|
|
|
|
#!/bin/bash
|
|
|
|
|
set -e
|
|
|
|
|
|
|
|
|
|
self=`realpath $0`
|
|
|
|
|
self_dir=`dirname "$self"`
|
|
|
|
|
|
|
|
|
|
DATA_SUBDIR=data
|
|
|
|
|
get_from_annex() {
|
|
|
|
|
annex_get 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200/*.ckpt*'
|
|
|
|
|
annex_get 'tesseract-models/GT4HistOCR/*.traineddata'
|
|
|
|
|
annex_get 'textline_detection/*.h5'
|
|
|
|
|
annex_get 'mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz'
|
|
|
|
|
}
|
|
|
|
|
get_from_web() {
|
|
|
|
|
download_to 'https://qurator-data.de/calamari-models/GT4HistOCR/model.tar.xz' 'calamari-models/GT4HistOCR/2019-07-22T15_49+0200'
|
|
|
|
|
download_to 'https://qurator-data.de/tesseract-models/GT4HistOCR/models.tar' 'tesseract-models/GT4HistOCR'
|
|
|
|
|
download_to 'https://qurator-data.de/sbb_textline_detector/models.tar.gz' 'textline_detection'
|
|
|
|
|
download_to --no-unpack 'https://qurator-data.de/mirror/github.com/tesseract-ocr/tessdata_best/archive/4.0.0-repacked.tar.gz' 'mirror/github.com/tesseract-ocr/tessdata_best/archive/'
|
|
|
|
|
}
|
|
|
|
|
. $self_dir/qurator_data_lib.sh
|
|
|
|
|
handle_data
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
docker build --cache-from my_ocrd_workflow -t my_ocrd_workflow .
|
