Question

我有什么：

windows 10 x64，anaconda3（python 3.6.4），apache2.4，mod_wsgi，flask webapp

我想要什么：

使用mod_wsgi在flask上部署flask web，flask web使用scikit-learn和tensorflow / keras模型

我尝试了什么：

打开命令提示符并激活tf15 ， python runserver.py

from myFlaskApp import app
if __name__ == '__main__':
    app.run('localhost', 8081)

一切正常

打开命令提示符并激活tf15 ， mod_wsgi-express module-config

我在apache httpd.conf中添加了这个

LoadFile "c:/anaconda3/envs/tf15/python36.dll"
LoadModule wsgi_module "c:/anaconda3/envs/tf15/lib/site-packages/mod_wsgi/server/mod_wsgi.cp36-win_amd64.pyd"
WSGIPythonHome "c:/anaconda3/envs/tf15"

一切正常，主页显示， sklearn模型工作但不是TF keras.models.load_model（“path_to_model_file.h5”）抛出异常：会话图是空的

[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     self.model = K.models.load_model(modelFilePath)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\models.py", line 246, in load_model\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     topology.load_weights_from_hdf5_group(f['model_weights'], model.layers)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\engine\\topology.py", line 3166, in load_weights_from_hdf5_group\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     K.batch_set_value(weight_value_tuples)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py", line 2370, in batch_set_value\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     get_session().run(assign_ops, feed_dict=feed_dict)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\tensorflow\\python\\client\\session.py", line 895, in run\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     run_metadata_ptr)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\tensorflow\\python\\client\\session.py", line 1055, in _run\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     raise RuntimeError('The Session graph is empty.  Add operations to the '\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579] RuntimeError: The Session graph is empty.  Add operations to the graph before calling run().
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579] \r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     self.model = K.models.load_model(modelFilePath)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\models.py", line 246, in load_model\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     topology.load_weights_from_hdf5_group(f['model_weights'], model.layers)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\engine\\topology.py", line 3166, in load_weights_from_hdf5_group\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     K.batch_set_value(weight_value_tuples)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\keras\\backend\\tensorflow_backend.py", line 2370, in batch_set_value\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     get_session().run(assign_ops, feed_dict=feed_dict)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\tensorflow\\python\\client\\session.py", line 895, in run\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     run_metadata_ptr)\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]   File "c:\\anaconda3\\envs\\tf15\\lib\\site-packages\\tensorflow\\python\\client\\session.py", line 1055, in _run\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579]     raise RuntimeError('The Session graph is empty.  Add operations to the '\r
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579] RuntimeError: The Session graph is empty.  Add operations to the graph before calling run().
[wsgi:error] [pid 7108:tid 1404] [client ::1:55579] \r

我很确定它不是空的，并且模型文件存在，因为它在烧瓶托管时起作用

tensorflow / keras不会在apache2.4上加载模型文件

0 个答案: