Tensorflow入门之Mnist数字识别(CNN实现)

# coding=utf-8
"""
卷积神经网络
一般的架构为：image,conv,max pooling,conv,max pooling,fully connected,fully connected,classifier
"""
import tensorflow as tf
import os
os.environ['CUDA_VISIBLE_DEVICES']='0'
from tensorflow.examples.tutorials.mnist import input_data
mnist = input_data.read_data_sets('MNIST_data',one_hot=True)


#计算精度，参数值传入test数据
def compute_accuracy(v_xs,v_ys):
    #sess.run()计算预测结果
    y_prediction = sess.run(prediction,feed_dict={xs:v_xs,keep_prob:1})
    #进行预测与真实的判断
    correct_prediciton = tf.equal(tf.argmax(y_prediction,1),tf.argmax(v_ys,1))
    #通过判断结果计算精确度
    accuracy = tf.reduce_mean(tf.cast(correct_prediciton,tf.float32))
    #sess.run()启动计算
    result = sess.run(accuracy,feed_dict={xs:v_xs,ys:v_ys,keep_prob:1})
    return result

def weight_variable(shape):
    """
    tf.truncated_normal(shape,mean,stddev)
    其中shape为维度，mean为均值，sttdv为标准差
    """
    initial = tf.truncated_normal(shape,stddev=0.1)
    return tf.Variable(initial)

def biases_variable(shape):
    initial = tf.constant(0.1,shape=shape)
    return tf.Variable(initial)

def conv2d(x,W):
    #stride[1,x_movement,y_movement,1]
    #必须有stride[0]=stride[3]=1
    return tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')

def max_pooling_2x2(x):
    #padding='valids',可能舍弃边上的一些元素.padding='SAME'不会舍弃边上的某些元素
    return tf.nn.max_pool(x,ksize=[1,2,2,1],strides=[1,2,2,1],padding='SAME')

xs = tf.placeholder(tf.float32,[None,784])
ys = tf.placeholder(tf.float32,[None,10])
keep_prob = tf.placeholder(tf.float32)
#-1表示导入的图片数，28x28为图片的像素点，1为channels表示灰度图片
x_image = tf.reshape(xs,[-1,28,28,1])

#conv1 layer
#patch5x5 insize1 outsize8(卷积核个数)
W_conv1 = weight_variable([5,5,1,8])
b_conv1 = biases_variable([8])
h_conv1 = tf.nn.relu(conv2d(x_image,W_conv1)+b_conv1)#output size 28x28x8
h_pool1 = max_pooling_2x2(h_conv1)#output size 14x14x8


#conv2 layer
W_conv2 = weight_variable([5,5,8,16])
b_conv2 = biases_variable([16])
h_conv2 = tf.nn.relu(conv2d(h_pool1,W_conv2)+b_conv2)#output size 14x14x16
h_pool2 = max_pooling_2x2(h_conv2)#output size 7x7x16

#func1 layer
W_fc1 = weight_variable([7*7*16,256])
b_fc1 = biases_variable([256])
#[n_samples,7,7,64]->[n_samples,7*7*64]
h_pool1_flat = tf.reshape(h_pool2,[-1,7*7*16])
h_fc1 = tf.nn.relu(tf.matmul(h_pool1_flat,W_fc1)+b_fc1)
h_fc1_drop = tf.nn.dropout(h_fc1,keep_prob)

#func2 layer
W_fc2 = weight_variable([256,10])
b_fc2 = biases_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1_drop,W_fc2)+b_fc2)

#定义交叉熵
cross_entropy = tf.reduce_mean(-tf.reduce_sum(ys*tf.log(prediction),reduction_indices=[1]))
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)

sess = tf.Session()
sess.run(tf.global_variables_initializer())

for i in range(1000):
    batch_xs,batch_ys = mnist.train.next_batch(50)
    sess.run(train_step,feed_dict={xs:batch_xs,ys:batch_ys,keep_prob:0.5})
    if i%50 ==0:
        print(compute_accuracy(mnist.test.images,mnist.test.labels))