From c7a21cf68607327631f4c7833bacfdeb0de32bf3 Mon Sep 17 00:00:00 2001 From: luyishisi <543429245@qq.com> Date: Wed, 8 Mar 2017 19:13:58 +0800 Subject: [PATCH] =?UTF-8?q?=E6=B5=8B=E8=AF=95TensorFlow=5Fcnn=E2=80=94?= =?UTF-8?q?=E2=80=94?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- .DS_Store | Bin 10244 -> 14340 bytes .../tensorflow/README.md" | 12 ++ .../tensorflow/gen_captcha.py" | 46 +++++ .../tensorflow/gen_captcha.pyc" | Bin 0 -> 2174 bytes .../tensorflow/tensorflow_cnn.py" | 195 ++++++++++++++++++ .../tensorflow/text_model.py" | 23 +++ 6 files changed, 276 insertions(+) create mode 100644 "1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.py" create mode 100644 "1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.pyc" create mode 100644 "1.\351\252\214\350\257\201\347\240\201/tensorflow/tensorflow_cnn.py" create mode 100644 "1.\351\252\214\350\257\201\347\240\201/tensorflow/text_model.py" diff --git a/.DS_Store b/.DS_Store index 67b87010c1e5905b320f2495ac2b47b68682f257..a9505c36df58d81e11e2cadf5d0d2b74db2c87f9 100644 GIT binary patch delta 484 zcmZn(Xem%&U|?W$DortDU@!nOIe-{M3-ADmHUV90te)q0bG6E%8Cg;gWO}-~9z)_OSz;NZJrPyXc4l$O^3M~4}o7p*7I2hG8 zzmq+{xLKI#0^?)>CXvkw0^-9jThiA&d4wKOwP{>C`v8M zOf8>W#S|bLonDlg;-8jQoLUkIQj%PhpOX`iUz}NznV&cLIg@XFL?%#6aAs91SZQ!U zVsd6)y0ZNTpnrgXQ`tcS!V+LmhA@D^BfuDvm!Fpklnj7~DLZ&WB{>+h85|jc8Dba; z85$U-Gc0CU&v2089K&OVuMGbgc^SnRl^C@dbr?+<%^4jS9T{1j89f+1Cm&>1+Pq9c zj&<`B5p`y0h@?ugLJT$F2GXv`(Xm;O@TRe@gg(JW_Au14n_sA_y*zE?2Iat zc@$P`PEi(Moj5^j;smM9Y68m`Ctit^WCiL60tKLz3_zO@R&Tuco_R9AiLL=i8e~fY iga$Fe0LWkgF_<6@*8s7BVg{3QOeRiVt9~5CqA&ooHZWrV diff --git "a/1.\351\252\214\350\257\201\347\240\201/tensorflow/README.md" "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/README.md" index 532b6c99..6f041e41 100644 --- "a/1.\351\252\214\350\257\201\347\240\201/tensorflow/README.md" +++ "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/README.md" @@ -4,3 +4,15 @@ 当前进度： 验证码自动生成---完成 + + +直接运行 ： +python TensorFlow_cnn.py + +当前程序设定一个成功率0.5或者0.7 + +我设置为当训练达到0.7 准确性就停止并且保留其模型。。 + +大概训练到 5k多次就能达到这个准确度。 + +还在研究如何提高。 diff --git "a/1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.py" "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.py" new file mode 100644 index 00000000..1dcbbfa5 --- /dev/null +++ "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.py" @@ -0,0 +1,46 @@ +#coding:utf-8 +from captcha.image import ImageCaptcha # pip install captcha +import numpy as np +import matplotlib.pyplot as plt +from PIL import Image +import random,time + +# 验证码中的字符, 就不用汉字了 +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'] +# 验证码一般都无视大小写；验证码长度4个字符 +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 + +# 生成字符对应的验证码 +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') # 写到文件 + + captcha_image = Image.open(captcha) + captcha_image = np.array(captcha_image) + return captcha_text, captcha_image + +if __name__ == '__main__': + # 测试 + while(1): + text, image = gen_captcha_text_and_image() + print 'begin ',time.ctime(),type(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() + print 