TensorFlow北大公开课学习笔记7.1-lenet5代码

1998年Y.LeCun等提出的GradientBased Learning Applied to Document Recognitionhttps://ieeexplore.ieee.org/document/726791

                                     

 今天学习针对28*28*1的手写数字进行复现lenet网络。比0.9808高呀。lenet是CNN的鼻祖呀 卷积神经网络。

这个函数img.resize((width, height),Image.ANTIALIAS)
第二个参数：
Image.NEAREST ：低质量
Image.BILINEAR：双线性
Image.BICUBIC ：三次样条插值
Image.ANTIALIAS：高质量

#coding:utf-8
import tensorflow as tf
#每张图片分辨率为28*28
IMAGE_SIZE = 28
#Mnist数据集为灰度图，故输入图片通道数NUM_CHANNELS取值为1
NUM_CHANNELS = 1
#第一层卷积核大小为5
CONV1_SIZE = 5
#卷积核个数为32
CONV1_KERNEL_NUM = 32
#第二层卷积核大小为5
CONV2_SIZE = 5
#卷积核个数为64
CONV2_KERNEL_NUM = 64
#全连接层第一层为 512 个神经元
FC_SIZE = 512
#全连接层第二层为 10 个神经元
OUTPUT_NODE = 10

#权重w计算
def get_weight(shape, regularizer):
	w = tf.Variable(tf.truncated_normal(shape,stddev=0.1))
	if regularizer != None: tf.add_to_collection('losses', tf.contrib.layers.l2_regularizer(regularizer)(w)) 
	return w

#偏置b计算
def get_bias(shape): 
	b = tf.Variable(tf.zeros(shape))  
	return b

#卷积层计算
def conv2d(x,w):  
	return tf.nn.conv2d(x, w, strides=[1, 1, 1, 1], padding='SAME')

#最大池化层计算
def max_pool_2x2(x):  
	return tf.nn.max_pool(x, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME') 

def forward(x, train, regularizer):

    # 实现第一层卷积
    conv1_w = get_weight([CONV1_SIZE, CONV1_SIZE, NUM_CHANNELS, CONV1_KERNEL_NUM], regularizer) 
    conv1_b = get_bias([CONV1_KERNEL_NUM]) 
    conv1 = conv2d(x, conv1_w)
    # 非线性激活
    relu1 = tf.nn.relu(tf.nn.bias_add(conv1, conv1_b))
    # 最大池化
    pool1 = max_pool_2x2(relu1)

    # 实现第二层卷积
    conv2_w = get_weight([CONV2_SIZE, CONV2_SIZE, CONV1_KERNEL_NUM, CONV2_KERNEL_NUM],regularizer) 
    conv2_b = get_bias([CONV2_KERNEL_NUM])
    conv2 = conv2d(pool1, conv2_w) 
    relu2 = tf.nn.relu(tf.nn.bias_add(conv2, conv2_b))
    pool2 = max_pool_2x2(relu2)

    # 获取一个张量的维度
    pool_shape = pool2.get_shape().as_list()
    # pool_shape[1] 为长 pool_shape[2] 为宽 pool_shape[3]为高
    nodes = pool_shape[1] * pool_shape[2] * pool_shape[3]
    # 得到矩阵被拉长后的长度，pool_shape[0]为batch值
    reshaped = tf.reshape(pool2, [pool_shape[0], nodes])

    # 实现第三层全连接层
    fc1_w = get_weight([nodes, FC_SIZE], regularizer) 
    fc1_b = get_bias([FC_SIZE]) 
    fc1 = tf.nn.relu(tf.matmul(reshaped, fc1_w) + fc1_b)
    # 如果是训练阶段，则对该层输出使用dropout
    if train: fc1 = tf.nn.dropout(fc1, 0.5)

    # 实现第四层全连接层
    fc2_w = get_weight([FC_SIZE, OUTPUT_NODE], regularizer)
    fc2_b = get_bias([OUTPUT_NODE])
    y = tf.matmul(fc1, fc2_w) + fc2_b
    return y 

#coding:utf-8
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import mnist_lenet5_forward
import os
import numpy as np

#batch的数量
BATCH_SIZE = 100
#初始学习率
LEARNING_RATE_BASE =  0.005
#学习率衰减率
LEARNING_RATE_DECAY = 0.99
#正则化
REGULARIZER = 0.0001
#最大迭代次数
STEPS = 50000
#滑动平均衰减率
MOVING_AVERAGE_DECAY = 0.99
#模型保存路径
MODEL_SAVE_PATH="./model/" #model会自动生成不需要自己加
#模型名称
MODEL_NAME="mnist_model" 

def backward(mnist):
    # 卷积层输入为四阶张量
    # 第一阶表示每轮喂入的图片数量，第二阶和第三阶分别表示图片的行分辨率和列分辨率，第四阶表示通道数
    x = tf.placeholder(tf.float32,[
	BATCH_SIZE,
	mnist_lenet5_forward.IMAGE_SIZE,
	mnist_lenet5_forward.IMAGE_SIZE,
	mnist_lenet5_forward.NUM_CHANNELS]) 
    y_ = tf.placeholder(tf.float32, [None, mnist_lenet5_forward.OUTPUT_NODE])
    # 前向传播过程
    y = mnist_lenet5_forward.forward(x,True, REGULARIZER)
    # 声明一个全局计数器
    global_step = tf.Variable(0, trainable=False)
    # 对网络最后一层的输出y做softmax，求取输出属于某一类的概率
    ce = tf.nn.sparse_softmax_cross_entropy_with_logits(logits=y, labels=tf.argmax(y_, 1))
    # 向量求均值
    cem = tf.reduce_mean(ce)
    # 正则化的损失值
    loss = cem + tf.add_n(tf.get_collection('losses'))

