Use ocrd-tesserocr-segment-*

my_ocrd_workflow

@@ -40,18 +40,24 @@ do_fontident() {
 }
 
 do_linesegmentation() {
+  remove_filegrp OCR-D-SEG-REGION mets.xml
   remove_filegrp OCR-D-SEG-LINE mets.xml
-  ocrd-ocropy-segment -l $LOG_LEVEL \
-    -m mets.xml -I OCR-D-IMG -O OCR-D-SEG-LINE
+  #ocrd-ocropy-segment -l $LOG_LEVEL \
+  #  -m mets.xml -I OCR-D-IMG -O OCR-D-SEG-LINE
+  # XXX ocrd-ocropy-segment throws an exception for buerger_gedichte_1778.ocrd
+  #ocrd workspace validate mets.xml
+
+  ocrd-tesserocr-segment-region -l $LOG_LEVEL \
+    -m mets.xml -I OCR-D-IMG -O OCR-D-SEG-REGION
   #ocrd workspace validate mets.xml
+  ocrd-tesserocr-segment-line -l $LOG_LEVEL \
+    -m mets.xml -I OCR-D-SEG-REGION -O OCR-D-SEG-LINE
+  #ocrd workspace validate mets.xml
+  # XXX compare ocrd-tesserocr-segment* vs tesseract native
 
   # XXX This leaves copies of the images at the top level of the workspace, because it "downloads" the "remote" files.
-  #     Clean it up.
+  #     Clean it up. (Maybe only affects ocrd-ocropy-segment)
   find . -maxdepth 1 -name "OCR-D-IMG*" -type f -exec rm -v {} \;
-
-  # XXX ocrd-tesserocr-segment-line does not seem to produce any line segmentation
-  # XXX mv {ocrd-ocropy-segment,-line}
-  # XXX ocrd-ocropy-segment throws an exception for buerger_gedichte_1778.ocrd
 }
 
 do_ocr() {