第四讲网络八股拓展

目标：

1. 自制数据集，解决本领域应用
2. 数据增强，扩充数据集
3. 断点续训，存取模型
4. 参数提取，把参数存入文本
5. acc/loss可视化，查看训练效果
6. 应用程序，给图识物

1、自制数据集

# 我们的目标是把图片路径和标签文件输入generateds函数
# 由该函数返回输入特征和标签
def generateds(图片路径, 标签文件)

在自制本地数据集之前，先观察数据集的结构。

由上图，第一列value[0]用于索引到每张图片，value[1]这一列就是每张图片对应的标签。
代码实现：

def generateds(path, txt):

	# 以只读形式打开txt文件
	f = open(txt, 'r')
	
	# 读取文件中所有行
	contents = f.readlines()
	f.close()
	x, y_ = [], []
	# 逐行读出
	for content in contents:

		# 以空格分开，图片名为value[0]，图片标签为value[1]
		value = content.split()
		# 图片路径+图片名拼接出图片的索引路径
		img_path = path + value[0]
		# 读入图片
		img = Image.open(img_path)
		# 图片变为8位宽度的灰度值的np.array格式
		img = np.array(img.convert('L'))
		# 数据归一化
		img = img / 255.
		# 归一化后的数据贴到列表x
		x.append(img)
		# 标签贴到列表y
		y_.append(value[1])
		# 打印状态提示
		print('loading:' + content)
	
	# 变为np.array格式
	x = np.array(x)
	# 变为np.array格式
	y_ = np.array(y_)
	y_ = y_.astype(np.int64)
	return x, y_

import tensorflow as tf
from PIL import Image
import numpy as np
import os

train_path = './mnist_image_label/mnist_train_jpg_60000/'
train_txt = './mnist_image_label/mnist_train_jpg_60000.txt'
# 训练集输入特征存储文件，训练集标签存储文件
x_train_savepath = './mnist_image_label/mnist_x_train.npy'
y_train_savepath = './mnist_image_label/mnist_y_train.npy'

test_path = './mnist_image_label/mnist_test_jpg_10000/'
test_txt = './mnist_image_label/mnist_test_jpg_10000.txt'
x_test_savepath = './mnist_image_label/mnist_x_test.npy'
y_test_savepath = './mnist_image_label/mnist_y_test.npy'


def generateds(path, txt):
    f = open(txt, 'r')  # 以只读形式打开txt文件
    contents = f.readlines()  # 读取文件中所有行
    f.close()  # 关闭txt文件
    x, y_ = [], []  # 建立空列表
    for content in contents:  # 逐行取出
        value = content.split()  # 以空格分开，图片路径为value[0] , 标签为value[1] , 存入列表
        img_path = path + value[0]  # 拼出图片路径和文件名
        img = Image.open(img_path)  # 读入图片
        img = np.array(img.convert('L'))  # 图片变为8位宽灰度值的np.array格式
        img = img / 255.  # 数据归一化 （实现预处理）
        x.append(img)  # 归一化后的数据，贴到列表x
        y_.append(value[1])  # 标签贴到列表y_
        print('loading : ' + content)  # 打印状态提示

    x = np.array(x)  # 变为np.array格式
    y_ = np.array(y_)  # 变为np.array格式
    y_ = y_.astype(np.int64)  # 变为64位整型
    return x, y_  # 返回输入特征x，返回标签y_


# 判断这四个文件是否存在
if os.path.exists(x_train_savepath) and os.path.exists(y_train_savepath) and os.path.exists(
        x_test_savepath) and os.path.exists(y_test_savepath):
    print('-------------Load Datasets-----------------')
    x_train_save = np.load(x_train_savepath)
    y_train = np.load(y_train_savepath)
    x_test_save = np.load(x_test_savepath)
    y_test = np.load(y_test_savepath)
    x_train = np.reshape(x_train_save, (len(x_train_save), 28, 28))
    x_test = np.reshape(x_test_save, (len(x_test_save), 28, 28))
# 不存在则调用generateds函数创建，并分别给x_train, y_train x_test, y_test赋值
else:
    print('-------------Generate Datasets-----------------')
    x_train, y_train = generateds(train_path, train_txt)
    x_test, y_test = generateds(test_path, test_txt)

    print('-------------Save Datasets-----------------')
    x_train_save = np.reshape(x_train, (len(x_train), -1))
    x_test_save = np.reshape(x_test, (len(x_test), -1))
    np.save(x_train_savepath, x_train_save)
    np.save(y_train_savepath, y_train)
    np.save(x_test_savepath, x_test_save)
    np.save(y_test_savepath, y_test)

