You cannot select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
175 lines
4.3 KiB
Bash
175 lines
4.3 KiB
Bash
#!/bin/bash
|
|
|
|
set -e # Abort on error
|
|
|
|
# Configuration
|
|
export LOG_LEVEL=${LOG_LEVEL:-DEBUG} # /etc/ocrd_logging.py uses this to set level for all OCR-D modules
|
|
export TEXTEQUIV_LEVEL=glyph
|
|
|
|
# Command line parameters
|
|
OPTS=`getopt -o I: --long input-file-grp:,skip-validation -- "$@"`
|
|
eval set -- "$OPTS"
|
|
INPUT_FILE_GRP=OCR-D-IMG
|
|
SKIP_VALIDATION=false
|
|
while true; do
|
|
case "$1" in
|
|
-I|--input-file-grp) INPUT_FILE_GRP=$2; shift 2;;
|
|
--skip-validation) SKIP_VALIDATION=true; shift;;
|
|
|
|
--) shift; break;;
|
|
*) break;;
|
|
esac
|
|
done
|
|
|
|
# Set up logging
|
|
if [ "$LOG_LEVEL" = "DEBUG" -o "$LOG_LEVEL" = "TRACE" ]; then
|
|
set -x
|
|
fi
|
|
|
|
|
|
do_validate() {
|
|
# Validate the workspace
|
|
|
|
# Both ocrd_tesserocr + ocrd_calamari produce segment coordinates that are not strictly within their parent's
|
|
# coordinates:
|
|
#
|
|
# INCONSISTENCY in [...] coords [...] not within parent coords
|
|
#
|
|
# → --page-coordinate-consistency off
|
|
#
|
|
# ocrd_tesserocr sometimes produces segment text results that aren't concatenating as expected by the validator:
|
|
#
|
|
# INCONSISTENCY in [...]: text results '[...]' != concatenated '[...]'
|
|
#
|
|
# → --page-strictness lax
|
|
#
|
|
validate_options='
|
|
--skip dimension
|
|
--skip pixel_density
|
|
--page-strictness lax
|
|
--page-coordinate-consistency off'
|
|
if [ "$SKIP_VALIDATION" = false ]; then
|
|
ocrd workspace validate $validate_options
|
|
fi
|
|
}
|
|
|
|
do_binarization() {
|
|
# Binarize the images
|
|
|
|
ocrd_olena_binarize_parameters='{
|
|
"impl": "sauvola-ms-split"
|
|
}'
|
|
ocrd workspace remove-group -rf OCR-D-IMG-BINPAGE
|
|
ocrd workspace remove-group -rf OCR-D-IMG-BIN
|
|
ocrd-olena-binarize -I $INPUT_FILE_GRP -O OCR-D-IMG-BINPAGE,OCR-D-IMG-BIN -p "$ocrd_olena_binarize_parameters"
|
|
}
|
|
|
|
do_linesegmentation_tesserocr() {
|
|
# Segment the lines in the binarized images
|
|
|
|
ocrd workspace remove-group -rf OCR-D-SEG-REGION
|
|
ocrd workspace remove-group -rf OCR-D-SEG-LINE
|
|
ocrd-tesserocr-segment-region -I OCR-D-IMG-BINPAGE -O OCR-D-SEG-REGION
|
|
ocrd-tesserocr-segment-line -I OCR-D-SEG-REGION -O OCR-D-SEG-LINE
|
|
}
|
|
|
|
do_linesegmentation_sbb() {
|
|
# Segment the lines in the images
|
|
|
|
ocrd_sbb_textline_detector_parameters='{
|
|
"model": "/var/lib/textline_detection"
|
|
}'
|
|
ocrd workspace remove-group -rf OCR-D-SEG-REGION
|
|
ocrd workspace remove-group -rf OCR-D-SEG-LINE
|
|
ocrd-sbb-textline-detector -I OCR-D-IMG-BINPAGE -O OCR-D-SEG-LINE -p "$ocrd_sbb_textline_detector_parameters"
|
|
}
|
|
|
|
do_ocr() {
|
|
# Perform OCR on the segmented lines
|
|
|
|
ocrd_tesserocr_recognize_parameters='{
|
|
"model": "GT4HistOCR_2000000",
|
|
"textequiv_level": "'$TEXTEQUIV_LEVEL'"
|
|
}'
|
|
ocrd workspace remove-group -rf OCR-D-OCR-TESS
|
|
ocrd-tesserocr-recognize -I OCR-D-SEG-LINE -O OCR-D-OCR-TESS -p "$ocrd_tesserocr_recognize_parameters"
|
|
}
|
|
|
|
do_ocr_calamari() {
|
|
# Perform OCR on the segmented lines
|
|
|
|
ocrd_calamari_recognize_parameters='{
|
|
"checkpoint": "/var/lib/calamari-models/GT4HistOCR/2019-07-22T15:49+0200/*.ckpt.json",
|
|
"textequiv_level": "'$TEXTEQUIV_LEVEL'"
|
|
}'
|
|
ocrd workspace remove-group -rf OCR-D-OCR-CALAMARI
|
|
ocrd-calamari-recognize -I OCR-D-SEG-LINE -O OCR-D-OCR-CALAMARI -p "$ocrd_calamari_recognize_parameters"
|
|
}
|
|
|
|
page_validate_xml() {
|
|
# Validate all PAGE XML against the XML schema
|
|
|
|
filegrp=$1
|
|
|
|
local file
|
|
for file in `ocrd workspace find -G $filegrp`; do
|
|
XSD_DIR=`dirname $0`/xsd
|
|
if [ ! -d "$XSD_DIR" ]; then
|
|
XSD_DIR=/usr/share/xml
|
|
fi
|
|
xmllint --noout --schema $XSD_DIR/pagecontent.2019-07-15.xsd $file
|
|
done
|
|
}
|
|
|
|
page_upgrade_to_2019() {
|
|
# Upgrade PAGE files to 2019-07-15
|
|
|
|
filegrp=$1
|
|
|
|
local file
|
|
for file in `ocrd workspace find -G $filegrp`; do
|
|
sed -i 's#pagecontent/[0-9-]*#pagecontent/2019-07-15#g' $file
|
|
done
|
|
}
|
|
|
|
|
|
main() {
|
|
do_binarization
|
|
do_validate
|
|
|
|
|
|
do_linesegmentation_sbb
|
|
page_upgrade_to_2019 OCR-D-SEG-LINE
|
|
page_validate_xml OCR-D-SEG-REGION
|
|
page_validate_xml OCR-D-SEG-LINE
|
|
do_validate
|
|
|
|
|
|
do_ocr_calamari
|
|
|
|
|
|
do_ocr
|
|
|
|
|
|
for ocr_filegrp in OCR-D-OCR-CALAMARI OCR-D-OCR-TESS; do
|
|
|
|
page_validate_xml $ocr_filegrp
|
|
do_validate
|
|
|
|
if ocrd workspace list-group | grep -q OCR-D-GT-PAGE; then
|
|
ocrd workspace remove-group -rf $ocr_filegrp-EVAL
|
|
ocrd-dinglehopper -I OCR-D-GT-PAGE,$ocr_filegrp -O $ocr_filegrp-EVAL
|
|
fi
|
|
|
|
done
|
|
}
|
|
|
|
|
|
if [ "$LOG_LEVEL" = "DEBUG" -o "$LOG_LEVEL" = "TRACE" ]; then
|
|
pip3 list
|
|
fi
|
|
main
|
|
|
|
|
|
# vim:tw=120:
|