驗(yàn)證碼生成
# coding: utf-8
# 驗(yàn)證碼生成庫
from captcha.image import ImageCaptcha # pip install captcha
import random
import sys
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
# alphabet = ['a','b','c','d','e','f','g','h','i','j','k','l','m','n','o','p','q','r','s','t','u','v','w','x','y','z']
# ALPHABET = ['A','B','C','D','E','F','G','H','I','J','K','L','M','N','O','P','Q','R','S','T','U','V','W','X','Y','Z']
def random_captcha_text(char_set=number, captcha_size=4):
# 驗(yàn)證碼列表（驗(yàn)證碼長度為4位，captcha_size=4）
captcha_text = []
for i in range(captcha_size):
# 隨機(jī)選擇
c = random.choice(char_set)
# 加入驗(yàn)證碼列表
captcha_text.append(c)
return captcha_text
# 生成字符對應(yīng)的驗(yàn)證碼
def gen_captcha_text_and_image():
image = ImageCaptcha()
# 獲得隨機(jī)生成的驗(yàn)證碼
captcha_text = random_captcha_text()
# 把驗(yàn)證碼列表轉(zhuǎn)為字符串
captcha_text = ''.join(captcha_text)
# 生成驗(yàn)證碼
captcha = image.generate(captcha_text)
image.write(captcha_text, 'captcha/images/' + captcha_text + '.jpg') # 寫到文件
# 數(shù)量少于10000，因?yàn)橹孛?num = 10000
if __name__ == '__main__':
for i in range(num):
gen_captcha_text_and_image()
sys.stdout.write('\r>> Creating image %d/%d' % (i + 1, num))
sys.stdout.flush()
sys.stdout.write('\n')
sys.stdout.flush()
print("生成完畢")
生成結(jié)果：
四位隨機(jī)數(shù)字驗(yàn)證碼
生成tfrecord 文件
# coding: utf-8
import tensorflow as tf
import os
import random
import sys
from PIL import Image
import numpy as np
# 驗(yàn)證集數(shù)量
_NUM_TEST = 500
# 隨機(jī)種子
_RANDOM_SEED = 0
# 數(shù)據(jù)集路徑
DATASET_DIR = "captcha/images/"
# tfrecord文件存放路徑
TFRECORD_DIR = "captcha/"
# 判斷tfrecord文件是否存在
def _dataset_exists(dataset_dir):
for split_name in ['train', 'test']:
output_filename = os.path.join(dataset_dir, split_name + '.tfrecords')
if not tf.gfile.Exists(output_filename):
return False
return True
# 獲取所有驗(yàn)證碼圖片
def _get_filenames_and_classes(dataset_dir):
photo_filenames = []
for filename in os.listdir(dataset_dir):
# 獲取文件路徑
path = os.path.join(dataset_dir, filename)
photo_filenames.append(path)
return photo_filenames
def int64_feature(values):
if not isinstance(values, (tuple, list)):
values = [values]
return tf.train.Feature(int64_list=tf.train.Int64List(value=values))
def bytes_feature(values):
return tf.train.Feature(bytes_list=tf.train.BytesList(value=[values]))
def image_to_tfexample(image_data, label0, label1, label2, label3):
# Abstract base class for protocol messages.
return tf.train.Example(features=tf.train.Features(feature={
'image': bytes_feature(image_data),
'label0': int64_feature(label0),
'label1': int64_feature(label1),
'label2': int64_feature(label2),
'label3': int64_feature(label3),
}))
# 把數(shù)據(jù)轉(zhuǎn)為TFRecord格式
def _convert_dataset(split_name, filenames, dataset_dir):
assert split_name in ['train', 'test']
with tf.Session() as sess:
# 定義tfrecord文件的路徑+名字
output_filename = os.path.join(TFRECORD_DIR, split_name + '.tfrecords')
with tf.python_io.TFRecordWriter(output_filename) as tfrecord_writer:
for i, filename in enumerate(filenames):
try:
sys.stdout.write('\r>> Converting image %d/%d' % (i + 1, len(filenames)))
sys.stdout.flush()
# 讀取圖片
image_data = Image.open(filename)
# 根據(jù)模型的結(jié)構(gòu)resize
image_data = image_data.resize((224, 224))
# 灰度化
image_data = np.array(image_data.convert('L'))
# 將圖片轉(zhuǎn)化為bytes
image_data = image_data.tobytes()
# 獲取label
labels = filename.split('/')[-1][0:4]
