def segment_draw(segment_org_path, labeled_img_path, label_path, show=True):
label = pd.read_csv(label_path)
if len(label) == 0:
print("No component needs to be labeled \n")
return
for s in range(label.iloc[-1].segment_no + 1):
seg_input_path = os.path.join(segment_org_path, str(s) + '.png')
seg_output_path = os.path.join(labeled_img_path, str(s) + '.png')
seg_img = cv2.imread(seg_input_path)
seg_label = label[label['segment_no'] == s]
draw.label(seg_label, seg_img, seg_output_path, show)
print('Labeled img saved in ' + labeled_img_path)
img_segment_path = os.path.join(segment_root, str(index))
label_path = os.path.join(label_root, str(index) + '.csv')
# catch label and screenshot img and segment them into smaller size
img, label = None, None
catch_success = False
if is_catch_element and '.com' in links.iloc[index]:
# set the format of libel
libel_format = pd.read_csv(os.path.join(data_position, 'format.csv'), index_col=0)
url = 'http://' + links.iloc[index] if 'http://' not in links.iloc[index] else links.iloc[index]
try:
img, label = catch.catch(url, label_path, img_org_path, libel_format, driver_path)
except FunctionTimedOut:
print('Catch Time Out')
continue
# segment the lengthy images
if is_segment and img is not None:
seg.segment_img(img, 600, img_segment_path, 0)
# read and draw label on segment img
if is_draw_label and img is not None and label is not None:
draw.label(label, img, img_drawn_path)
end_time = time.clock()
print("*** %d Time taken:%ds ***\n" % (index, int(end_time - start_time)))
if index > end_pos:
break
if is_read_existed_links:
# read links
csv = pd.read_csv(os.path.join(data_position, 'preset_500_2.csv'))
links = csv.link
print("*** Links Fetched ***\n")
start_pos = 1
for index in range(start_pos, len(links)):
start_time = time.clock()
# set path
org_img_path = os.path.join(img_root, str(index) + '.png')
drawn_img_path = os.path.join(drawn_root, str(index) + '.png')
label_path = os.path.join(label_root, str(index) + '.csv')
# catch label and screenshot img and segment them into smaller size
catch_success = False
if is_catch_element:
# set the format of libel
libel_format = pd.read_csv(os.path.join(data_position, 'format.csv'), index_col=0)
url = 'http://' + links.iloc[index]
img, label = catch.catch(url, label_path, org_img_path, libel_format, driver_path)
# read and draw label on segment img
if is_draw_label and img is not None and label is not None:
draw.label(label, img, drawn_img_path)
end_time = time.clock()
print("*** Time taken:%ds ***\n" % int(end_time - start_time))