def marked_by_accuracy(request):
char = request.POST.get('char', None)
if char is None:
return JsonResponse({'status': 'error', 'msg': 'no char'})
l_value = int(request.POST.get('min_value'))
r_value = int(request.POST.get('max_value'))
if ((r_value <= 0) or (l_value == r_value)):
_mark_based_scope(l_value, char)
elif (l_value > r_value):
count = 0
else:
if (r_value > 500):
updateNum = Character.objects.filter(
char=char,
is_correct=0,
accuracy__gte=l_value,
accuracy__lte=r_value).update(is_correct=1)
else:
updateNum = Character.objects.filter(
char=char,
is_correct=0,
accuracy__gte=l_value,
accuracy__lte=r_value).update(is_correct=-1)
Character.update_statistics(char)
return JsonResponse({'status': 'ok'})
def run_seg(request, pk):
page = get_object_or_404(Page, pk=pk)
img_path = page.get_image_path()
raw_data = serializers.serialize("python",
Region.objects.filter(page_id=pk))
region_lst = [d['fields'] for d in raw_data]
char_lst = charseg(img_path, region_lst, pk)
for ch in char_lst:
char_id = '{0}L{1:02}R{2:02}C{3:03}'.format(pk, ch['line_no'],
ch['region_no'],
ch['char_no'])
character = Character(
id=char_id,
page_id=pk,
char=ch['char'],
left=ch['left'],
right=ch['right'],
top=ch['top'],
bottom=ch['bottom'],
line_no=ch['line_no'],
region_no=ch['region_no'],
char_no=ch['char_no'],
is_correct=-9 # has not chut char image
)
character.save()
cut_char.delay(pk) #offline task cut char image
return JsonResponse(char_lst, safe=False)
def filter_mark_by_recog(self, request):
checked_chars = Character.objects.filter(id__in=request.data["checked_ids"])
unchecked_chars = Character.objects.filter(id__in=request.data["unchecked_ids"])
result_type = request.data["type"]
if result_type == 0:
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_same = -2
char.save()
elif result_type == 1:
for char in checked_chars:
char.is_correct = -1
char.save()
for char in unchecked_chars:
char.is_same = 2
char.save()
elif result_type == 2:
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_correct = -1
char.is_same = 2
char.save()
Character.update_statistics(request.data["char"])
return Response({"status": "ok"})
def runSegment(request,page_id):
page = Page.objects.get(id=page_id)
image_name = page.image.url
#print image_name
text = page.text
image = io.imread(image_name, 0)
total_char_lst = process_page(image, text, page_id)
character_lst = []
temp_lst = []
line_lst = []
cur_line_no = 0
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(), page_id=page_id, char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left, right=ch.right,
top=ch.top, bottom=ch.bottom,
line_no=ch.line_no, char_no=ch.char_no,
is_correct=False)
character.width = character.right - character.left
character.height = character.bottom - character.top
if character.line_no != cur_line_no:
if temp_lst:
line = CharacterLine(cur_line_no, temp_lst[0].left, temp_lst[0].right, temp_lst)
line_lst.append(line)
cur_line_no = character.line_no
temp_lst = [character]
else:
temp_lst.append(character)
if temp_lst:
line = CharacterLine(cur_line_no, temp_lst[0].left, temp_lst[0].right, temp_lst)
line_lst.append(line)
json_line_lst = json.dumps(line_lst,cls=MyJsonEncoder)
return JsonResponse({ u'line_lst': json_line_lst}, safe=False)
def filter_mark_by_recog(self, request):
checked_chars = Character.objects.filter(
id__in=request.data['checked_ids'])
unchecked_chars = Character.objects.filter(
id__in=request.data['unchecked_ids'])
result_type = request.data['type']
if (result_type == 0):
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_same = -2
char.save()
elif (result_type == 1):
for char in checked_chars:
char.is_correct = -1
char.save()
for char in unchecked_chars:
char.is_same = 2
char.save()
elif (result_type == 2):
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_correct = -1
char.is_same = 2
char.save()
Character.update_statistics(request.data['char'])
return Response({'status': 'ok'})
def filter_mark_by_recog(self, request):
checked_chars = Character.objects.filter(id__in=request.data['checked_ids'])
