#!/usr/bin/env python3

#-*- coding:utf-8 -*-

############################

#File Name: save_restore.py

#Brief:

#Author: frank

#Mail: frank0903@aliyun.com

#Created Time:2018-06-26 20:30:09

############################

import tensorflow as tf

from tensorflow.python.framework import graph_util

#TF提供了convert_variables_to_constants函数,通过这个函数可以将计算图中的变量及其取值通过常量的方式保存.

v1 = tf.Variable(tf.constant(1.0, shape=[1]), name="v1")

v2 = tf.Variable(tf.constant(2.0, shape=[1]), name="v2")

result = v1 + v2

init_op = tf.global_variables_initializer()

with tf.Session() as sess:

    sess.run(init_op)

    #导出当前计算图的GraphDef部分,只需要这一步就可以完成从输入层到输出层的计算过程

    graph_def = tf.get_default_graph().as_graph_def()

    #将图中的变量及其取值转化为常量,同时将图中不必要的节点去掉.本程序只关心加法运算,所以这里只保存['add']节点,其他和该计算无关的节点就不保存了

    output_graph_def = graph_util.convert_variables_to_constants(sess, graph_def, ['add'])

    with tf.gfile.GFile("./combine/combined_model.pb", "wb") as f:

    ¦   f.write(output_graph_def.SerializeToString())

 #!/usr/bin/env python3

 #-*- coding:utf-8 -*-

 ############################

 #File Name: save_restore2.py

 #Brief:

 #Author: frank

 #Mail: frank0903@aliyun.com

 #Created Time:2018-06-26 20:30:09

 ############################                                                                          

 import tensorflow as tf

 from tensorflow.python.platform import gfile                                                          

 with tf.Session() as sess:

     model_filename = "combine/combined_model.pb"

     #读取保存的模型文件,并将文件解析成对应的GraphDef Protocol Buffer.

     with gfile.FastGFile(model_filename, "rb") as f:

     ¦   graph_def = tf.GraphDef()

     ¦   graph_def.ParseFromString(f.read())                                                           

     #将graph_def中保存的图加载到当前图中.return_elements=["add:0"]给出了返回的张量的名称.在保存的时候给出的是计算节点的名称,所以为"add".在加载的时候给出的是张量的名称,所以是add:0.

     result = tf.import_graph_def(graph_def, return_elements=["add:0"])

     print(sess.run(result))