tensorflow实战应用大全

迁移学习inception-v3各种图像的识别

1.下载inception-v3模型

2.下载好测试数据集，我存放在images下

# coding: UTF-8
import tensorflow as tf
import os
import numpy as np
import re
from PIL import Image
import matplotlib.pyplot as plt
 
 
class NodeLookup(object):
    def __init__(self):
        label_lookup_path = './inception_model/imagenet_2012_challenge_label_map_proto.pbtxt'
        uid_lookup_path = './inception_model/imagenet_synset_to_human_label_map.txt'
        self.node_lookup = self.load(label_lookup_path, uid_lookup_path)
 
    def load(self, label_lookup_path, uid_lookup_path):
        # 加载分类字符串n***********对应分类名称的文件
        proto_as_ascii_lines = tf.gfile.GFile(uid_lookup_path).readlines()
        uid_to_human = {}
        for line in proto_as_ascii_lines:
            line = line.strip('\n')
            parsed_items = line.split('\t')
            uid = parsed_items[0]  # n15092227
            human_string = parsed_items[1]
            uid_to_human[uid] = human_string
 
        proto_as_ascii = tf.gfile.GFile(label_lookup_path).readlines()
        node_id_to_uid = {}
        for line in proto_as_ascii:
            if line.startswith('  target_class:'):
                target_class = int(line.split(': ')[1])
            if line.startswith('  target_class_string:'):
                target_class_string = line.split(': ')[1]
                node_id_to_uid[target_class] = target_class_string[1: -2]
 
        node_id_to_name = {}
        for key, val in node_id_to_uid.items():
            name = uid_to_human[val]
            node_id_to_name[key] = name
 
        return node_id_to_name
 
    def id_to_string(self, node_id):
        if node_id not in self.node_lookup:
            return ''
        return self.node_lookup[node_id]
 
 
# 创建一个图来存放google调整好的模型
with tf.gfile.FastGFile('./inception_model/classify_image_graph_def.pb', 'rb') as f:
    graph_def = tf.GraphDef()
    graph_def.ParseFromString(f.read())
    tf.import_graph_def(graph_def, name='')
 
with tf.Session() as sess:
    softmax_tensor = sess.graph.get_tensor_by_name('softmax:0')
    # 遍历目录
    for root, dirs, files in os.walk('./images/'):
        for file in files:
            image_data = tf.gfile.FastGFile(os.path.join(root, file), 'rb').read()
            predictions = sess.run(softmax_tensor, {'DecodeJpeg/contents:0': image_data})
            predictions = np.squeeze(predictions)
 
            image_path = os.path.join(root, file)
            print (image_path)
 
            img = Image.open(image_path)
            plt.imshow(img)
            plt.axis('off')
            plt.show()
 
            top_k = predictions.argsort()[-5:][::-1]
            node_lookup = NodeLookup()
            for node_id in top_k:
                human_string = node_lookup.id_to_string(node_id)
                score = predictions[node_id]
                print('%s (score=%.5f)' % (human_string, score))
            print()

结果：识别出来这条狗是samoyede

tensorboard 可视化

定义要可视化的name_scope

import tensorflow as tf
import numpy as np


def add_layer(inputs,in_size,out_size,n_layer,activation_function=None):
    layer_name = "layer%s" % n_layer
    with tf.name_scope(layer_name):
        with tf.name_scope("Weights"):
            Weights = tf.Variable(tf.random_normal([in_size,out_size]),name='W')
            #概率分布的形式
            tf.summary.histogram(layer_name+'/weights',Weights)
        with tf.name_scope("biases"):
            biases = tf.Variable(tf.zeros([1,out_size])+0.1,name='b')
            tf.summary.histogram(layer_name + '/biases', biases)
        with tf.name_scope("Wx_plus_b"):
            Wx_plus_b = tf.add(tf.matmul(inputs,Weights),biases)
        if activation_function is None:
            outputs = Wx_plus_b
        else:
            outputs = activation_function(Wx_plus_b)
        return outputs


x_data = np.linspace(-1,1,300)[:,np.newaxis]
noise = np.random.normal(0,0.05,x_data.shape).astype(np.float32)
y_data = np.square(x_data) - 0.5 + noise

