【TensorFlow】4位纯数字或数字字母组合验证码识别
本文做了一个识别简单验证码的小例子，具体思路如下：
- 使用captcha类库随机生成4位纯数字或数字、大小写组合的验证码；
- 以第一步生成的验证码为训练数据，对3层卷积、2层全连接模型进行训练；
- 当训练的正确率达到规定的阈值时，保存模型，并停止训练。
说明：
- 若验证码为4位纯数字，则模型要完成的分类数为4*10=40分类任务；
- 若验证码为4为数字+大写+小写组成，则为4*(10+26+26)=248分类任务；
- 本文未对4位验证码进行切割操作（即：把一个验证码切割成4个子图），而是把整个验证码进行训练、预测；
- 本文的模型若想应用在实际工程中，可以使用爬虫工具下载工程中的验证码，并人工对其打标签，标签可以以验证码的命名体现。在读取验证码时，只需修改少许函数即可；
- 4位纯数字验证码识别和数字、字母混合识别类似，只是分类结果后者较多。
captcha生成验证码示例
captchaIdentify.py：
import tensorflow as tf
from captcha.image import ImageCaptcha # 生成验证码的库
import numpy as np
import matplotlib.pyplot as plt
from PIL import Image
import random
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 + alphabet + ALPHABET, captcha_size=4):
captcha_text = []
for i in range(captcha_size):
c = random.choice(char_set)
captcha_text.append(c)
return captcha_text
def gen_captcha_text_and_image():
image = ImageCaptcha()
captcha_text = random_captcha_text()
captcha_text = ''.join(captcha_text) # list转字符串
captcha = image.generate(captcha_text)
# image.write(captcha_text, captcha_text + '.jpg')
captcha_image = Image.open(captcha)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
if __name__ == '__main__':
text, image = gen_captcha_text_and_image()
f = plt.figure()
ax = f.add_subplot(111) # 等价于f.add_subplot(1,1,1)，画布分成1行1列。从左至右，从上往下数，图片在第一块处显示
ax.text(0.1, 0.9, text, ha='center', va='center', transform=ax.transAxes) # text显示在坐标为(0.1,0.9),va：垂直分布情况,ha：水平分布情况
plt.imshow(image)
plt.show()
效果如下：
4位纯数字验证码网络模型
onlyNumber.py：
import numpy as np
import tensorflow as tf
from captcha.image import ImageCaptcha
import matplotlib.pyplot as plt
from PIL import Image
import random
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
def random_captcha_text(char_set=number, captcha_size=4):
captcha_text = []
for i in range(captcha_size):
c = random.choice(char_set)
captcha_text.append(c)
return captcha_text
def gen_captcha_text_and_image():
image = ImageCaptcha()
captcha_text = random_captcha_text()
captcha_text = ''.join(captcha_text)
captcha = image.generate(captcha_text)
# image.write(captcha_text, captcha_text + '.jpg')
captcha_image = Image.open(captcha)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def convert2gray(img):
if len(img.shape) > 2:
gray = np.mean(img, -1)
# 上面的转法较快，正规转法如下
# r, g, b = img[:, :, 0], img[:, :, 1], img[:, :, 2]
# gray = 0.2989 * r + 0.5870 * g + 0.1140 * b
return gray
else:
return img
def text2vec(text):
text_len = len(text)
if text_len > MAX_CAPTCHA:
raise ValueError('验证码最长4个字符')
vector = np.zeros(MAX_CAPTCHA * CHAR_SET_LEN)
'''
def char2pos(c):
if c == '_':
k = 62
return k
k = ord(c) - 48
if k > 9:
k = ord(c) - 55
if k > 35:
k = ord(c) - 61
if k > 61:
raise ValueError('No Map')
return k
'''
for i, c in enumerate(text):
idx = i * CHAR_SET_LEN + int(c)
vector[idx] = 1
return vector
# 向量转回文本
def vec2text(vec):
'''
char_pos = vec.nonzero()[0]
text = []
for i, c in enumerate(char_pos):
char_at_pos = i # c/63
char_idx = c % CHAR_SET_LEN
if char_idx < 10:
char_code = char_idx + ord('0')
elif char_idx < 36:
char_code = char_idx - 10 + ord('A')
elif char_idx < 62:
char_code = char_idx - 36 + ord('a')
