qurator-spk/page2tsv
1
0
Fork 0
mirror of https://github.com/qurator-spk/page2tsv.git synced 2025-07-03 23:29:52 +02:00
Code Issues Releases Wiki Activity

use max confidence instead of mean

Browse source
This commit is contained in:
Kai 2021-02-04 07:37:58 +01:00
parent 85ec36218e
commit 5d55ba24a3
1 changed files with 4 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

11
tsvtools/cli.py
View file

@ -100,12 +100,9 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
tsv = [] tsv = []
line_info = [] line_info = []
line_number = 0 for rgn_number, region in enumerate(tree.findall('.//{%s}TextRegion' % xmlns)):
rgn_number = 0
for region in tree.findall('.//{%s}TextRegion' % xmlns): for line_number, text_line in enumerate(region.findall('.//{%s}TextLine' % xmlns)):
rgn_number += 1
for text_line in region.findall('.//{%s}TextLine' % xmlns):
line_number += 1
points = [int(scale_factor * float(pos)) for coords in text_line.findall('./{%s}Coords' % xmlns) for p in points = [int(scale_factor * float(pos)) for coords in text_line.findall('./{%s}Coords' % xmlns) for p in
coords.attrib['points'].split(' ') for pos in p.split(',')] coords.attrib['points'].split(' ') for pos in p.split(',')]
@ -115,7 +112,7 @@ def page2tsv(page_xml_file, tsv_out_file, purpose, image_url, ner_rest_endpoint,
left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points) left, right, top, bottom = min(x_points), max(x_points), min(y_points), max(y_points)
if min_confidence is not None and max_confidence is not None: if min_confidence is not None and max_confidence is not None:
conf = np.mean([float(text.attrib['conf']) for text in text_line.findall('./{%s}TextEquiv' % xmlns)]) conf = np.max([float(text.attrib['conf']) for text in text_line.findall('./{%s}TextEquiv' % xmlns)])
else: else:
conf = np.nan conf = np.nan