    # 指数衰减学习率
    learning_rate = tf.train.exponential_decay( 
        LEARNING_RATE_BASE,
        global_step,
        mnist.train.num_examples / BATCH_SIZE, 
		LEARNING_RATE_DECAY,
        staircase=True)

    # 梯度下降算法的优化器
    # train_step = tf.train.GradientDescentOptimizer(learning_rate).minimize(loss, global_step=global_step)
    train_step = tf.train.MomentumOptimizer(learning_rate, 0.9).minimize(loss, global_step=global_step)
    # 采用滑动平均的方法更新参数
    ema = tf.train.ExponentialMovingAverage(MOVING_AVERAGE_DECAY, global_step)
    ema_op = ema.apply(tf.trainable_variables())
    # 将train_step和ema_op两个训练操作绑定到train_op上
    with tf.control_dependencies([train_step, ema_op]): 
        train_op = tf.no_op(name='train')

    # 实例化一个保存和恢复变量的saver
    saver = tf.train.Saver()
    # 创建一个会话
    with tf.Session() as sess: 
        init_op = tf.global_variables_initializer() 
        sess.run(init_op)
        # 通过 checkpoint 文件定位到最新保存的模型，若文件存在，则加载最新的模型
        ckpt = tf.train.get_checkpoint_state(MODEL_SAVE_PATH) 
        if ckpt and ckpt.model_checkpoint_path:
        	saver.restore(sess, ckpt.model_checkpoint_path) 

        for i in range(STEPS):
            # 读取一个batch数据，将输入数据xs转成与网络输入相同形状的矩阵
            xs, ys = mnist.train.next_batch(BATCH_SIZE) 
            reshaped_xs = np.reshape(xs,(  
		    BATCH_SIZE,
        	mnist_lenet5_forward.IMAGE_SIZE,
        	mnist_lenet5_forward.IMAGE_SIZE,
        	mnist_lenet5_forward.NUM_CHANNELS))
            # 读取一个batch数据，将输入数据xs转成与网络输入相同形状的矩阵
            _, loss_value, step = sess.run([train_op, loss, global_step], feed_dict={x: reshaped_xs, y_: ys}) 
            if i % 100 == 0: 
                print("After %d training step(s), loss on training batch is %g." % (step, loss_value))
                saver.save(sess, os.path.join(MODEL_SAVE_PATH, MODEL_NAME), global_step=global_step)

def main():
    mnist = input_data.read_data_sets("./data/", one_hot=True) 
    backward(mnist)

if __name__ == '__main__':
    main()

 

#coding:utf-8
import time
import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import mnist_lenet5_forward
import mnist_lenet5_backward
import numpy as np

TEST_INTERVAL_SECS = 5

#创建一个默认图，在该图中执行以下操作
def Test(mnist):
    with tf.Graph().as_default() as g: 
        x = tf.placeholder(tf.float32,[
            mnist.test.num_examples,
            mnist_lenet5_forward.IMAGE_SIZE,
            mnist_lenet5_forward.IMAGE_SIZE,
            mnist_lenet5_forward.NUM_CHANNELS]) 
        y_ = tf.placeholder(tf.float32, [None, mnist_lenet5_forward.OUTPUT_NODE])
        # 训练好的网络，故不使用 dropout
        y = mnist_lenet5_forward.forward(x,False,None)

        ema = tf.train.ExponentialMovingAverage(mnist_lenet5_backward.MOVING_AVERAGE_DECAY)
        ema_restore = ema.variables_to_restore()
        saver = tf.train.Saver(ema_restore)

        # 判断预测值和实际值是否相同
        correct_prediction = tf.equal(tf.argmax(y, 1), tf.argmax(y_, 1))
        ## 求平均得到准确率
        accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float32)) 

        while True:
            with tf.Session() as sess:
                ckpt = tf.train.get_checkpoint_state(mnist_lenet5_backward.MODEL_SAVE_PATH)
                if ckpt and ckpt.model_checkpoint_path:
                    saver.restore(sess, ckpt.model_checkpoint_path)

                    # 根据读入的模型名字切分出该模型是属于迭代了多少次保存的
                    global_step = ckpt.model_checkpoint_path.split('/')[-1].split('-')[-1] 
                    reshaped_x = np.reshape(mnist.test.images,(
                    mnist.test.num_examples,
        	        mnist_lenet5_forward.IMAGE_SIZE,
        	        mnist_lenet5_forward.IMAGE_SIZE,
        	        mnist_lenet5_forward.NUM_CHANNELS))

                    # 利用多线程提高图片和标签的批获取效率
                    coord = tf.train.Coordinator()  # 3
                    threads = tf.train.start_queue_runners(sess=sess, coord=coord)  # 4
                    accuracy_score = sess.run(accuracy, feed_dict={x:reshaped_x,y_:mnist.test.labels}) 
                    print("After %s training step(s), test accuracy = %g" % (global_step, accuracy_score))
                    coord.request_stop()  # 6
                    coord.join(threads)  # 7
                    # 关闭线程协调器
                else:
                    print('No checkpoint file found')
                    return
            time.sleep(TEST_INTERVAL_SECS) 

def main():
    mnist = input_data.read_data_sets("./data/", one_hot=True)
    Test(mnist)

if __name__ == '__main__':
    main()