News for hackers. Stuff that matters.

A comment on Creating a Modern OCR Pipeline Using Computer Vision and Deep Learning. Parent comment

Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by tensor on Thursday November 09, 2017 @02:28PM (#15665139)
The unreleased v4 version of tesseract has a new engine based on state of the art deep learning techniques. In some initial poking at it, it looks like it might blow the existing commercial solutions away in terms of quality.
[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by tobltobs on Thursday November 09, 2017 @03:48PM (#15665755)
In mumble experience it is much better then v3 but still not as good as Abby.
[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by tensor on Thursday November 09, 2017 @04:23PM (#15665974)
Can you elaborate on this a bit? It still has some edge case quality issues, but I haven't seen anything where Abby does better so far. That said, the default app is missing a bunch of preprocessing that you have to add (e.g. page deskew, flipping, etc).
[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by ktta on Thursday November 09, 2017 @09:49PM (#15667564)
Do you have an idea of when it is going to be released? Or this is only going to be a Google internal thing?
[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by wwarner on Thursday November 09, 2017 @09:52PM (#15667580)
it's in the master branch, it just hasn't been tagged for release yet.
[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by ktta on Thursday November 09, 2017 @10:49PM (#15667800)
Oh, thanks. I had taken a looked at the number of commits which didn't have many after the last release so I thought it hadn't been merged yet.

https://github.com/tesseract-ocr/tesseract/graphs/commit-act...

[ Parent | on HN ]
Re:Creating a Modern OCR Pipeline Using Computer Vision and ... by ocrcustomserver on Thursday November 09, 2017 @10:53PM (#15667826)
It uses LSTM for the line recognizer.
[ Parent | on HN ]