#None表示给多少个sample都可以
with tf.name_scope("input"):
    xs = tf.placeholder(tf.float32,[None,1],name='x_input')
    ys = tf.placeholder(tf.float32,[None,1],name='y_input')

l1 = add_layer(xs,1,10,1,activation_function=tf.nn.relu)
prediction = add_layer(l1,10,1,2,activation_function=None)

with tf.name_scope('loss'):
    loss = tf.reduce_mean(tf.reduce_sum(tf.square(ys - prediction),
                         reduction_indices=[1]))
    tf.summary.scalar("loss",loss)


with tf.name_scope('train'):
    train_step = tf.train.GradientDescentOptimizer(learning_rate=0.1).minimize(loss)


init = tf.global_variables_initializer()

with tf.Session() as sess:
    # 1.2之前 tf.train.SummaryWriter("logs/",sess.graph)
    merged = tf.summary.merge_all()
    writer = tf.summary.FileWriter('logs/',sess.graph)
    sess.run(init)
    for i in range(1000):
        sess.run(train_step,feed_dict={xs:x_data,ys:y_data})
        if i % 50 == 0:
            result = sess.run(merged,feed_dict={xs:x_data,ys:y_data})
            writer.add_summary(result,i)

结果：

CNN手写数字识别

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
#读取mnist数据
mnist = input_data.read_data_sets('./MNIST_data',one_hot=True)
def compute_accuracy(v_xs,v_ys,sess):
    #prediction 变为全剧变量
    global prediction
    y_pre = sess.run(prediction,feed_dict={xs:v_xs,keep_prob:1})
    #预测值每行是10列，tf.argmax(数据，axis），相等为1，不想等为0
    correct_prediction = tf.equal(tf.argmax(y_pre,1),tf.argmax(v_ys,1))
    # 计算平均值，即计算准确率
    accuracy = tf.reduce_mean(tf.cast(correct_prediction,tf.float32))
    # 运行我们的accuracy这一步
    result = sess.run(accuracy,feed_dict={xs:v_xs,ys:v_ys,keep_prob:1})
    return result
#初始化权重
def weight_variable(shape):
    inital = tf.truncated_normal(shape,stddev=0.1)
    return tf.Variable(inital)
#初始化偏置
def bias_variable(shape):
    inital = tf.constant(0.1,shape=shape)
    return tf.Variable(inital)
#卷积层
def conv2d(x,W):
    #strides是步长，四维的列表 [1,x_movement,y_movement,1]
    #x:输入数据 w:权重
    return tf.nn.conv2d(x,W,strides=[1,1,1,1],padding='SAME')
#池化层
def max_pool_2x2(x):
    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])
#dopout率
keep_prob = tf.placeholder(tf.float32)

#将图片reshape，第一位是张数的意思，二三位是图片的长和宽
#，第四位是channel，因为只有灰这一个channel，所以是1
x_image = tf.reshape(xs,[-1,28,28,1])

## conv1 layer ##
#patch是5 * 5 的，输入是1个，输出是32，也就是32个feature map
W_conv1 = weight_variable([5,5,1,32])
b_conv1 = bias_variable([32])
#卷积后经过relu激活再池化
h_conv1= tf.nn.relu(conv2d(x_image,W_conv1) + b_conv1) # 输出28 * 28 * 32
h_pool1 = max_pool_2x2(h_conv1) #输出 14 * 14 * 32

## conv2 layer ##
W_conv2 = weight_variable([5,5,32,64]) # 64个feature map
b_conv2 = bias_variable([64])
h_conv2= tf.nn.relu(conv2d(h_pool1,W_conv2) + b_conv2) # 输出14 * 14 * 64
h_pool2 = max_pool_2x2(h_conv2) #输出 7 * 7 * 64

## func1 layer ##
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)
h_fc1 = tf.nn.dropout(h_fc1,keep_prob)

## func2 layer ##
W_fc2 = weight_variable([1024,10])
b_fc2 = bias_variable([10])
prediction = tf.nn.softmax(tf.matmul(h_fc1,W_fc2)+b_fc2)

#prediction = add_layer(xs,784,10,activation_function=tf.nn.softmax)

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)

init = tf.global_variables_initializer()

with tf.Session() as sess:
    sess.run(init)
    for i in range(1000):
        batch_xs, batch_ys = mnist.train.next_batch(100)
        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,sess))

tensorflow实现简单的线性回归

import tensorflow as tf
import numpy as np
import matplotlib.pyplot as plt

#使用numpy生成200个随机点
x_data=np.linspace(-0.5,0.5,200)[:,np.newaxis]
noise=np.random.normal(0,0.02,x_data.shape)
y_data=np.square(x_data)+noise

#定义两个placeholder存放输入数据
x=tf.placeholder(tf.float32,[None,1])
y=tf.placeholder(tf.float32,[None,1])

