an issue for that project then and add more details about the problems which you encountered?

[jbreiden]

A contribution of direct ALTO support would be welcome. Recommend implementing in a separate file api/altorenderer.cpp rather than adding to api/baseapi.cpp.

[stweil]

There is now initial support for ALTO output in latest Git master. I keep this issue open nevertheless until issue altoxml/schema#54 was solved and more testing with Tesseract + ALTO was done.

[Shreeshrii]

@stweil ResultIteratorTest shows that tesseract can identify superscripts, subscripts, small caps and drop caps.

Alto schema seems to support these - see
https://github.com/kermitt2/pdfalto/blob/master/schema/alto.xsd#L762

Does tesseract ALTO output include such Font styles?

[amitdo]

This feature request was implemented a long time ago.