elif char_idx == 62:
char_code = ord('_')
else:
raise ValueError('error')
text.append(chr(char_code))
'''
text = []
char_pos = vec.nonzero()[0]
for i, c in enumerate(char_pos):
number = i % 10
text.append(str(number))
return ''.join(text)
'''
# 向量（大小MAX_CAPTCHA*CHAR_SET_LEN）用0,1编码 每63个编码一个字符，这样顺利有，字符也有
vec = text2vec('F5Sd')
text = vec2text(vec)
print(text) # F5Sd
vec = text2vec('SFd5')
text = vec2text(vec)
print(text) # SFd5
'''
# 生成一个训练batch
def get_next_batch(batch_size=128):
batch_x = np.zeros([batch_size, IMAGE_HEIGHT * IMAGE_WIDTH])
batch_y = np.zeros([batch_size, MAX_CAPTCHA * CHAR_SET_LEN])
# 有时生成图像大小不是（60,160,3）
def wrap_gen_captcha_text_and_image():
while True:
text, image = gen_captcha_text_and_image()
if image.shape == (60, 160, 3):
return text, image
for i in range(batch_size):
text, image = wrap_gen_captcha_text_and_image()
image = convert2gray(image)
batch_x[i, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0，取值在0和1之间
batch_y[i, :] = text2vec(text) # label转成独热编码
return batch_x, batch_y
# 定义CNN
def crack_captcha_cnn(w_alpha=0.01, b_alpha=0.1):
x = tf.reshape(X, shape=[-1, IMAGE_HEIGHT, IMAGE_WIDTH, 1]) # TensorFlow处理的格式为四维
# w_c1_alpha = np.sqrt(2.0 / (IMAGE_HEIGHT * IMAGE_WIDTH))
# w_c2_alpha = np.sqrt(2.0 / (3 * 3 * 32))
# w_c3_alpha = np.sqrt(2.0 / (3 * 3 * 64))
# w_d1_alpha = np.sqrt(2.0 / (8 * 32 * 64))
# out_alpha = np.sqrt(2.0 / 1024)
# 3 conv layer
W_c1 = tf.Variable(w_alpha * tf.random_normal([3, 3, 1, 32]))
b_c1 = tf.Variable(b_alpha * tf.random_normal([32]))
conv1 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(x, W_c1, strides=[1, 1, 1, 1], padding='SAME'), b_c1)) # padding经常用SAME
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv1 = tf.nn.dropout(conv1, keep_prob)
W_c2 = tf.Variable(w_alpha * tf.random_normal([3, 3, 32, 64]))
b_c2 = tf.Variable(b_alpha * tf.random_normal([64]))
conv2 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv1, W_c2, strides=[1, 1, 1, 1], padding='SAME'), b_c2))
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv2 = tf.nn.dropout(conv2, keep_prob)
W_c3 = tf.Variable(w_alpha * tf.random_normal([3, 3, 64, 64]))
b_c3 = tf.Variable(b_alpha * tf.random_normal([64]))
conv3 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv2, W_c3, strides=[1, 1, 1, 1], padding='SAME'), b_c3))
conv3 = tf.nn.max_pool(conv3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv3 = tf.nn.dropout(conv3, keep_prob)
# Fully connected layer
W_d = tf.Variable(w_alpha * tf.random_normal([8 * 20 * 64, 1024])) # 60->30->15->8；160->80->40->20
b_d = tf.Variable(b_alpha * tf.random_normal([1024]))
dense = tf.reshape(conv3, [-1, W_d.get_shape().as_list()[0]])
dense = tf.nn.relu(tf.add(tf.matmul(dense, W_d), b_d))
dense = tf.nn.dropout(dense, keep_prob)
W_out = tf.Variable(w_alpha * tf.random_normal([1024, MAX_CAPTCHA * CHAR_SET_LEN]))
b_out = tf.Variable(b_alpha * tf.random_normal([MAX_CAPTCHA * CHAR_SET_LEN]))
out = tf.add(tf.matmul(dense, W_out), b_out)
return out
# 训练
def train_crack_captcha_cnn():
output = crack_captcha_cnn()
loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=output, labels=Y))