#定义神经网络中间层
Weights_L1=tf.Variable(tf.random_normal([1,10]))
biases_L1=tf.Variable(tf.zeros([1,10]))    #加入偏置项
Wx_plus_b_L1=tf.matmul(x,Weights_L1)+biases_L1
L1=tf.nn.tanh(Wx_plus_b_L1)   #加入激活函数

#定义神经网络输出层
Weights_L2=tf.Variable(tf.random_normal([10,1]))
biases_L2=tf.Variable(tf.zeros([1,1]))  #加入偏置项
Wx_plus_b_L2=tf.matmul(L1,Weights_L2)+biases_L2
prediction=tf.nn.tanh(Wx_plus_b_L2)   #加入激活函数

#定义损失函数（均方差函数）
loss=tf.reduce_mean(tf.square(y-prediction))
#定义反向传播算法（使用梯度下降算法训练）
train_step=tf.train.GradientDescentOptimizer(0.1).minimize(loss)

with tf.Session() as sess:
    #变量初始化
    sess.run(tf.global_variables_initializer())
    #训练2000次
    for i in range(2000):
        sess.run(train_step,feed_dict={x:x_data,y:y_data})
    #获得预测值
    prediction_value=sess.run(prediction,feed_dict={x:x_data})
    #画图
    plt.figure()
    plt.scatter(x_data,y_data)   #散点是真实值
    plt.plot(x_data,prediction_value,'r-',lw=5)   #曲线是预测值
    plt.show()

手写数字生成对抗网络实战

import tensorflow as tf
from tensorflow.examples.tutorials.mnist import input_data
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
import os
#该函数将给出权重初始化的方法
def variable_init(size):
    in_dim = size[0]
    #计算随机生成变量所服从的正态分布标准差
    w_stddev = 1. / tf.sqrt(in_dim / 2.)
    return tf.random_normal(shape=size, stddev=w_stddev)

#定义一个可以生成m*n阶随机矩阵的函数，该矩阵的元素服从均匀分布，随机生成的z就为生成器的输入
def sample_Z(m, n):
    return np.random.uniform(-1., 1., size=[m, n])

#该函数用于输出生成图片
def plot(samples):
    fig = plt.figure(figsize=(4, 4))
    gs = gridspec.GridSpec(4, 4)
    gs.update(wspace=0.05, hspace=0.05)

    for i, sample in enumerate(samples):
        ax = plt.subplot(gs[i])
        plt.axis('off')
        ax.set_xticklabels([])
        ax.set_yticklabels([])
        ax.set_aspect('equal')
        plt.imshow(sample.reshape(28, 28), cmap='Greys_r')

    return fig
#定义输入矩阵的占位符，输入层单元为784，None代表批量大小的占位，X代表输入的真实图片。占位符的数值类型为32位浮点型
X = tf.placeholder(tf.float32, shape=[None, 784])

#定义判别器的权重矩阵和偏置项向量，由此可知判别网络为三层全连接网络
D_W1 = tf.Variable(variable_init([784, 128]))
D_b1 = tf.Variable(tf.zeros(shape=[128]))

D_W2 = tf.Variable(variable_init([128, 1]))
D_b2 = tf.Variable(tf.zeros(shape=[1]))

theta_D = [D_W1, D_W2, D_b1, D_b2]

#定义生成器的输入噪声为100维度的向量组，None根据批量大小确定
Z = tf.placeholder(tf.float32, shape=[None, 100])

#定义生成器的权重与偏置项。输入层为100个神经元且接受随机噪声，
#输出层为784个神经元，并输出手写字体图片。生成网络根据原论文为三层全连接网络
G_W1 = tf.Variable(variable_init([100, 128]))
G_b1 = tf.Variable(tf.zeros(shape=[128]))

G_W2 = tf.Variable(variable_init([128, 784]))
G_b2 = tf.Variable(tf.zeros(shape=[784]))

theta_G = [G_W1, G_W2, G_b1, G_b2]
# 定义生成器
def generator(z):
    # 第一层先计算 y=z*G_W1+G-b1,然后投入激活函数计算G_h1=ReLU（y）,G_h1 为第二次层神经网络的输出激活值
    G_h1 = tf.nn.relu(tf.matmul(z, G_W1) + G_b1)

