qurator-spk/page2tsv
1
0
Fork 0
mirror of https://github.com/qurator-spk/page2tsv.git synced 2025-06-09 19:39:54 +02:00
Code Issues Releases Wiki Activity

try other coordinate computation

Browse source
This commit is contained in:
Kai 2021-05-07 10:40:52 +02:00
parent 1eb05d0d62
commit f320904503
1 changed files with 21 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

24
tsvtools/cli.py
View file

@ -106,7 +106,15 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
for region_idx, region in enumerate(pcgts.get_Page().get_AllRegions(classes=['Text'], order='reading-order')):
for text_line in region.get_TextLine():
left, top, right, bottom = [int(scale_factor * x) for x in bbox_from_points(text_line.get_Coords().points)]
points = [int(scale_factor * float(pos)) for coords in text_line.get_Coords() for p in
coords.attrib['points'].split(' ') for pos in p.split(',')]
x_points, y_points = points[0::2], points[1::2]
left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points)
# left, top, right, bottom = [int(scale_factor * x) for x in bbox_from_points(text_line.get_Coords().points)]
if min_confidence is not None and max_confidence is not None:
conf = np.max([textequiv.conf for textequiv in text_line.get_TextEquiv()])
@ -119,7 +127,15 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
for text_equiv in word.get_TextEquiv():
# transform OCR coordinates using `scale_factor` to derive
# correct coordinates for the web presentation image
left, top, right, bottom = [int(scale_factor * x) for x in bbox_from_points(word.get_Coords().points)]
points = [int(scale_factor * float(pos)) for coords in text_line.get_Coords() for p in
coords.attrib['points'].split(' ') for pos in p.split(',')]
x_points, y_points = points[0::2], points[1::2]
left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points)
# left, top, right, bottom = [int(scale_factor * x) for x in bbox_from_points(word.get_Coords().points)]
tsv.append((region_idx, len(line_info) - 1, left + (right - left) / 2.0,
text_equiv.get_Unicode(), len(urls), left, right, top, bottom, text_line.id))
@ -181,8 +197,10 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
except requests.HTTPError as e:
print(e)
@click.command()
@click.option('--output-filename', '-o', help="Output filename. If omitted, PAGE-XML filename with .corrected.xml extension")
@click.option('--output-filename', '-o', help="Output filename. "
"If omitted, PAGE-XML filename with .corrected.xml extension")
@click.option('--keep-words', '-k', is_flag=True, help="Keep (out-of-date) Words of TextLines")
@click.argument('page-file')
@click.argument('tsv-file')