def run_all_uwiii():
# run all UW-III images
output_dir = "fullpage/"
result_list = []
for imgfilename in sys.argv[1:]:
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename(basename)
# zone box files use "ZONE" instead of "BIN"
# e.g.,
# A00ABIN_300_010_2990.png -> A00AZONE_300_010_2990.xml
gtruth_xml_filename = uwiii_xmldir + "{0}.xml".format(
basename.replace("BIN", "ZONE"))
# print basename, document_id, gtruth_xml_filename
result = run_image_with_gtruth(imgfilename, gtruth_xml_filename)
result_list.append(result)
# save pickled file so can do interesting things with the results later
# (especially if we crash)
output = open("uwiii.pkl", "wb")
pickle.dump(result_list, output)
output.close()
xmlfilename = result["output_xml_file"]
# did we crash? if so don't bother trying to draw the XML
if xmlfilename == "failed":
print "{0} failed!".format(imgfilename)
continue
# draw the resulting XML onto the original input image
out_imgfilename = output_dir + "{0}_zones.png".format(document_id)
if os.path.exists(out_imgfilename):
print "{0} already exists; not redrawing".format(out_imgfilename)
continue
drawxml.draw_zones(xmlfilename, imgfilename, out_imgfilename)
print "wrote", out_imgfilename
outfile = open("uwiii.dat", "w")
for result in result_list:
print >> outfile, "{0} {1} {2}".format(result["output_image_file"],
result["output_xml_file"],
result["metric"])
outfile.close()
def run_file_list(img_filelist, xml_filelist, output_dir, output_basename):
result_list = []
pickle_filename = os.path.join(output_dir, output_basename + ".pkl")
dat_filename = os.path.join(output_dir, output_basename + ".dat")
for imgfilename, gtruth_xml_filename in zip(img_filelist, xml_filelist):
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename(basename)
result = run_image_with_gtruth(imgfilename, gtruth_xml_filename,
output_dir)
result_list.append(result)
# save pickled file so can do interesting things with the results later
# (especially if we crash)
output = open(pickle_filename, "wb")
pickle.dump(result_list, output)
output.close()
xmlfilename = result["output_xml_file"]
# did we crash? if so don't bother trying to draw the XML
if xmlfilename == "failed":
print "{0} failed!".format(imgfilename)
continue
# draw the resulting XML onto the original input image
out_imgfilename = os.path.join(output_dir,
"{0}_zones.png".format(basename))
if os.path.exists(out_imgfilename):
print "{0} already exists; not redrawing".format(out_imgfilename)
continue
drawxml.draw_zones(xmlfilename, imgfilename, out_imgfilename)
print "wrote", out_imgfilename
outfile = open(dat_filename, "w")
for result in result_list:
print >> outfile, "{0} {1} {2}".format(result["output_image_file"],
result["output_xml_file"],
result["metric"])
outfile.close()
def run_all_uwiii():
# run all UW-III images
output_dir = "fullpage/"
result_list = []
for imgfilename in sys.argv[1:]:
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename(basename)
# zone box files use "ZONE" instead of "BIN"
# e.g.,
# A00ABIN_300_010_2990.png -> A00AZONE_300_010_2990.xml
gtruth_xml_filename = uwiii_xmldir + "{0}.xml".format(basename.replace("BIN", "ZONE"))
# print basename, document_id, gtruth_xml_filename
result = run_image_with_gtruth(imgfilename, gtruth_xml_filename)
result_list.append(result)
# save pickled file so can do interesting things with the results later
# (especially if we crash)
output = open("uwiii.pkl", "wb")
pickle.dump(result_list, output)
output.close()
xmlfilename = result["output_xml_file"]
# did we crash? if so don't bother trying to draw the XML
if xmlfilename == "failed":
print "{0} failed!".format(imgfilename)
continue
# draw the resulting XML onto the original input image
out_imgfilename = output_dir + "{0}_zones.png".format(document_id)
if os.path.exists(out_imgfilename):
print "{0} already exists; not redrawing".format(out_imgfilename)
continue
drawxml.draw_zones(xmlfilename, imgfilename, out_imgfilename)
print "wrote", out_imgfilename
outfile = open("uwiii.dat", "w")
for result in result_list:
print >> outfile, "{0} {1} {2}".format(result["output_image_file"], result["output_xml_file"], result["metric"])
outfile.close()
def run_file_list(img_filelist, xml_filelist, output_dir, output_basename):
result_list = []
pickle_filename = os.path.join(output_dir, output_basename + ".pkl")
dat_filename = os.path.join(output_dir, output_basename + ".dat")
for imgfilename, gtruth_xml_filename in zip(img_filelist, xml_filelist):
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename(basename)