unchecked_chars = Character.objects.filter(id__in=request.data['unchecked_ids'])
result_type = request.data['type']
if (result_type == 0):
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_same = -2
char.save()
elif (result_type == 1):
for char in checked_chars:
char.is_correct = -1
char.save()
for char in unchecked_chars:
char.is_same = 2
char.save()
elif (result_type == 2):
for char in checked_chars:
char.is_correct = 1
char.save()
for char in unchecked_chars:
char.is_correct = -1
char.is_same = 2
char.save()
Character.update_statistics(request.data['char'])
return Response({'status': 'ok'})
def segment_one_page(page_id, image_name, text):
if text == text1:
print 'equal----'
try:
image = io.imread(image_name, 0)
total_char_lst = process_page(image, text, page_id)
character_lst = []
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(),
page_id=page_id,
char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left,
right=ch.right,
top=ch.top,
bottom=ch.bottom,
line_no=ch.line_no,
char_no=ch.char_no,
is_correct=False)
#character_lst.append(character)
print character
character.save()
#Character.objects.bulk_create(character_lst)
except:
print 'missing image file'
def _mark_based_scope(l_value, char):
if l_value > 500:
updateNum = Character.objects.filter(
char=char, is_correct=0, accuracy=l_value).update(is_correct=1)
else:
updateNum = Character.objects.filter(
char=char, is_correct=0, accuracy=l_value).update(is_correct=-1)
Character.update_statistics(char)
def set_correct(request):
if 'id' in request.POST:
char_id = request.POST['id']
is_correct = int(request.POST['is_correct'])
char = request.POST['char']
char.encode('utf-8')
query_set = Character.objects.filter(id=char_id)
query_set.update(is_correct=is_correct)
Character.update_statistics(char)
record = CharMarkRecord.create(request.user, char_id, is_correct,
timezone.now())
record.save()
data = {'status': 'ok'}
elif (('e_charArr[]' in request.POST)
or ('c_charArr[]' in request.POST)): # uncheck -> check
check_char_number = request.session.get('check_char_number', 0)
request.session['check_char_number'] = check_char_number + 1
charArr = request.POST.getlist('e_charArr[]')
char = request.POST['char']
time = timezone.now()
records = []
if charArr:
query_set = Character.objects.filter(id__in=charArr)
query_set.update(is_correct=-1)
Character.update_statistics(char)
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, -1, time)
records.append(record)
CharMarkRecord.objects.bulk_create(records)
charArr = request.POST.getlist('c_charArr[]')
if charArr:
query_set = Character.objects.filter(id__in=charArr)
query_set.update(is_correct=1)
Character.update_statistics(char)
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, 1, time)
records.append(record)
CharMarkRecord.objects.bulk_create(records)
data = {'status': 'ok'}
elif ('cl_charArr[]' in request.POST):
c_num = int(request.POST['c_num'])
e_num = int(request.POST['e_num'])
unset_num = e_num + c_num
char = request.POST['char']
charArr = request.POST.getlist('cl_charArr[]')
Character.objects.filter(id__in=charArr).update(is_correct=0)
Character.update_statistics(char)
time = timezone.now()
records = []
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, 0, time)
records.append(record)
if charArr:
CharMarkRecord.objects.bulk_create(records)
data = {'status': 'ok', 'clear': 'ok'}
else:
data = {'status': 'error'}
return JsonResponse(data)
def set_correct(request):
if 'id' in request.POST:
char_id = request.POST['id']
is_correct = int(request.POST['is_correct'])
char = request.POST['char']
char.encode('utf-8')
query_set = Character.objects.filter(id=char_id)
query_set.update(is_correct=is_correct)
Character.update_statistics(char)
record = CharMarkRecord.create(request.user, char_id, is_correct, timezone.now())
record.save()
data = {'status': 'ok'}
elif (('e_charArr[]' in request.POST) or ('c_charArr[]' in request.POST)): # uncheck -> check
check_char_number = request.session.get('check_char_number', 0)
request.session['check_char_number'] = check_char_number+1
charArr = request.POST.getlist('e_charArr[]')
