search

OCR-D / ocrd_tesserocr

Run tesseract with the tesserocr bindings with @OCR-D's interfaces
MIT License
38 stars 11 forks source link

Incorrect model path with current master #202

Closed mikegerber closed 4 months ago

mikegerber commented 6 months ago

With current master (possible the same as the current release), and an install like this:

make install-tesseract
make deps
make install

I get this as the model path (aka "module dir" or "module resource location"):

❯ ocrd-tesserocr-recognize -D
./

I've asked @bertsky in the OCR-D/Lobby, and tried PR #200 -> That gives the correct path:

❯ ocrd-tesserocr-recognize -D
/home/b-mg106/.pyenv/versions/3.11.7/envs/ocrd_tesserocr/share/tessdata/
mikegerber commented 6 months ago

Possibly useful for debugging:

python -c "import tesserocr; print(tesserocr.get_languages()[0])" gives the same paths. In an earlier test -- I think -- at least tesseract --list-langs gave the correct languages for the broken installation.

mikegerber commented 6 months ago

dpkg -l | grep tesser is empty.

bertsky commented 4 months ago

dpkg -l | grep tesser is empty.

That's because we always build our own Tesseract from src now, to be on the safe side.

Since #200 already fixed this, and has been merged, can we close @mikegerber ?

mikegerber commented 4 months ago

dpkg -l | grep tesser is empty.

That's because we always build our own Tesseract from src now, to be on the safe side.

I commented that to document that the system does not have any other Tesseract besides the one installed using make install-tesseract.

Since #200 already fixed this, and has been merged, can we close @mikegerber ?

I think so!