qurator-spk/train-calamari-gt4histocr
1
0
Fork 0
mirror of https://github.com/qurator-spk/train-calamari-gt4histocr.git synced 2025-10-31 17:24:13 +01:00
Code Issues Releases Wiki Activity

⬆ train-calamari-gt4histocr: Update to Calamari 1

Browse source
This commit is contained in:
Gerber, Mike 2019-12-09 14:28:07 +01:00
parent 5a51ca4f20
commit ccb7f34ab7
3 changed files with 14 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

2
data

@ -1 +1 @@
Subproject commit f817209ba765464adb132a132774ea7856d53f4e
Subproject commit b0c0cd08551ba1828ea73833034188047f3e14ab

4
requirements.txt
View file

@ -1,2 +1,2 @@
calamari-ocr==0.3.5
tensorflow-gpu==1.13.1
calamari-ocr==1.0.1
tensorflow-gpu==2.0.0

11
train.sh
View file

@ -35,6 +35,17 @@ done
echo "Removing dta19/1882-keller_sinngedicht/04970.nrm.png (Broken PNG)"
rm -f $TMPDIR/dta19/1882-keller_sinngedicht/04970.*
# If we're just testing, keep just some files
if [ "$TEST" = 1 ]; then
num_pngs_wanted=2000
num_pngs=`find "$TMPDIR" -path "$TMPDIR/*/*/*.png" | wc -l`
num_pngs_to_delete=$(($num_pngs-$num_pngs_wanted))
echo "TEST = 1, Reducing dataset from $num_pngs to $num_pngs_wanted PNG files"
find "$TMPDIR" -path "$TMPDIR/*/*/*.png" | shuf -n $num_pngs_to_delete | xargs rm
fi
export PYTHONUNBUFFERED=1 # For python + tee
outdir=$DATA_SUBDIR/calamari-models/GT4HistOCR