optimizer = tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
predict = tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN])
max_idx_p = tf.argmax(predict, 2) # 在矩阵第2轴找最大值的索引，即预测值
max_idx_l = tf.argmax(tf.reshape(Y, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
correct_pred = tf.equal(max_idx_p, max_idx_l)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
saver = tf.train.Saver()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
step = 0
while True:
batch_x, batch_y = get_next_batch(64)
_, loss_ = sess.run([optimizer, loss], feed_dict={X: batch_x, Y: batch_y, keep_prob: 0.75})
print(step, loss_)
# 每100 step计算一次准确率
if step % 100 == 0:
batch_x_test, batch_y_test = get_next_batch(100)
acc = sess.run(accuracy, feed_dict={X: batch_x_test, Y: batch_y_test, keep_prob: 1.})
print(step, acc)
# 如果准确率大于90%, 保存模型，完成训练
if acc > 0.90:
saver.save(sess, './model/crack_capche.model', global_step=step) # 保存session计算域所有的内容。迭代次数step作为模型的名字
break
step += 1
def crack_captcha(captcha_image):
output = crack_captcha_cnn()
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess, './model/crack_capche.model-2150')
predict = tf.argmax(tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
text_list = sess.run(predict, feed_dict={X: [captcha_image], keep_prob: 1})
text = text_list[0].tolist()
return text
if __name__ == '__main__':
train = 0 # 0：训练；1：验证。使用变量控制训练或验证，可以复用代码
if train == 0:
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
text, image = gen_captcha_text_and_image()
print('验证码图像channel:', image.shape) # (60, 160, 3)
# 图像大小,验证码长、宽
IMAGE_HEIGHT = 60
IMAGE_WIDTH = 160
MAX_CAPTCHA = len(text)
print('验证码文本最长字符数', MAX_CAPTCHA)
# 文本转向量
# char_set = number +alphabet + ALPHABET + ['_'] # 如果验证码长度小于4，'_'用来补齐
char_set = number
CHAR_SET_LEN = len(char_set)
X = tf.placeholder(tf.float32, [None, IMAGE_HEIGHT * IMAGE_WIDTH])
Y = tf.placeholder(tf.float32, [None, MAX_CAPTCHA * CHAR_SET_LEN])
keep_prob = tf.placeholder(tf.float32) # dropout
train_crack_captcha_cnn()
if train == 1:
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
IMAGE_HEIGHT = 60
IMAGE_WIDTH = 160
char_set = number
CHAR_SET_LEN = len(char_set)
text, image = gen_captcha_text_and_image()
f = plt.figure()
ax = f.add_subplot(111)
ax.text(0.1, 0.9, text, ha='center', va='center', transform=ax.transAxes)
plt.imshow(image)
plt.show()
MAX_CAPTCHA = len(text)
image = convert2gray(image)
image = image.flatten() / 255
X = tf.placeholder(tf.float32, [None, IMAGE_HEIGHT * IMAGE_WIDTH])
Y = tf.placeholder(tf.float32, [None, MAX_CAPTCHA * CHAR_SET_LEN]) # 验证码由4个数字组成，每个数字用独热编码，所以每个Y的长度为4*10
keep_prob = tf.placeholder(tf.float32) # dropout
predict_text = crack_captcha(image)
print('正确：{} 预测：{}'.format(text, predict_text))
部分运行结果：
验证码图像channel: (60, 160, 3)
验证码文本最长字符数 4
0 0.68183994
0 0.11
1 0.51985824
2 0.34787816
3 0.41765136
4 0.41940632
5 0.3634311
6 0.34169158
7 0.34222344
8 0.36967105
9 0.3652032
10 0.3539911
11 0.34141412
12 0.33093944
13 0.33646545
14 0.33852464
15 0.34308758
16 0.34396297
17 0.3367651
18 0.33524507
19 0.330844
20 0.332375474位数字+大小写字母验证码网络模型
train.py：
import numpy as np
import tensorflow as tf
from captcha.image import ImageCaptcha
