Fix bugs in traintessv4.sh

1 files changed, 7 insertions, 4 deletions

diff --git a/traintessv4.sh b/traintessv4.sh
index 948397b..9e4e8f4 100755
--- a/traintessv4.sh
+++ b/traintessv4.sh
@@ -37,11 +37,14 @@ done
 
 echo "Making box files"
 find "$name/gt" "$name/eval" -type f -name '*txt' | while read i; do
-	b=`basename "$i"`
+	b=`basename "$i" .txt`
 	d=`dirname "$i"`
-	test -f "$b.tif" && n="$b.tif"
-	test -f "$b.png" && n="$b.png"
-	python generate_line_box.py -i "$d/$n" -t "$i" > "$d/$b.box" || exit 1
+	n=""
+	test -f "$d/$b.tif" && n="$b.tif"
+	test -f "$d/$b.png" && n="$b.png"
+	test -z "$n" && echo "Skipping $i as no corresponding image found" && continue
+	test -f "$d/$b.box" && echo "Skipping $i as box file already present" && continue
+	python ~/training/generate_line_box.py -i "$d/$n" -t "$i" > "$d/$b.box" || exit 1
 done
 
 echo "Making unicharset"