二、寒假学习计划：Tensorflow之保存、加载模型

想要使用Movidius，但是Movidius只支持两种库，一个是Caffe另一个就是Tensorflow，相对来说的话，个人比较擅长Tensorflow，所以就选择了Tensorflow。

一、保存模型

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

#载入数据
mnist = input_data.read_data_sets('./MNIST_data/',one_hot=True)
#定义数据大小
batch_size = 64
n_batch = mnist.train.num_examples // batch_size
print(batch_size,n_batch)

# 定义占位符，x和y
x = tf.placeholder(tf.float32,shape=[None,784],name='input_x')
y = tf.placeholder(tf.float32,shape=[None,10],name='input_y')
# 定义权重和偏置量
W = tf.Variable(tf.truncated_normal([784,10]),name='W')
b = tf.Variable(tf.zeros([10]) + 0.1,name='bc')
# 模型的输出
prediction = tf.nn.softmax(tf.matmul(x,W) + b,name='output')
# 定义熵代价函数
loss = tf.losses.softmax_cross_entropy(y,prediction,)
# 优化器
train_step = tf.train.AdamOptimizer(0.001).minimize(loss,name='train')
# 初始化
init = tf.global_variables_initializer()
# 计算准确率
correct_prediction = tf.equal(tf.argmax(y,1),tf.argmax(prediction,1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32),name='accuracy')

# 保存模型
saver = tf.train.Saver()
with tf.Session() as sess:
    sess.run(init)
    for epoch in range(10):
        for batch in range(n_batch):
            batch_xs,batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step,feed_dict={x:batch_xs,y:batch_ys})
        acc = sess.run(accuracy,feed_dict={x:mnist.test.images,y:mnist.test.labels})
        print("Iter: {},Accuracy: {}".format(epoch+1,acc))
    # 保存模型
    saver.save(sess,save_path="./model/my_model.ckpt")
    sess.close()

二、加载保存的模型

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data

mnist = input_data.read_data_sets('./MNIST_data',one_hot=True)
batch_size = 64
n_batch = mnist.train.num_examples // batch_size
print(batch_size,n_batch)

with tf.Session() as sess:
    saver = tf.train.import_meta_graph("./model/my_model.ckpt.meta")
    saver.restore(sess,"./model/my_model.ckpt")

    # 根据tensor的名字获取对应的tensor
    output = sess.graph.get_tensor_by_name('output:0')
    accuracy = sess.graph.get_tensor_by_name('accuracy:0')
    train_step = sess.graph.get_operation_by_name('train')
    print(sess.run(accuracy,feed_dict={'input_x:0':mnist.test.images,'input_y:0':mnist.test.labels}))
    for epoch in range(10):
        for batch in range(n_batch):
            batch_xs,batch_ys = mnist.train.next_batch(batch_size)
            sess.run(train_step, feed_dict={'input_x:0': batch_xs, 'input_y:0': batch_ys})
        acc = sess.run(accuracy,feed_dict={'input_x:0':mnist.test.images,'input_y:0':mnist.test.labels})
        print("Iter: {},Accuracy: {}".format(epoch+1,acc))
    # 保存模型
    sess.close()