def get_train_word_lens_data(count):
logger.info('Start get train word lens')
X_train = []
Y_train = []
fn_lst = get_fn_lst(conf.CAPTCHA_SRC_DIR)
num = 0
for fn in fn_lst[:count]:
word = get_word_from_fn(fn)
image = get_image(fn)
train_image = get_word_train_image(image)
X_train.append([train_image])
Y_train.append(len(word))
num += 1
if num % 1000 == 0:
logger.info('Get train word lens: complete %s', num)
imsave('/opt/data/log_train_word_lens_data' + conf.CAPTCHA_EXTENSION,
X_train[0][0])
return X_train, Y_train
def make_letters_from_all():
for index, fn in enumerate(get_fn_lst(conf.CAPTCHA_SRC_DIR)):
word = get_word_from_fn(fn)
word_len = len(word)
image = get_image(fn)
word_image = get_word_image_for_letters(image)
try:
letter_images = make_letters(word_image, word_len)
except LettersError:
continue
for p in range(word_len):
letter = word[p]
letter_folder = conf.CAPTCHA_TRAIN_DIR + letter.encode('utf-8') + '/'
try:
os.mkdir(letter_folder)
except OSError:
pass
letter_image = letter_images[p]
letter_fn = letter_folder + str(p) + '-' + word.encode('utf-8') + conf.CAPTCHA_EXTENSION
imsave(letter_fn, letter_image)
if index % 100 == 0:
logger.info('Process on image: %s', index)
def get_train_letters_data(count):
X_train = []
Y_train = []
logger.info('Start get train letters')
sorted_letters_folders_lst = filter(
lambda l: len(l.decode('utf-8')) == 1,
sorted(os.listdir(conf.CAPTCHA_TRAIN_DIR)))
letters_folders_lst = list(enumerate(sorted_letters_folders_lst))
random.shuffle(letters_folders_lst)
for num, letter in letters_folders_lst:
letter_folder = conf.CAPTCHA_TRAIN_DIR + letter + '/'
letters_fn_lst = os.listdir(letter_folder)
random.shuffle(letters_fn_lst)
for fn in letters_fn_lst[:count]:
letter_fn = letter_folder + fn
image = get_image(letter_fn)
train_image = get_letter_train_image(image)
X_train.append([train_image])
Y_train.append(num)
imsave('/opt/data/log_train_letter_image' + conf.CAPTCHA_EXTENSION,
train_image)
logger.info('Get train letters %s: complete %s', letter, num)
return X_train, Y_train