num_labels = []
for j in range(4):
num_labels.append(int(labels[j]))
# 生成protocol數(shù)據(jù)類型
example = image_to_tfexample(image_data, num_labels[0], num_labels[1], num_labels[2], num_labels[3])
tfrecord_writer.write(example.SerializeToString())
except IOError as e:
print('Could not read:', filename)
print('Error:', e)
print('Skip it\n')
sys.stdout.write('\n')
sys.stdout.flush()
# 判斷tfrecord文件是否存在
if _dataset_exists(TFRECORD_DIR):
print('tfcecord文件已存在')
else:
# 獲得所有圖片
photo_filenames = _get_filenames_and_classes(DATASET_DIR)
# 把數(shù)據(jù)切分為訓(xùn)練集和測試集,并打亂
random.seed(_RANDOM_SEED)
random.shuffle(photo_filenames)
training_filenames = photo_filenames[_NUM_TEST:]
testing_filenames = photo_filenames[:_NUM_TEST]
# 數(shù)據(jù)轉(zhuǎn)換
_convert_dataset('train', training_filenames, DATASET_DIR)
_convert_dataset('test', testing_filenames, DATASET_DIR)
print('生成tfcecord文件')
生成的tfrecord文件
驗(yàn)證碼識別的兩種方式
-
把標(biāo)簽轉(zhuǎn)為向量，向量長度為40。比如有一個(gè)驗(yàn)證碼為0782，
它的標(biāo)簽可以轉(zhuǎn)為長度為40的向量：1000000000 0000000100 0000000010 0010000000
訓(xùn)練方法跟0-9手寫數(shù)字識別類似。 使用多任務(wù)學(xué)習(xí)的方式
拆分為4個(gè)標(biāo)簽，比如有一個(gè)驗(yàn)證碼為0782
Label0：1000000000
Label1：0000000100
Label2：0000000010
Label3：0010000000
多任務(wù)學(xué)習(xí)一般有兩種方式：
1 .Multi-task Learning - 交替訓(xùn)練
適用于不同的任務(wù)有不同的訓(xùn)練集
2 .Multi-task Learning – 聯(lián)合訓(xùn)練
適用于不同的任務(wù)有相同的數(shù)據(jù)集
訓(xùn)練
# coding: utf-8
import tensorflow as tf
from nets import nets_factory
# 不同字符數(shù)量
CHAR_SET_LEN = 10
# 圖片高度
IMAGE_HEIGHT = 60
# 圖片寬度
IMAGE_WIDTH = 160
# 批次
BATCH_SIZE = 25
# tfrecord文件存放路徑
TFRECORD_FILE = "captcha/train.tfrecords"
# placeholder
x = tf.placeholder(tf.float32, [None, 224, 224])
y0 = tf.placeholder(tf.float32, [None])
y1 = tf.placeholder(tf.float32, [None])
y2 = tf.placeholder(tf.float32, [None])
y3 = tf.placeholder(tf.float32, [None])
# 學(xué)習(xí)率
lr = tf.Variable(0.003, dtype=tf.float32)
# 從tfrecord讀出數(shù)據(jù)
def read_and_decode(filename):
# 根據(jù)文件名生成一個(gè)隊(duì)列
filename_queue = tf.train.string_input_producer([filename])
reader = tf.TFRecordReader()
# 返回文件名和文件
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(serialized_example,
features={
'image': tf.FixedLenFeature([], tf.string),
'label0': tf.FixedLenFeature([], tf.int64),
'label1': tf.FixedLenFeature([], tf.int64),
'label2': tf.FixedLenFeature([], tf.int64),
'label3': tf.FixedLenFeature([], tf.int64),
})
# 獲取圖片數(shù)據(jù)
image = tf.decode_raw(features['image'], tf.uint8)
# tf.train.shuffle_batch必須確定shape
image = tf.reshape(image, [224, 224])
# 圖片預(yù)處理
image = tf.cast(image, tf.float32) / 255.0
image = tf.subtract(image, 0.5)
image = tf.multiply(image, 2.0)
# 獲取label
label0 = tf.cast(features['label0'], tf.int32)
label1 = tf.cast(features['label1'], tf.int32)
label2 = tf.cast(features['label2'], tf.int32)
label3 = tf.cast(features['label3'], tf.int32)
return image, label0, label1, label2, label3
# 獲取圖片數(shù)據(jù)和標(biāo)簽
image, label0, label1, label2, label3 = read_and_decode(TFRECORD_FILE)
# 使用shuffle_batch可以隨機(jī)打亂
image_batch, label_batch0, label_batch1, label_batch2, label_batch3 = tf.train.shuffle_batch(
[image, label0, label1, label2, label3], batch_size=BATCH_SIZE,
capacity=50000, min_after_dequeue=10000, num_threads=1)
# 定義網(wǎng)絡(luò)結(jié)構(gòu)
train_network_fn = nets_factory.get_network_fn(
'alexnet_v2',
