Discuz验证码识别

Discuz/get_discuz.py

@@ -0,0 +1,48 @@
+#-*- coding:utf-8 -*-
+from urllib.request import urlretrieve
+import time, random, os
+
+class Discuz():
+	def __init__(self):
+		# Discuz验证码生成图片地址
+		self.url = 'http://cuijiahua.com/tutrial/discuz/index.php?label='
+
+	def random_captcha_text(self, captcha_size = 4):
+		"""
+		验证码一般都无视大小写；验证码长度4个字符
+		Parameters:
+			captcha_size:验证码长度
+		Returns:
+			captcha_text:验证码字符串
+		"""
+		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']
+		char_set = number + alphabet
+		captcha_text = []
+		for i in range(captcha_size):
+			c = random.choice(char_set)
+			captcha_text.append(c)
+		captcha_text = ''.join(captcha_text)
+		return captcha_text
+
+	def download_discuz(self, nums = 5000):
+		"""
+		下载验证码图片
+		Parameters:
+			nums:下载的验证码图片数量
+		"""
+		dirname = './Discuz'
+		if dirname not in os.listdir():
+			os.mkdir(dirname)
+		for i in range(nums):
+			label = self.random_captcha_text()
+			print('第%d张图片:%s下载' % (i + 1,label))
+			urlretrieve(url = self.url + label, filename = dirname + '/' + label + '.jpg')
+			# 请至少加200ms延时，避免给我的服务器造成过多的压力，如发现影响服务器正常工作，我会关闭此功能。
+			# 你好我也好，大家好才是真的好！
+			time.sleep(0.2)
+		print('恭喜图片下载完成！')
+
+if __name__ == '__main__':
+	dz = Discuz()
+	dz.download_discuz()

Discuz/test.py

@@ -0,0 +1,36 @@
+#-*- coding:utf-8 -*-
+import tensorflow as tf
+import numpy as np
+import train
+
+def crack_captcha(captcha_image, captcha_label):
+	"""
+	使用模型做预测
+	Parameters:
+		captcha_image:数据
+		captcha_label:标签
+	"""
+
+	output = dz.crack_captcha_cnn()
+	saver = tf.train.Saver()
+	with tf.Session(config=dz.config) as sess:
+
+		saver.restore(sess, tf.train.latest_checkpoint('.'))
+		for i in range(len(captcha_label)):
+			img = captcha_image[i].flatten()
+			label = captcha_label[i]
+			predict = tf.argmax(tf.reshape(output, [-1, dz.max_captcha, dz.char_set_len]), 2)
+			text_list = sess.run(predict, feed_dict={dz.X: [img], dz.keep_prob: 1})
+			text = text_list[0].tolist()
+			vector = np.zeros(dz.max_captcha*dz.char_set_len)
+			i = 0
+			for n in text:
+					vector[i*dz.char_set_len + n] = 1
+					i += 1
+			prediction_text = dz.vec2text(vector)
+			print("正确: {}  预测: {}".format(dz.vec2text(label), prediction_text))
+
+if __name__ == '__main__':
+	dz = train.Discuz()
+	batch_x, batch_y = dz.get_next_batch(False, 5)
+	crack_captcha(batch_x, batch_y)