CNN网站验证码训练后重复预测异常 about anti-anti-spider HOT 8 OPEN

重新写个测试更明显点

def test():
    captcha = Captcha()
    text, image = captcha.gen()
    image = convert2gray(image)
    image = image.flatten() / 255
    predict_text = captcha.crack_captcha(image)
    print("正确: {}  预测: {}".format(text, predict_text))
    time.sleep(1)
    test()

2017-11-02 14:34:54.950657: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use SSE4.2 instructions, but these are available on your machine and could speed up CPU computations.
2017-11-02 14:34:54.950681: W tensorflow/core/platform/cpu_feature_guard.cc:45] The TensorFlow library wasn't compiled to use AVX instructions, but these are available on your machine and could speed up CPU computations.
正确: 51170 预测: 51170
2017-11-02 14:34:56.887373: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_11 not found in checkpoint
2017-11-02 14:34:56.887503: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_10 not found in checkpoint
2017-11-02 14:34:56.889094: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_13 not found in checkpoint
2017-11-02 14:34:56.889150: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_12 not found in checkpoint
2017-11-02 14:34:56.890027: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_14 not found in checkpoint
2017-11-02 14:34:56.891680: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_15 not found in checkpoint
2017-11-02 14:34:56.891723: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_16 not found in checkpoint
2017-11-02 14:34:56.891904: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_19 not found in checkpoint
2017-11-02 14:34:56.894900: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_17 not found in checkpoint
2017-11-02 14:34:56.894900: W tensorflow/core/framework/op_kernel.cc:1152] Not found: Key Variable_18 not found in checkpoint
Traceback (most recent call last):
File "captcha.py", line 252, in
test()
File "captcha.py", line 248, in test
test()
File "captcha.py", line 244, in test
predict_text = captcha.crack_captcha(image) #导入模型识别
File "captcha.py", line 217, in crack_captcha
saver.restore(sess, tf.train.latest_checkpoint('.'))
File "/Library/Python/2.7/site-packages/tensorflow/python/training/saver.py", line 1457, in restore
{self.saver_def.filename_tensor_name: save_path})
File "/Library/Python/2.7/site-packages/tensorflow/python/client/session.py", line 778, in run
run_metadata_ptr)
File "/Library/Python/2.7/site-packages/tensorflow/python/client/session.py", line 982, in _run
feed_dict_string, options, run_metadata)
File "/Library/Python/2.7/site-packages/tensorflow/python/client/session.py", line 1032, in _do_run
target_list, options, run_metadata)
File "/Library/Python/2.7/site-packages/tensorflow/python/client/session.py", line 1052, in _do_call
raise type(e)(node_def, op, message)
tensorflow.python.framework.errors_impl.NotFoundError

from anti-anti-spider.

问题好像是：重试时不要重复定义CNN，即crack_captcha_cnn只执行一次复用即可解决

from anti-anti-spider.

你解决了吗？好像问题不在这

from anti-anti-spider.

我每次都是创建一个新的对象去处理，肯定是执行了一次

from anti-anti-spider.

@1513 crack_captcha_cnn只执行一次，如果在对象里就要处理成单例模式

from anti-anti-spider.

@xiefg 我已经解决了需要重置tf.reset_default_graph()，每次识别前需要进行这个操作，感谢您的回复，另外，要不要设置成单例，多线程下要不要同步处理，不清楚，我试过python多线程跑，好像没有问题

from anti-anti-spider.

@ 1513 ，重置tf.reset_default_graph（）代码会报错，你的重置放哪了？

from anti-anti-spider.

@1513 请问一下tf.reset_default_graph()这个函数放哪里啊，放预测代码中会报错

目前我的解决方案如 @xiefg 所说的，直接使用的单例

from anti-anti-spider.