Unable to perform chinese language OCR using ocrmypdf-polyglot

Using the lastest ocrmypdf-polyglot docker image, I ran docker run -v "$(pwd):/home/docker" jbarlow83/ocrmypdf-polyglot -v -l chi_sim textbook.pdf textbook-ocr.pdf according to instructions on a single page A4 simplified chinese PDF with ~1.5 line spacing. The result was mainly solid black square characters (■■■■■).

Is there a reference PDF for which I can test on for simplified chinese (or any asian language)?

[jbarlow83]

存轻乡圣王里人的领导力

v4.2 offers a partial solution if you specify the tesseract renderer and regular PDF output as opposed to PDF/A. This bypasses two of the bugs described above.

ocrmypdf -f -l chi_sim --pdf-renderer tesseract --output-type pdf   input.pdf output.pdf

You might need to rescan that image if the OCR was not what you hoped for. There is some warping in the text baseline. (I can’t help with OCR quality, that’s an issue for tesseract-ocr).

[jbarlow83]

I tried manually retriggering the Docker build. (It’s usually automatic.) Hopefully that will do it.

On Thu, 4 Aug 2016 at 23:56 wallclock [e-mail] wrote:

Great! I will be happy to try out the new release. Unfortunately, the latest polyglot docker image released 2 days ago seems to be version 4.1.3.

— You are receiving this because you commented.

Reply to this email directly, view it on GitHub https://github.com/jbarlow83/OCRmyPDF/issues/81#issuecomment-237771077, or mute the thread https://github.com/notifications/unsubscribe-auth/ABvcM3Ci1WoV6DSYuR2cIlBB0gB6grz-ks5qct6IgaJpZM4JVDyv .