num_classes=CHAR_SET_LEN,
weight_decay=0.0005,
is_training=True)
with tf.Session() as sess:
# inputs: a tensor of size [batch_size, height, width, channels]
X = tf.reshape(x, [BATCH_SIZE, 224, 224, 1])
# 數(shù)據(jù)輸入網(wǎng)絡(luò)得到輸出值
logits0, logits1, logits2, logits3, end_points = train_network_fn(X)
# 把標(biāo)簽轉(zhuǎn)成one_hot的形式
one_hot_labels0 = tf.one_hot(indices=tf.cast(y0, tf.int32), depth=CHAR_SET_LEN)
one_hot_labels1 = tf.one_hot(indices=tf.cast(y1, tf.int32), depth=CHAR_SET_LEN)
one_hot_labels2 = tf.one_hot(indices=tf.cast(y2, tf.int32), depth=CHAR_SET_LEN)
one_hot_labels3 = tf.one_hot(indices=tf.cast(y3, tf.int32), depth=CHAR_SET_LEN)
# 計(jì)算loss
loss0 = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits0, labels=one_hot_labels0))
loss1 = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits1, labels=one_hot_labels1))
loss2 = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits2, labels=one_hot_labels2))
loss3 = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits=logits3, labels=one_hot_labels3))
# 計(jì)算總的loss
total_loss = (loss0 + loss1 + loss2 + loss3) / 4.0
# 優(yōu)化total_loss
optimizer = tf.train.AdamOptimizer(learning_rate=lr).minimize(total_loss)
# 計(jì)算準(zhǔn)確率
correct_prediction0 = tf.equal(tf.argmax(one_hot_labels0, 1), tf.argmax(logits0, 1))
accuracy0 = tf.reduce_mean(tf.cast(correct_prediction0, tf.float32))
correct_prediction1 = tf.equal(tf.argmax(one_hot_labels1, 1), tf.argmax(logits1, 1))
accuracy1 = tf.reduce_mean(tf.cast(correct_prediction1, tf.float32))
correct_prediction2 = tf.equal(tf.argmax(one_hot_labels2, 1), tf.argmax(logits2, 1))
accuracy2 = tf.reduce_mean(tf.cast(correct_prediction2, tf.float32))
correct_prediction3 = tf.equal(tf.argmax(one_hot_labels3, 1), tf.argmax(logits3, 1))
accuracy3 = tf.reduce_mean(tf.cast(correct_prediction3, tf.float32))
# 用于保存模型
saver = tf.train.Saver()
# 初始化
sess.run(tf.global_variables_initializer())
# 創(chuàng)建一個(gè)協(xié)調(diào)器，管理線程
coord = tf.train.Coordinator()
# 啟動QueueRunner, 此時(shí)文件名隊(duì)列已經(jīng)進(jìn)隊(duì)
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
for i in range(6001):
# 獲取一個(gè)批次的數(shù)據(jù)和標(biāo)簽
b_image, b_label0, b_label1, b_label2, b_label3 = sess.run(
[image_batch, label_batch0, label_batch1, label_batch2, label_batch3])
# 優(yōu)化模型
sess.run(optimizer, feed_dict={x: b_image, y0: b_label0, y1: b_label1, y2: b_label2, y3: b_label3})
# 每迭代20次計(jì)算一次loss和準(zhǔn)確率
if i % 20 == 0:
# 每迭代2000次降低一次學(xué)習(xí)率
if i % 2000 == 0:
sess.run(tf.assign(lr, lr / 3))
acc0, acc1, acc2, acc3, loss_ = sess.run([accuracy0, accuracy1, accuracy2, accuracy3, total_loss],
feed_dict={x: b_image,
y0: b_label0,
y1: b_label1,
y2: b_label2,
y3: b_label3})
learning_rate = sess.run(lr)
print("Iter:%d Loss:%.3f Accuracy:%.2f,%.2f,%.2f,%.2f Learning_rate:%.4f" % (
i, loss_, acc0, acc1, acc2, acc3, learning_rate))
# 保存模型
# if i == 6000:
if acc0 > 0.90 and acc1 > 0.90 and acc2 > 0.90 and acc3 > 0.90:
saver.save(sess, "captcha/models/crack_captcha.model", global_step=i)
break
# 通知其他線程關(guān)閉
coord.request_stop()
# 其他所有線程關(guān)閉之后，這一函數(shù)才能返回
coord.join(threads)
- 測試
# coding: utf-8
import tensorflow as tf
from PIL import Image
from nets import nets_factory
import matplotlib.pyplot as plt
# 不同字符數(shù)量
CHAR_SET_LEN = 10
# 圖片高度
IMAGE_HEIGHT = 60
# 圖片寬度
IMAGE_WIDTH = 160
# 批次
BATCH_SIZE = 1