result = run_image_with_gtruth(imgfilename, gtruth_xml_filename, output_dir)
result_list.append(result)
# save pickled file so can do interesting things with the results later
# (especially if we crash)
output = open(pickle_filename, "wb")
pickle.dump(result_list, output)
output.close()
xmlfilename = result["output_xml_file"]
# did we crash? if so don't bother trying to draw the XML
if xmlfilename == "failed":
print "{0} failed!".format(imgfilename)
continue
# draw the resulting XML onto the original input image
out_imgfilename = os.path.join(output_dir, "{0}_zones.png".format(basename))
if os.path.exists(out_imgfilename):
print "{0} already exists; not redrawing".format(out_imgfilename)
continue
drawxml.draw_zones(xmlfilename, imgfilename, out_imgfilename)
print "wrote", out_imgfilename
outfile = open(dat_filename, "w")
for result in result_list:
print >> outfile, "{0} {1} {2}".format(result["output_image_file"], result["output_xml_file"], result["metric"])
outfile.close()
def run(imgfilename):
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename(basename)
# destination for the output files
make_output_dir(document_id)
# stripnum = get_stripnum_from_filename( basename )
out_imgfilename = output_dir + "{0}_rast.png".format(basename)
xml_filename = output_dir + "{0}_rast.xml".format(basename)
input_dir = "{0}/{1}/".format(num_rows_in_strip, document_id)
# zone box files use "ZONE" instead of "BIN"
# e.g.,
# A00ABIN_300_010_2990.png -> A00AZONE_300_010_2990.xml
gtruth_xml_filename = input_dir + "{0}.xml".format(
basename.replace("BIN", "ZONE"))
print "imgfilename=", imgfilename
print "out_imgfilename=", out_imgfilename
print "xml_filename=", xml_filename
print "gtruth_xml_filename=", gtruth_xml_filename
# sys.exit(0)
# segment the image
cmd = "./rast-ocropus {0} {1}".format(imgfilename, out_imgfilename)
print cmd
try:
result = subprocess.check_output(cmd, shell=True)
except subprocess.CalledProcessError:
return (imgfilename, "failed")
# remove some clutter
os.unlink(out_imgfilename)
# write the XML results
with open(xml_filename, "w") as outfile:
print >> outfile, result
print "wrote", xml_filename
# run the compare
cmd = "runZoneComp -g {0} -d {1}".format(gtruth_xml_filename, xml_filename)
print cmd
result = subprocess.check_output(cmd, shell=True)
# get the segmentation metric from the output
metric = parse_runZoneComp_result(result)
print "metric={0}".format(metric)
# draw the experimental result onto the input image
out_imgfilename = output_dir + "{0}_rast_zone.png".format(basename)
fname = drawxml.draw_zones(xml_filename, imgfilename, out_imgfilename)
print "wrote", fname
# remove some clutter
# os.unlink(xml_filename)
return (imgfilename, metric)
def run( imgfilename ) :
basename = get_basename(imgfilename)
document_id = get_document_id_from_basename( basename )
# destination for the output files
make_output_dir(document_id)
# stripnum = get_stripnum_from_filename( basename )
out_imgfilename = output_dir + "{0}_rast.png".format( basename )
xml_filename = output_dir + "{0}_rast.xml".format( basename )
input_dir = "{0}/{1}/".format( num_rows_in_strip, document_id )
# zone box files use "ZONE" instead of "BIN"
# e.g.,
# A00ABIN_300_010_2990.png -> A00AZONE_300_010_2990.xml
gtruth_xml_filename = input_dir + "{0}.xml".format( basename.replace("BIN","ZONE") )
print "imgfilename=",imgfilename
print "out_imgfilename=",out_imgfilename
print "xml_filename=",xml_filename
print "gtruth_xml_filename=",gtruth_xml_filename
# sys.exit(0)
# segment the image
cmd = "./rast-ocropus {0} {1}".format( imgfilename, out_imgfilename )
print cmd
try :
result = subprocess.check_output( cmd, shell=True )
except subprocess.CalledProcessError :
return (imgfilename,"failed")
# remove some clutter
os.unlink(out_imgfilename)
# write the XML results
with open(xml_filename,"w") as outfile :
print >>outfile, result
print "wrote", xml_filename
# run the compare
cmd = "runZoneComp -g {0} -d {1}".format( gtruth_xml_filename, xml_filename )
print cmd
result = subprocess.check_output( cmd, shell=True )
# get the segmentation metric from the output
metric = parse_runZoneComp_result( result )
print "metric={0}".format( metric )
# draw the experimental result onto the input image
out_imgfilename = output_dir + "{0}_rast_zone.png".format( basename )
fname = drawxml.draw_zones( xml_filename, imgfilename, out_imgfilename )
print "wrote", fname
# remove some clutter
# os.unlink(xml_filename)
return (imgfilename,metric)