    # 以下两个语句计算第二层传播到第三层的激活结果，第三层的激活结果是含有784个元素的向量，该向量转化28×28就可以表示图像
    G_log_prob = tf.matmul(G_h1, G_W2) + G_b2
    G_prob = tf.nn.sigmoid(G_log_prob)
    return G_prob
# 定义判别器
def discriminator(x):
    # 计算D_h1=ReLU（x*D_W1+D_b1）,该层的输入为含784个元素的向量
    D_h1 = tf.nn.relu(tf.matmul(x, D_W1) + D_b1)

    # 计算第三层的输出结果。因为使用的是Sigmoid函数，则该输出结果是一个取值为[0,1]间的标量（见上述权重定义）
    # 即判别输入的图像到底是真（=1）还是假（=0）
    D_logit = tf.matmul(D_h1, D_W2) + D_b2
    D_prob = tf.nn.sigmoid(D_logit)

    # 返回判别为真的概率和第三层的输入值，输出D_logit是为了将其输入tf.nn.sigmoid_cross_entropy_with_logits()以构建损失函数
    return D_prob, D_logit
#输入随机噪声z而输出生成样本
G_sample = generator(Z)

#分别输入真实图片和生成的图片，并投入判别器以判断真伪
D_real, D_logit_real = discriminator(X)
D_fake, D_logit_fake = discriminator(G_sample)

#以下为原论文的判别器损失和生成器损失，但本实现并没有使用该损失函数
# D_loss = -tf.reduce_mean(tf.log(D_real) + tf.log(1. - D_fake))
# G_loss = -tf.reduce_mean(tf.log(D_fake))

# 我们使用交叉熵作为判别器和生成器的损失函数，因为sigmoid_cross_entropy_with_logits内部会对预测输入执行Sigmoid函数，
#所以我们取判别器最后一层未投入激活函数的值，即D_h1*D_W2+D_b2。
#tf.ones_like(D_logit_real)创建维度和D_logit_real相等的全是1的标注，真实图片。
D_loss_real = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_real, labels=tf.ones_like(D_logit_real)))
D_loss_fake = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_fake, labels=tf.zeros_like(D_logit_fake)))

#损失函数为两部分，即E[log(D(x))]+E[log(1-D(G(z)))]，将真的判别为假和将假的判别为真
D_loss = D_loss_real + D_loss_fake

#同样使用交叉熵构建生成器损失函数
G_loss = tf.reduce_mean(tf.nn.sigmoid_cross_entropy_with_logits(logits=D_logit_fake, labels=tf.ones_like(D_logit_fake)))
#定义判别器和生成器的优化方法为Adam算法，关键字var_list表明最小化损失函数所更新的权重矩阵
D_solver = tf.train.AdamOptimizer().minimize(D_loss, var_list=theta_D)
G_solver = tf.train.AdamOptimizer().minimize(G_loss, var_list=theta_G)
#选择训练的批量大小和随机生成噪声的维度
mb_size = 128
Z_dim = 100

#读取数据集MNIST，并放在当前目录data文件夹下MNIST文件夹中，如果该地址没有数据，则下载数据至该文件夹
mnist = input_data.read_data_sets("./MNIST_data", one_hot=True)

# 打开一个会话运行计算图
sess = tf.Session()

# 初始化所有定义的变量
sess.run(tf.global_variables_initializer())

# 如果当前目录下不存在out文件夹，则创建该文件夹
if not os.path.exists('out/'):
    os.makedirs('out/')

# 初始化，并开始迭代训练，100W次
i = 0
for it in range(20000):

    # 每2000次输出一张生成器生成的图片
    if it % 2000 == 0:
        samples = sess.run(G_sample, feed_dict={Z: sample_Z(16, Z_dim)})

        fig = plot(samples)
        plt.savefig('out/{}.png'.format(str(i).zfill(3)), bbox_inches='tight')
        i += 1
        plt.close(fig)

    # next_batch抽取下一个批量的图片，该方法返回一个矩阵，即shape=[mb_size，784]，每一行是一张图片，共批量大小行
    X_mb, _ = mnist.train.next_batch(mb_size)

    # 投入数据并根据优化方法迭代一次，计算损失后返回损失值
    _, D_loss_curr = sess.run([D_solver, D_loss], feed_dict={X: X_mb, Z: sample_Z(mb_size, Z_dim)})
    _, G_loss_curr = sess.run([G_solver, G_loss], feed_dict={Z: sample_Z(mb_size, Z_dim)})

    # 每迭代2000次输出迭代数、生成器损失和判别器损失
    if it % 2000 == 0:
        print('Iter: {}'.format(it))
        print('D loss: {:.4}'.format(D_loss_curr))
        print('G_loss: {:.4}'.format(G_loss_curr))
        print()

生成的图像：