Open stweil opened 3 years ago
Running make all PIP_OPTIONS="--pre --extra-index-url=https://digi.bib.uni-mannheim.de/pypi" -k
now creates at least these processors:
venv/bin/ocrd venv/bin/ocrd-import venv/bin/ocrd-tesserocr-binarize venv/bin/ocrd-tesserocr-segment-line
venv/bin/ocrd-dinglehopper venv/bin/ocrd-make venv/bin/ocrd-tesserocr-crop venv/bin/ocrd-tesserocr-segment-region
venv/bin/ocrd-dummy venv/bin/ocrd-olahd-client venv/bin/ocrd-tesserocr-deskew venv/bin/ocrd-tesserocr-segment-table
venv/bin/ocrd-export-larex venv/bin/ocrd-olena-binarize venv/bin/ocrd-tesserocr-fontshape venv/bin/ocrd-tesserocr-segment-word
venv/bin/ocrd-fileformat-transform venv/bin/ocrd-pagetopdf venv/bin/ocrd-tesserocr-recognize
venv/bin/ocrd-im6convert venv/bin/ocrd-repair-inconsistencies venv/bin/ocrd-tesserocr-segment
Currently OCR-D cannot be built natively on MacOS for M1 processors.
This issue tries to track the known problems and solutions.
Working:
make all
builds tesseract.Smaller hacks required:
make all
).Failing:
numpy
fails to build:clang: error: the clang compiler does not support 'faltivec', please use -maltivec and include altivec.h explicitly
.opencv-python-headless
fails to build.numpy
.tensorflow
nortensorflow-gpu
are available in any version.