char = request.POST['char']
time = timezone.now()
records = []
if charArr:
query_set = Character.objects.filter(id__in =charArr)
query_set.update(is_correct=-1)
Character.update_statistics(char)
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, -1, time)
records.append(record)
CharMarkRecord.objects.bulk_create(records)
charArr = request.POST.getlist('c_charArr[]')
if charArr:
query_set = Character.objects.filter(id__in =charArr)
query_set.update(is_correct=1)
Character.update_statistics(char)
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, 1, time)
records.append(record)
CharMarkRecord.objects.bulk_create(records)
data = {'status': 'ok'}
elif ('cl_charArr[]' in request.POST):
c_num = int(request.POST['c_num'])
e_num = int(request.POST['e_num'])
unset_num = e_num + c_num;
char = request.POST['char']
charArr = request.POST.getlist('cl_charArr[]')
Character.objects.filter(id__in =charArr).update(is_correct=0)
Character.update_statistics(char)
time = timezone.now()
records = []
for char_id in charArr:
record = CharMarkRecord.create(request.user, char_id, 0, time)
records.append(record)
if charArr:
CharMarkRecord.objects.bulk_create(records)
data = {'status': 'ok', 'clear': 'ok'}
else:
data = {'status': 'error'}
return JsonResponse(data)
def marked_by_accuracy(request):
char = request.POST.get('char', None)
if char is None:
return JsonResponse({'status': 'error', 'msg': 'no char'})
l_value = int(request.POST.get('min_value'))
r_value = int(request.POST.get('max_value'))
if ((r_value <= 0) or (l_value == r_value)):
_mark_based_scope(l_value, char)
elif (l_value > r_value):
count = 0
else:
if (r_value > 500):
updateNum = Character.objects.filter(char=char, is_correct=0, accuracy__gte=l_value, accuracy__lte=r_value).update(is_correct=1)
else:
updateNum = Character.objects.filter(char=char, is_correct=0, accuracy__gte=l_value, accuracy__lte=r_value).update(is_correct=-1)
Character.update_statistics(char)
return JsonResponse({'status': 'ok'})
def test_file():
vol_no = u'01'
page_no = u'001'
output = subprocess.check_output(
'cd /media/DATA/work/image-preprocessing/binary_image/%s/%s/; ls *.txt 2>/dev/null'
% (vol_no, page_no),
shell=True)
pos = output.find('.txt')
if pos == -1:
sys.exit(-1)
page_id_prefix = output[:pos - 1]
page_id = page_id_prefix + 'a'
image_name = u'/media/DATA/work/image-preprocessing/binary_image/%s/%s/%s.jpg' % (
vol_no, page_no, page_id)
text_name = u'/media/DATA/work/image-preprocessing/binary_image/%s/%s/%s.txt' % (
vol_no, page_no, page_id)
image = io.imread(image_name, 0)
# image = io.imread(u'/home/xianbu/custom/1a.jpg', 0)
text = u''
with open(text_name, 'r') as f:
text = f.read().decode('utf-8')
text1 = text
process_page(image, text, page_id)
total_char_lst = process_page(image, text, page_id)
import django
django.setup()
from segmentation.models import Character
character_lst = []
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(),
page_id=page_id,
char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left,
right=ch.right,
top=ch.top,
bottom=ch.bottom,
line_no=ch.line_no,
char_no=ch.char_no,
is_correct=False)
#character_lst.append(character)
print character
character.save()
def runSegment(request, page_id):
page = Page.objects.get(id=page_id)
image_name = page.image.url
#print image_name
text = page.text
image = io.imread(image_name, 0)
total_char_lst = process_page(image, text, page_id)
character_lst = []
temp_lst = []
line_lst = []
cur_line_no = 0
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(),
page_id=page_id,
char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left,
right=ch.right,
top=ch.top,
bottom=ch.bottom,
line_no=ch.line_no,
char_no=ch.char_no,
is_correct=False)
character.width = character.right - character.left
character.height = character.bottom - character.top
if character.line_no != cur_line_no:
if temp_lst:
line = CharacterLine(cur_line_no, temp_lst[0].left,
temp_lst[0].right, temp_lst)
line_lst.append(line)
cur_line_no = character.line_no
temp_lst = [character]
else:
temp_lst.append(character)
if temp_lst:
