Extend totaltext converter to support text fields (#728)

* Extend totaltext converter to support text fieldols/

* fix bug

* fix comment typo

Co-authored-by: Tong Gao <gaotongxiao@gmail.com>

Co-authored-by: Tong Gao <gaotongxiao@gmail.com>
pull/691/merge
liukuikun 2022-01-14 16:00:53 +08:00 committed by GitHub
parent c736989615
commit 2f429d5e40
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 9 additions and 10 deletions

View File

@ -87,7 +87,8 @@ def get_contours_mat(gt_path):
contours = []
words = []
data = scio.loadmat(gt_path)
data_polygt = data['polygt']
# 'gt' for the latest version; 'polygt' for the legacy version
data_polygt = data.get('polygt', data['gt'])
for i, lines in enumerate(data_polygt):
X = np.array(lines[1])
@ -129,9 +130,9 @@ def load_mat_info(img_info, gt_file):
assert isinstance(img_info, dict)
assert isinstance(gt_file, str)
contours, words = get_contours_mat(gt_file)
contours, texts = get_contours_mat(gt_file)
anno_info = []
for contour in contours:
for contour, text in zip(contours, texts):
if contour.shape[0] == 2:
continue
category_id = 1
@ -149,6 +150,7 @@ def load_mat_info(img_info, gt_file):
category_id=category_id,
bbox=bbox,
area=area,
text=text,
segmentation=[contour])
anno_info.append(anno)
@ -239,11 +241,7 @@ def get_contours_txt(gt_path):
if tmp_line != '':
contours, words = process_line(tmp_line, contours, words)
for word in words:
if word == '#':
word = '###'
continue
words = ['###' if word == '#' else word for word in words]
return contours, words
@ -259,9 +257,9 @@ def load_txt_info(gt_file, img_info):
img_info(dict): The dict of the img and annotation information
"""
contours, words = get_contours_txt(gt_file)
contours, texts = get_contours_txt(gt_file)
anno_info = []
for contour in contours:
for contour, text in zip(contours, texts):
if contour.shape[0] == 2:
continue
category_id = 1
@ -279,6 +277,7 @@ def load_txt_info(gt_file, img_info):
category_id=category_id,
bbox=bbox,
area=area,
text=text,
segmentation=[contour])
anno_info.append(anno)