import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
##查看训练数据集的大小
#print(mnist.train.images.shape)
#print(mnist.train.labels.shape)
def weight_variable(shape):
initial = tf.truncated_normal(shape, stddev=0.1)
return tf.Variable(initial)
###偏置
def bias_variable(shape):
initial = tf.constant(0.1, shape=shape)
return tf.Variable(initial)
##卷积层函数(使用激励函数+卷积层运算)
def conv2d(x, W):
return tf.nn.conv2d(x, W, strides=[1, 1, 1, 1], padding='SAME')
##padding是指对边缘如何处理
###池化层函数(对数据进行抽样,数据池化后,数据量减小)
def max_pool_2x2(x):
return tf.nn.max_pool(x, ksize=[1, 2, 2, 1],
strides=[1, 2, 2, 1], padding='SAME')
if __name__ == '__main__':
# 读入数据
mnist = input_data.read_data_sets("MNIST", one_hot=True)
# x为训练图像的占位符、y_为训练图像标签的占位符
x = tf.placeholder(tf.float32, [None, 784])
y_ = tf.placeholder(tf.float32, [None, 10])
# 数据维度调整 n*784->28*28*1*n
##输入28X28=784维度
# 将单张图片从784维向量重新还原为28x28的矩阵图片
x_image = tf.reshape(x, [-1, 28, 28, 1])
# 第一层卷积层
##卷积大小:5x5
W_conv1 = weight_variable([5, 5, 1, 32])
b_conv1 = bias_variable([32])
h_conv1 = tf.nn.relu(conv2d(x_image, W_conv1) + b_conv1)
##将卷积运算的结果放到一个激励函数里面实现映射
h_pool1 = max_pool_2x2(h_conv1)
# 第二层卷积层
##卷积核:5x5
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)
## [N * 7 * 7 * 64][7 * 7 * 64, 1024] = N * 1024
# 全连接层,输出为1024维的向量
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)
# 使用Dropout,keep_prob是一个占位符,训练时为0.5,测试时为1
keep_prob = tf.placeholder(tf.float32)
###防止过拟合,添加dropout函数
h_fc1_drop = tf.nn.dropout(h_fc1, keep_prob)
# 把1024维的向量转换成10维,对应10个类别
W_fc2 = weight_variable([1024, 10])
b_fc2 = bias_variable([10])
y_conv = tf.matmul(h_fc1_drop, W_fc2) + b_fc2
# 不采用先Softmax再计算交叉熵的方法,而是直接用tf.nn.softmax_cross_entropy_with_logits直接计算
##softmax回归模型:设x为(N,784),w(784,10) b(10,)
##wx+b(N,10)->每一行是一个10维的向量,这些向量来表示最终预测数据的分布概率
###求出loss(构建交叉熵损失)
cross_entropy = tf.reduce_mean(
tf.nn.softmax_cross_entropy_with_logits(labels=y_, logits=y_conv))
# 同样定义train_step
train_step = tf.train.AdamOptimizer(1e-4).minimize(cross_entropy)
# 定义测试的准确率(得到正确的预测结果)
correct_prediction = tf.equal(tf.argmax(y_conv, 1), tf.argmax(y_, 1))
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))
# 创建Session和变量初始化
sess = tf.InteractiveSession()
sess.run(tf.global_variables_initializer())
# 训练20000步
for i in range(20000):
batch = mnist.train.next_batch(50)
##每次读取50步照片
# 每100步报告一次在验证集上的准确度
if i % 100 == 0:
train_accuracy = accuracy.eval(feed_dict={
x: batch[0], y_: batch[1], keep_prob: 1.0})
print("step %d, 训练集准确率%g" % (i, train_accuracy))
train_step.run(feed_dict={x: batch[0], y_: batch[1], keep_prob: 0.5})
# 训练结束后报告在测试集上的准确度
print("测试集准确率 %g" % accuracy.eval(feed_dict={
x: mnist.test.images, y_: mnist.test.labels, keep_prob:1.0}))
手写数字识别代码实现(准确度99%)
最新推荐文章于 2024-07-23 17:35:52 发布