line = CharacterLine(cur_line_no, temp_lst[0].left, temp_lst[0].right,
temp_lst)
line_lst.append(line)
json_line_lst = json.dumps(line_lst, cls=MyJsonEncoder)
return JsonResponse({u'line_lst': json_line_lst}, safe=False)
def segment_one_page(page_id, image_name, text):
if text == text1:
print 'equal----'
try:
image = io.imread(image_name, 0)
total_char_lst = process_page(image, text, page_id)
character_lst = []
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(), page_id=page_id, char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left, right=ch.right,
top=ch.top, bottom=ch.bottom,
line_no=ch.line_no, char_no=ch.char_no,
is_correct=False)
#character_lst.append(character)
print character
character.save()
#Character.objects.bulk_create(character_lst)
except:
print 'missing image file'
def test_file():
vol_no = u'01'
page_no = u'001'
output = subprocess.check_output(
'cd /media/DATA/work/image-preprocessing/binary_image/%s/%s/; ls *.txt 2>/dev/null' % (vol_no, page_no),
shell=True)
pos = output.find('.txt')
if pos == -1:
sys.exit(-1)
page_id_prefix = output[:pos - 1]
page_id = page_id_prefix + 'a'
image_name = u'/media/DATA/work/image-preprocessing/binary_image/%s/%s/%s.jpg' % (vol_no, page_no, page_id)
text_name = u'/media/DATA/work/image-preprocessing/binary_image/%s/%s/%s.txt' % (vol_no, page_no, page_id)
image = io.imread(image_name, 0)
# image = io.imread(u'/home/xianbu/custom/1a.jpg', 0)
text = u''
with open(text_name, 'r') as f:
text = f.read().decode('utf-8')
text1 = text
process_page(image, text, page_id)
total_char_lst = process_page(image, text, page_id)
import django
django.setup()
from segmentation.models import Character
character_lst = []
for ch in total_char_lst:
character = Character(id=ch.char_id.strip(), page_id=page_id, char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left, right=ch.right,
top=ch.top, bottom=ch.bottom,
line_no=ch.line_no, char_no=ch.char_no,
is_correct=False)
#character_lst.append(character)
print character
character.save()
def run_seg(request,pk):
page = get_object_or_404(Page, pk=pk)
img_path = page.get_image_path()
raw_data = serializers.serialize("python",Region.objects.filter(page_id=pk))
region_lst = [d['fields'] for d in raw_data]
char_lst = charseg(img_path,region_lst, pk)
for ch in char_lst:
char_id = '{0}L{1:02}R{2:02}C{3:03}'.format(pk,ch['line_no'],ch['region_no'],ch['char_no'])
character = Character(
id=char_id,
page_id= pk,
char=ch['char'],
left=ch['left'],
right=ch['right'],
top=ch['top'],
bottom=ch['bottom'],
line_no=ch['line_no'],
region_no = ch['region_no'],
char_no=ch['char_no'],
is_correct=-9 # has not chut char image
)
character.save()
cut_char.delay(pk)#offline task cut char image
return JsonResponse(char_lst, safe=False)
def segment_one_page(page_id, image_name, text):
image_path = u'/home/share/dzj_characters/page_images/%s' % image_name
image = io.imread(image_path, 0)
total_char_lst = process_page(image, text, page_id)
character_lst = []
for ch in total_char_lst:
path = u'/home/share/dzj_characters/character_images/%s.jpg' % ch.char_id.strip(
)
ch.cut_char_from_page(image, path)
character = Character(id=ch.char_id.strip(),
page_id=page_id,
char=ch.char,
image=ch.char_id.strip() + u'.jpg',
left=ch.left,
right=ch.right,
top=ch.top,
bottom=ch.bottom,
line_no=ch.line_no,
char_no=ch.char_no,
is_correct=0)
character_lst.append(character)
Character.objects.filter(page_id=page_id).delete()
Character.objects.bulk_create(character_lst)
def recog_all():
for char in CharacterStatistics.objects.order_by('-total_cnt').values_list('char',flat=True)[0:500]:
print(char)
Character.recog_characters(char)
def recog_all():
for char in CharacterStatistics.objects.order_by('-total_cnt').values_list(
'char', flat=True)[0:500]:
print(char)
Character.recog_characters(char)
def _mark_based_scope(l_value, char):
if l_value > 500:
updateNum = Character.objects.filter(char=char, is_correct=0, accuracy=l_value).update(is_correct=1)
else:
updateNum = Character.objects.filter(char=char, is_correct=0, accuracy=l_value).update(is_correct=-1)
Character.update_statistics(char)