import matplotlib.pyplot as plt
from PIL import Image
import random
number = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']
def random_captcha_text(char_set=number, captcha_size=4):
captcha_text = []
for i in range(captcha_size):
c = random.choice(char_set)
captcha_text.append(c)
return captcha_text
def gen_captcha_text_and_image():
image = ImageCaptcha()
captcha_text = random_captcha_text()
captcha_text = ''.join(captcha_text)
captcha = image.generate(captcha_text)
# image.write(captcha_text, captcha_text + '.jpg')
captcha_image = Image.open(captcha)
captcha_image = np.array(captcha_image)
return captcha_text, captcha_image
def convert2gray(img):
if len(img.shape) > 2:
gray = np.mean(img, -1)
# 上面的转法较快，正规转法如下
# r, g, b = img[:, :, 0], img[:, :, 1], img[:, :, 2]
# gray = 0.2989 * r + 0.5870 * g + 0.1140 * b
return gray
else:
return img
def text2vec(text):
text_len = len(text)
if text_len > MAX_CAPTCHA:
raise ValueError('验证码最长4个字符')
vector = np.zeros(MAX_CAPTCHA * CHAR_SET_LEN)
def char2pos(c):
if c == '_':
k = 62
return k
k = ord(c) - 48
if k > 9:
k = ord(c) - 55
if k > 35:
k = ord(c) - 61
if k > 61:
raise ValueError('No Map')
return k
for i, c in enumerate(text):
idx = i * CHAR_SET_LEN + char2pos(c)
vector[idx] = 1
return vector
# 向量转回文本
def vec2text(vec):
char_pos = vec.nonzero()[0]
text = []
for i, c in enumerate(char_pos):
char_at_pos = i # c/63
char_idx = c % CHAR_SET_LEN
if char_idx < 10:
char_code = char_idx + ord('0')
elif char_idx < 36:
char_code = char_idx - 10 + ord('A')
elif char_idx < 62:
char_code = char_idx - 36 + ord('a')
elif char_idx == 62:
char_code = ord('_')
else:
raise ValueError('error')
text.append(chr(char_code))
return ''.join(text)
'''
# 向量（大小MAX_CAPTCHA*CHAR_SET_LEN）用0,1编码 每63个编码一个字符，这样顺利有，字符也有
vec = text2vec('F5Sd')
text = vec2text(vec)
print(text) # F5Sd
vec = text2vec('SFd5')
text = vec2text(vec)
print(text) # SFd5
'''
# 生成一个训练batch
def get_next_batch(batch_size=128):
batch_x = np.zeros([batch_size, IMAGE_HEIGHT * IMAGE_WIDTH])
batch_y = np.zeros([batch_size, MAX_CAPTCHA * CHAR_SET_LEN])
# 有时生成图像大小不是（60,160,3）
def wrap_gen_captcha_text_and_image():
while True:
text, image = gen_captcha_text_and_image()
if image.shape == (60, 160, 3):
return text, image
for i in range(batch_size):
text, image = wrap_gen_captcha_text_and_image()
image = convert2gray(image)
batch_x[i, :] = image.flatten() / 255 # (image.flatten()-128)/128 mean为0
batch_y[i, :] = text2vec(text)
return batch_x, batch_y
# 定义CNN
def crack_captcha_cnn(w_alpha=0.01, b_alpha=0.1):
x = tf.reshape(X, shape=[-1, IMAGE_HEIGHT, IMAGE_WIDTH, 1])
# w_c1_alpha = np.sqrt(2.0 / (IMAGE_HEIGHT * IMAGE_WIDTH))
# w_c2_alpha = np.sqrt(2.0 / (3 * 3 * 32))
# w_c3_alpha = np.sqrt(2.0 / (3 * 3 * 64))
# w_d1_alpha = np.sqrt(2.0 / (8 * 32 * 64))
# out_alpha = np.sqrt(2.0 / 1024)
# 3 conv layer
W_c1 = tf.Variable(w_alpha * tf.random_normal([3, 3, 1, 32]))
b_c1 = tf.Variable(b_alpha * tf.random_normal([32]))
conv1 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(x, W_c1, strides=[1, 1, 1, 1], padding='SAME'), b_c1))
conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv1 = tf.nn.dropout(conv1, keep_prob)