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(10, activation='softmax')
])

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
              metrics=['sparse_categorical_accuracy'])

model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1)
model.summary()

第一次运行路径中没有npy数据集，创建数据集

2、数据增强

数据增强可以帮助拓展数据集，对图像的增强就是对图像的简单形变，用来因对拍照角度不同引起的图片变形。
TensorFlow2给出了数据增强函数：

image_gen_train = tf.keras.preprocessing.image.ImageDataGenerator(
	rescale = 所有数据集将乘以该数值,
	rotation_range = 随机旋转角度数范围,
	width_shift_range = 随机宽度偏移量,
	height_shift_range = 随机高度偏移量,
	水平翻转：horizontal_flip = 是否随机水平翻转,
	随机缩放：zoom_range = 随机缩放的范围[1-n, 1+n]
)
# 这个fit里面只接受四维数据
image_gen_train.fit(x_train)
# model.fit里同步升级为.flow，将x_train, y_train以batch打包
model.fit(image_gen_train.flow(x_train, y_train, batch_size=32), ...)

例：
# 给数据增加一个维度,从(60000, 28, 28)reshape为(60000, 28, 28, 1)
x_train = x_train.reshape(x_train.shape[0], 28, 28, 1)  
image_gen_train = ImageDataGenerator(
    rescale=1. / 1.,  # 如为图像，分母为255时，可归至0～1
    rotation_range=45,  # 随机45度旋转
    width_shift_range=.15,  # 宽度偏移
    height_shift_range=.15,  # 高度偏移
    horizontal_flip=False,  # 水平翻转
    zoom_range=0.5  # 将图像随机缩放阈量50％
)
image_gen_train.fit(x_train)
model.fit(image_gen_train.flow(x_train, y_train, batch_size=32), 
			epochs=5,
			 validation_data=(x_test, y_test),
			validation_freq=1)

3、断点续训

断点续训可以存取模型。
读取模型：

load_weights(路径文件名)
# 定义出文件存放的路径和文件名.ckpt
checkpoint_save_path = "./checkpoint/mnist.ckpt"
# 在生成ckpt文件时会同步生成索引表，所以可以通过判断是否有索引，就知道是不是已经保存过模型参数
if os.path.exists(checkpoint_save_path + '.index'):
    print('-------------load the model-----------------')
    # 若有索引表，则调用load_weights读取模型参数
    model.load_weights(checkpoint_save_path)

保存模型：

tf.keras.callbacks.ModelCheckpoint(
	filepath=路径文件名,
	# 是否只保留模型参数
    save_weights_only=True/False,
    # 是否只保留最右结果
    save_best_only=True/False)
# 在fit中加入回调选项，返回给history
history = model.fit(callbacks=[cp_callback])

例：

cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_save_path,
                                                 save_weights_only=True,
                                                 save_best_only=True)

history = model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), 
					validation_freq=1,
                    callbacks=[cp_callback])

完整代码：

import tensorflow as tf
import os

mnist = tf.keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(10, activation='softmax')
])

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
              metrics=['sparse_categorical_accuracy'])

checkpoint_save_path = "./checkpoint/mnist.ckpt"
if os.path.exists(checkpoint_save_path + '.index'):
    print('-------------load the model-----------------')
    model.load_weights(checkpoint_save_path)

cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_save_path,
                                                 save_weights_only=True,
                                                 save_best_only=True)

history = model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1,
                    callbacks=[cp_callback])
model.summary()

保存模型之后再次运行会提升准确率。

4、参数提取

把参数存入文本
提取可训练参数：

# 返回模型中可训练的参数
model.trainable_variables

设置print输出格式：

np.set_printoptions(threshold=超过多少省略显示)  # np.inf表示无限大

np.set_printoptions(threshold=np.inf)
print(model.trainable_variables)
# 通过循环写入文件中
file = open('./weights.txt', 'w')
for v in model.trainable_variables:
    file.write(str(v.name) + '\n')
    file.write(str(v.shape) + '\n')
    file.write(str(v.numpy()) + '\n')
file.close()

