def batch_hack_captcha():
output = crack_captcha_cnn()
predict = tf.argmax(tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
saver = tf.train.Saver()
with tf.Session() as sess:
# saver = tf.train.import_meta_graph(save_model + ".meta")
saver.restore(sess, tf.train.latest_checkpoint(model_path))
stime = time.time()
task_cnt = 1000
right_cnt = 0
for i in range(task_cnt):
text, image = gen_captcha_text_and_image()
image = convert2gray(image)
image = image.flatten() / 255
predict_text = hack_function(sess, predict, image)
if text == predict_text:
right_cnt += 1
else:
print("标记: {} 预测: {}".format(text, predict_text))
pass
# print("标记: {} 预测: {}".format(text, predict_text))
print('task:', task_cnt, ' cost time:', (time.time() - stime), 's')
print('right/total-----', right_cnt, '/', task_cnt)
def crack_captcha(captcha_image=random_im):
if captcha_image.shape != (IMAGE_HEIGHT, IMAGE_WIDTH, 3):
captcha_image = captcha_image.resize((IMAGE_HEIGHT, IMAGE_WIDTH, 3))
plt.imshow(captcha_image)
plt.show()
# 定义预测计算图
output = crack_captcha_cnn()
predict = tf.argmax(tf.reshape(output, [-1, MAX_CAPTCHA, CHAR_SET_LEN]), 2)
saver = tf.train.Saver()
with tf.Session() as sess:
saver.restore(sess=sess,
save_path=os.path.join(os.getcwd(), 'checkpoints'))
captcha_image = convert2gray(captcha_image)
captcha_image = captcha_image.flatten() / 255
o, p = sess.run([output, predict],
feed_dict={
X: [captcha_image],
keep_prob: 1
})
v = np.zeros(MAX_CAPTCHA * CHAR_SET_LEN)
for i in p[0]:
v1 = np.argwhere(p[0] == i)
idx = v1 * CHAR_SET_LEN + i
v[idx] = 1
return vec2text(v)
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))
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(max_to_keep=1)
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_)
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)
if acc >= 0.98:
save_path = saver.save(sess,
model_path + "crack_captcha.model",
global_step=step)
print("Save to path: ", save_path)
break
step += 1
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(os.path.dirname(CKPT_PATH)))
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)
def train_crack_captcha_cnn():
output = crack_captcha_cnn()
# loss
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)
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
start = time.time()
paths = LOG_PATH.split('.')
log_path = '{}.{}.{}'.format('.'.join(paths[:-1]), str(int(start)),
paths[-1])
while True:
batch_x, batch_y = get_next_batch(BATCH_SIZE)
_, loss_ = sess.run([optimizer, loss],
feed_dict={
X: batch_x,
Y: batch_y,
keep_prob: 0.75
})
log('step:{} loss:{}'.format(step, loss_), log_path, start)
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.
})
log('step:{} ----accuracy:{}'.format(step, acc), log_path,
start)
if acc > STOP_ACC:
saver.save(sess, CKPT_PATH, global_step=step)
break
step += 1
tf.summary.FileWriter(BOARD_PATH, sess.graph)
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)
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))
print("Configuring TensorBoard and Saver...")
# 配置 Tensorboard，重新训练时，请将tensorboard文件夹删除，不然图会覆盖
tensorboard_dir = 'tensorboard'
if not os.path.exists(tensorboard_dir):
os.makedirs(tensorboard_dir)
tf.summary.scalar("loss", loss)
tf.summary.scalar("accuracy", accuracy)
merged_summary = tf.summary.merge_all()
writer = tf.summary.FileWriter(tensorboard_dir)
# 配置 Saver
saver = tf.train.Saver()
if not os.path.exists(save_dir):
os.makedirs(save_dir)
start_time = time.time()
best_acc_val = 0.0 # 最佳验证集准确率
last_improved = 0 # 记录上一次提升批次
#require_improvement = 1600 # 如果超过xx轮未提升，提前结束训练
with tf.Session() as sess:
writer.add_graph(sess.graph)
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: 0.5
})
# 每100 step计算一次准确率
if step % 100 == 0:
batch_x_test, batch_y_test = get_next_batch(100)
s, acc, losss = sess.run([merged_summary, accuracy, loss],
feed_dict={
X: batch_x_test,
Y: batch_y_test,
keep_prob: 1.
})
# 将训练结果写入tensorboard scalar
writer.add_summary(s, step)
if acc > best_acc_val:
# 保存最好结果
best_acc_val = acc
last_improved = step
saver.save(sess=sess, save_path=save_dir)
improved_str = '*'
else:
improved_str = ''
time_dif = get_time_dif(start_time)
msg = 'Step: {0:>6}, Train Loss: {1:>6.2}, Val Loss: {2:>6.2}, Val Acc: {3:>7.2%}, Time: {4} {5}'
print(
msg.format(step, loss_, losss, acc, time_dif,
improved_str))
"""
if total_batch - last_improved > require_improvement:
# 验证集正确率长期不提升，提前结束训练
print("No optimization for a long time, auto-stopping...")
break # 跳出循环
else:
step += 1
"""
step += 1