'end ',time.ctime() diff --git "a/1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.pyc" "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/gen_captcha.pyc" new file mode 100644 index 0000000000000000000000000000000000000000..be12e8b4519fd95d46ea190409e33327a16edabb GIT binary patch literal 2174 zcmb`IOLr7S6vzM79r7Z>BqWd!d?2EZ4>G(T3c?T|350-?AUH=)pU&LMq=)IArn^EC zPFQfl(&JuN?mWkhYd3xZ{RS@Fspq)!1L(bXCW3qVq<(YjR`so_|Gm}m|1J!k{O+gw zb#$Le{=LjC|IGlze+DfeHmDgG>3e8-sQEg^b`P4nQ0vvMk5(Vze$@K^n>v8nz|Pb` z)CNI4pj{w-p%2s#;!$?)1`UAplLi~Wq9vb@eHUH zVfh5t@W4a%tJ(Mpwuo)ubDlGT8XxOAX6zuEqvD;hW3YR%Z7?%yd(qq{&f0$1{b=%$ zXJn){0Xu;vYcdO(gJt6e=q6}hc}qtRVB7bNgqZ9GG_>&ILbDGS%)LYtxmdc*nU4P9 zr^MNY8)D4GVZ%;`sjIJs3mx~a9k-KN$CSsFCzL0ZrkPSSCY~y zSCkFqsxneGl@F9HWvon;ZDp!_sLYhP(ka)JkCf}m4dtftv9hAPqMTMK$*GrZ%CHOm+PXE{`n8t<$2Zcj_g`qsR7t4re!9b&RK1_b^J_I>;}x zRlb|sEHAe<%az1MM-_`{L^dlQpZfFrFaP-A>F?h@Ejyd!?QEshUN1Ln5_H#^N;lmo z2jPA5Zg;$aELdQ7KFEOSHKV3zMoe9La#m?7Ab-_Fw{6yt4OWj0BcD4_PQ~OdODM{9 zx~gcCI zt2)DQ;%~O2L&Q|r3%A_{u!bwdesSbx(7H&SzWh8gFdR;>HF}jMW zE<2ghDK9%8^3UUi6lpR7&QGAL?c>pWm;p234Va==G-KR~Mvof=aTp~*kV^?J+eVZe zirQ>8zWG|7A1^h2`B|R7F6qHn!%l&+4mMNIqtJ<*jte4RX=kyj%q4rMbWq$<6g#EU z(T5$K(ofdnbhCO(x+9D@bZM*YTG2Asm{__}*OV-#E$8I-&fL7Fq*{?zB`+?DJGX^H zF7v+RqJ*t*ZoIvzS>@Yo^iEa&NY;SU5MMX*0LYV`uP#Y fuo>di4x0)8h*>gA-gy5&VaW5@cR)u 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 + +""" +cnn在图像大小是2的倍数时性能最高, 如果你用的图像大小不是2的倍数，可以在图像边缘补无用像素。 +np.pad(image【,((2,3),(2,2)), 'constant', constant_values=(255,)) # 在图像上补2行，下补3行，左补2行，右补2行 +""" + +# 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_LEN = len(char_set) +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 + +#################################################################### + +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 + +# 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) + #out = tf.nn.softmax(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 + 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) + 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50%,保存模型,完成训练 + if acc > 0.5: + saver.save(sess, "crack_capcha.model", global_step=step) + break + step += 1 + +train_crack_captcha_cnn() diff --git "a/1.\351\252\214\350\257\201\347\240\201/tensorflow/text_model.py" "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/text_model.py" new file mode 100644 index 00000000..34133001 --- /dev/null +++ "b/1.\351\252\214\350\257\201\347\240\201/tensorflow/text_model.py" @@ -0,0 +1,23 @@ +def crack_captcha(captcha_image): + output = crack_captcha_cnn() + + saver = tf.train.Saver() + with tf.Session() as sess: + saver.restore(sess, tf.train.latest_checkpoint('.')) + + 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() + vector = np.zeros(MAX_CAPTCHA*CHAR_SET_LEN) + i = 0 + for n in text: + vector[i*CHAR_SET_LEN + n] = 1 + i += 1 + return vec2text(vector) + +text, image = gen_captcha_text_and_image() +image = convert2gray(image) +image = image.flatten() / 255 +predict_text = crack_captcha(image) +print("正确: {} 预测: {}".format(text, predict_text))