W_c2 = tf.Variable(w_alpha * tf.random_normal([3, 3, 32, 64]))
b_c2 = tf.Variable(b_alpha * tf.random_normal([64]))
conv2 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv1, W_c2, strides=[1, 1, 1, 1], padding='SAME'), b_c2))
conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv2 = tf.nn.dropout(conv2, keep_prob)
W_c3 = tf.Variable(w_alpha * tf.random_normal([3, 3, 64, 64]))
b_c3 = tf.Variable(b_alpha * tf.random_normal([64]))
conv3 = tf.nn.relu(tf.nn.bias_add(tf.nn.conv2d(conv2, W_c3, strides=[1, 1, 1, 1], padding='SAME'), b_c3))
conv3 = tf.nn.max_pool(conv3, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
conv3 = tf.nn.dropout(conv3, keep_prob)
# Fully connected layer
W_d = tf.Variable(w_alpha * tf.random_normal([8 * 20 * 64, 1024]))
b_d = tf.Variable(b_alpha * tf.random_normal([1024]))
dense = tf.reshape(conv3, [-1, W_d.get_shape().as_list()[0]])
dense = tf.nn.relu(tf.add(tf.matmul(dense, W_d), b_d))
dense = tf.nn.dropout(dense, keep_prob)
W_out = tf.Variable(w_alpha * tf.random_normal([1024, MAX_CAPTCHA * CHAR_SET_LEN]))
b_out = tf.Variable(b_alpha * tf.random_normal([MAX_CAPTCHA * CHAR_SET_LEN]))
out = tf.add(tf.matmul(dense, W_out), b_out)
return out
# 训练
def train_crack_captcha_cnn():
output = crack_captcha_cnn()
# loss
# loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(output, Y))
loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=output, labels=Y))
# 最后一层用来分类的softmax和sigmoid有什么不同？
# optimizer为了加快训练learning_rate应该开始大，然后慢慢衰减
# optimizer = tf.train.AdamOptimizer(learning_rate=0.001).minimize(loss)
optimizer = tf.train.AdamOptimizer().minimize(loss)
predict = tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN])
max_idx_p = tf.argmax(predict, 2)
max_idx_l = tf.argmax(tf.reshape(Y, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
correct_pred = tf.equal(max_idx_p, max_idx_l)
accuracy = tf.reduce_mean(tf.cast(correct_pred, tf.float32))
saver = tf.train.Saver()
with tf.Session() as sess:
sess.run(tf.global_variables_initializer())
step = 0
while True:
batch_x, batch_y = get_next_batch(64)
_, loss_ = sess.run([optimizer, loss], feed_dict={X: batch_x, Y: batch_y, keep_prob: 0.75})
print(step, loss_)
# 每100 step计算一次准确率
if step % 10 == 0:
batch_x_test, batch_y_test = get_next_batch(100)
acc = sess.run(accuracy, feed_dict={X: batch_x_test, Y: batch_y_test, keep_prob: 1.})
print(step, acc)
# 如果准确率大于50%, 保存模型，完成训练
if acc > 0.50:
saver.save(sess, './model/crack_capche.model', global_step=step)
break
step += 1
if __name__ == '__main__':
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']
text, image = gen_captcha_text_and_image()
print('验证码图像channel:', image.shape) # (60, 160, 3)
# 图像大小
IMAGE_HEIGHT = 60
IMAGE_WIDTH = 160
MAX_CAPTCHA = len(text)
print('验证码文本最长字符数', MAX_CAPTCHA)
# 文本转向量
# char_set = number +alphabet + ALPHABET + ['_'] # 如果验证码长度小于4，'_'用来补齐
char_set = number
CHAR_SET_LEN = len(char_set)
X = tf.placeholder(tf.float32, [None, IMAGE_HEIGHT * IMAGE_WIDTH])
Y = tf.placeholder(tf.float32, [None, MAX_CAPTCHA * CHAR_SET_LEN])
keep_prob = tf.placeholder(tf.float32) # dropout
train_crack_captcha_cnn()
运行结果同上。