Tensorflow框架下实现Mnist数字识别

下面是实现数字识别的程序代码，注意相应的路径：

#!/usr/bin/python  

import input_data #方法一，下载好，并保存到指定的路径下

mnist = input_data.read_data_sets('/home/yuan/TestMnist', one_hot=True)

import tensorflow as tf  

import sys  

from tensorflow.examples.tutorials.mnist import input_data  #方法二，网上下载

  

def weight_variable(shape):  

  initial = tf.truncated_normal(shape, stddev=0.1)  

  return tf.Variable(initial)  

  

def bias_variable(shape):  

  initial = tf.constant(0.1, shape=shape)  

  return tf.Variable(initial)  

  

def conv2d(x, W):  

  return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')  

  

def max_pool_2x2(x):  

  return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')  

  

#mnist = input_data.read_data_sets("MNIST_data/", one_hot=True)  

  

sess = tf.InteractiveSession()  

  

x = tf.placeholder("float", shape=[None, 784])  

y_ = tf.placeholder("float", shape=[None, 10])  

  

W = tf.Variable(tf.zeros([784,10]))  

b = tf.Variable(tf.zeros([10]))  

  

W_conv1 = weight_variable([5, 5, 1, 32])  

b_conv1 = bias_variable([32])  

  

x_image = tf.reshape(x, [-1, 28, 28, 1])  

  

h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)  

h_pool1 = max_pool_2x2(h_conv1)  

  

W_conv2 = weight_variable([5, 5, 32, 64])  

b_conv2 = bias_variable([64])  

  

h_conv2 = tf.nn.relu(conv2d(h_pool1, W_conv2) + b_conv2)  

h_pool2 = max_pool_2x2(h_conv2)  

  

# Now image size is reduced to 7*7  

W_fc1 = weight_variable([7 * 7 * 64, 1024])  

b_fc1 = bias_variable([1024])  

  

h_pool2_flat = tf.reshape(h_pool2, [-1, 7*7*64])  

h_fc1 = tf.nn.relu(tf.matmul(h_pool2_flat, W_fc1) + b_fc1)  

  

keep_prob = tf.placeholder("float")  

h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)  

  

W_fc2 = weight_variable([1024, 10])  

b_fc2 = bias_variable([10])  

y_conv=tf.nn.softmax(tf.matmul(h_fc1_drop, W_fc2) + b_fc2)    

  

cross_entropy = -tf.reduce_sum(y_*tf.log(y_conv))  

train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)  

correct_prediction = tf.equal(tf.argmax(y_conv,1), tf.argmax(y_,1))  

accuracy = tf.reduce_mean(tf.cast(correct_prediction, "float"))  

sess.run(tf.initialize_all_variables())  

  

for i in range(20000):  

  batch = mnist.train.next_batch(50)  

  if i%100 == 0:  

    train_accuracy = accuracy.eval(feed_dict={  

        x:batch[0], y_: batch[1], keep_prob: 1.0})  

    print "step %d, training accuracy %.3f"%(i, train_accuracy)  

  train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})  

  

print "Training finished"  

  

print "test accuracy %.3f" % accuracy.eval(feed_dict={  

    x: mnist.test.images, y_: mnist.test.labels, keep_prob: 1.0})  

下面是程序的部分结果：

Extracting /home/yuan/TestMnist/train-images-idx3-ubyte.gz

Extracting /home/yuan/TestMnist/train-labels-idx1-ubyte.gz

Extracting /home/yuan/TestMnist/t10k-images-idx3-ubyte.gz

Extracting /home/yuan/TestMnist/t10k-labels-idx1-ubyte.gz

step 0, training accuracy 0.100

step 100, training accuracy 0.800

step 200, training accuracy 0.860

step 300, training accuracy 0.800

step 400, training accuracy 0.980

step 500, training accuracy 0.880

step 600, training accuracy 1.000

step 700, training accuracy 0.960

step 800, training accuracy 0.900

step 900, training accuracy 1.000

..

..

..

..

step 19600, training accuracy 1.000

step 19700, training accuracy 1.000

step 19800, training accuracy 1.000

step 19900, training accuracy 1.000

Training finished

test accuracy 0.993

[Finished in 22256.5s]