Many, many changes made. Process is extensively commented in align_to_ocr.py. Biggest things to note:
Now works on python 3.5 (the current python version used in the py3-celery docker container in rodan-docker).
Now using Calamari OCR instead of Ocropus
All image processing using scikit-image instead of gamera
OCR models are no longer inserted into the job itself but are packaged directly in the git repo, and the user must select one from a drop-down menu in the rodan job settings
Significant reorganization of functions across the project so that the whole process makes more sense (and all runs through the align_to_ocr.py file).
Many, many changes made. Process is extensively commented in
align_to_ocr.py. Biggest things to note:align_to_ocr.pyfile).