# tfrecord文件存放路徑
TFRECORD_FILE = "captcha/test.tfrecords"
# placeholder
x = tf.placeholder(tf.float32, [None, 224, 224])
# 從tfrecord讀出數(shù)據(jù)
def read_and_decode(filename):
# 根據(jù)文件名生成一個(gè)隊(duì)列
filename_queue = tf.train.string_input_producer([filename])
reader = tf.TFRecordReader()
# 返回文件名和文件
_, serialized_example = reader.read(filename_queue)
features = tf.parse_single_example(serialized_example,
features={
'image': tf.FixedLenFeature([], tf.string),
'label0': tf.FixedLenFeature([], tf.int64),
'label1': tf.FixedLenFeature([], tf.int64),
'label2': tf.FixedLenFeature([], tf.int64),
'label3': tf.FixedLenFeature([], tf.int64),
})
# 獲取圖片數(shù)據(jù)
image = tf.decode_raw(features['image'], tf.uint8)
# 沒有經(jīng)過預(yù)處理的灰度圖
image_raw = tf.reshape(image, [224, 224])
# tf.train.shuffle_batch必須確定shape
image = tf.reshape(image, [224, 224])
# 圖片預(yù)處理
image = tf.cast(image, tf.float32) / 255.0
image = tf.subtract(image, 0.5)
image = tf.multiply(image, 2.0)
# 獲取label
label0 = tf.cast(features['label0'], tf.int32)
label1 = tf.cast(features['label1'], tf.int32)
label2 = tf.cast(features['label2'], tf.int32)
label3 = tf.cast(features['label3'], tf.int32)
return image, image_raw, label0, label1, label2, label3
# 獲取圖片數(shù)據(jù)和標(biāo)簽
image, image_raw, label0, label1, label2, label3 = read_and_decode(TFRECORD_FILE)
# 使用shuffle_batch可以隨機(jī)打亂
image_batch, image_raw_batch, label_batch0, label_batch1, label_batch2, label_batch3 = tf.train.shuffle_batch(
[image, image_raw, label0, label1, label2, label3], batch_size=BATCH_SIZE,
capacity=50000, min_after_dequeue=10000, num_threads=1)
# 定義網(wǎng)絡(luò)結(jié)構(gòu)
train_network_fn = nets_factory.get_network_fn(
'alexnet_v2',
num_classes=CHAR_SET_LEN,
weight_decay=0.0005,
is_training=False)
with tf.Session() as sess:
# inputs: a tensor of size [batch_size, height, width, channels]
X = tf.reshape(x, [BATCH_SIZE, 224, 224, 1])
# 數(shù)據(jù)輸入網(wǎng)絡(luò)得到輸出值
logits0, logits1, logits2, logits3, end_points = train_network_fn(X)
# 預(yù)測值
predict0 = tf.reshape(logits0, [-1, CHAR_SET_LEN])
predict0 = tf.argmax(predict0, 1)
predict1 = tf.reshape(logits1, [-1, CHAR_SET_LEN])
predict1 = tf.argmax(predict1, 1)
predict2 = tf.reshape(logits2, [-1, CHAR_SET_LEN])
predict2 = tf.argmax(predict2, 1)
predict3 = tf.reshape(logits3, [-1, CHAR_SET_LEN])
predict3 = tf.argmax(predict3, 1)
# 初始化
sess.run(tf.global_variables_initializer())
# 載入訓(xùn)練好的模型
saver = tf.train.Saver()
saver.restore(sess, 'captcha/models/crack_captcha.model-6000')
# 創(chuàng)建一個(gè)協(xié)調(diào)器，管理線程
coord = tf.train.Coordinator()
# 啟動QueueRunner, 此時(shí)文件名隊(duì)列已經(jīng)進(jìn)隊(duì)
threads = tf.train.start_queue_runners(sess=sess, coord=coord)
for i in range(10):
# 獲取一個(gè)批次的數(shù)據(jù)和標(biāo)簽
b_image, b_image_raw, b_label0, b_label1, b_label2, b_label3 = sess.run([image_batch,
image_raw_batch,
label_batch0,
label_batch1,
label_batch2,
label_batch3])
# 顯示圖片
img = Image.fromarray(b_image_raw[0], 'L')
plt.imshow(img)
plt.axis('off')
plt.show()
# 打印標(biāo)簽
print('label:', b_label0, b_label1, b_label2, b_label3)
# 預(yù)測
label0, label1, label2, label3 = sess.run([predict0, predict1, predict2, predict3], feed_dict={x: b_image})
# 打印預(yù)測值
print('predict:', label0, label1, label2, label3)
# 通知其他線程關(guān)閉
coord.request_stop()
# 其他所有線程關(guān)閉之后，這一函數(shù)才能返回
coord.join(threads)