5、acc和loss可视化

在fit函数执行过程中，已经记录了
训练集loss：loss
测试集loss：val_loss
训练集准确率：sparse_categorical_accuracy
测试集准确率：val_sparse_categorical_accuracy
可以用history.history提取出来

acc = history.history['sparse_categorical_accuracy']
val_acc = history.history['val_sparse_categorical_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']

画图代码

# 将图像分为一行两列，这一段画出第一列
plt.subplot(1, 2, 1)
# 画出acc和val_acc数据，设置出标题，画出图例
plt.plot(acc, label='Training Accuracy')
plt.plot(val_acc, label='Validation Accuracy')
plt.title('Training and Validation Accuracy')
plt.legend()

# 画出第二列
plt.subplot(1, 2, 2)
plt.plot(loss, label='Training Loss')
plt.plot(val_loss, label='Validation Loss')
plt.title('Training and Validation Loss')
plt.legend()

plt.show()

在断点续训的基础上加上画图

import tensorflow as tf
import os
import numpy as np
from matplotlib import pyplot as plt

np.set_printoptions(threshold=np.inf)

mnist = tf.keras.datasets.mnist
(x_train, y_train), (x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(10, activation='softmax')
])

model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=False),
              metrics=['sparse_categorical_accuracy'])

checkpoint_save_path = "./checkpoint/mnist.ckpt"
if os.path.exists(checkpoint_save_path + '.index'):
    print('-------------load the model-----------------')
    model.load_weights(checkpoint_save_path)

cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_save_path,
                                                 save_weights_only=True,
                                                 save_best_only=True)

history = model.fit(x_train, y_train, batch_size=32, epochs=5, validation_data=(x_test, y_test), validation_freq=1,
                    callbacks=[cp_callback])
model.summary()

print(model.trainable_variables)
file = open('./weights.txt', 'w')
for v in model.trainable_variables:
    file.write(str(v.name) + '\n')
    file.write(str(v.shape) + '\n')
    file.write(str(v.numpy()) + '\n')
file.close()

###############################################    show   ###############################################

# 显示训练集和验证集的acc和loss曲线
acc = history.history['sparse_categorical_accuracy']
val_acc = history.history['val_sparse_categorical_accuracy']
loss = history.history['loss']
val_loss = history.history['val_loss']

plt.subplot(1, 2, 1)
plt.plot(acc, label='Training Accuracy')
plt.plot(val_acc, label='Validation Accuracy')
plt.title('Training and Validation Accuracy')
plt.legend()

plt.subplot(1, 2, 2)
plt.plot(loss, label='Training Loss')
plt.plot(val_loss, label='Validation Loss')
plt.title('Training and Validation Loss')
plt.legend()
plt.show()

6、应用程序，给图识物

这里希望输入一张手写照片，神经网络能输别出数字结果

TensorFlow给出了predict函数，它可以根据输入特征给出预测结果。

predict(输入特征, batch_size = 整数)  # 返回前向传播计算结果

第一步：复现模型
model = tf.keras.models.Sequential([
	tf.keras.layers.Flatten(),
	tf.keras.layers.Dense(128, activation='relu'),
	tf.keras.layers.Dense(10, activation='softmax')])

第二步：加载参数
model.load_weights(model_save_path)

第三步：预测结果
result = model.predict(x_predict)

图片识别：

from PIL import Image
import numpy as np
import tensorflow as tf

model_save_path = './checkpoint/mnist.ckpt'

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dense(128, activation='relu'),
    tf.keras.layers.Dense(10, activation='softmax')])

model.load_weights(model_save_path)

preNum = int(input("input the number of test pictures:"))

for i in range(preNum):
    image_path = input("the path of test picture:")
    img = Image.open(image_path)
    img = img.resize((28, 28), Image.ANTIALIAS)
    img_arr = np.array(img.convert('L'))

    # 把图片变为只有黑色和白色的高对比度图片
    for i in range(28):
        for j in range(28):
            if img_arr[i][j] < 200:
                img_arr[i][j] = 255
            else:
                img_arr[i][j] = 0

    img_arr = img_arr / 255.0
    # 增加一个维度
    x_predict = img_arr[tf.newaxis, ...]
    result = model.predict(x_predict)

    pred = tf.argmax(result, axis=1)

    print('\n')
    tf.print(pred)