摘要

本篇文章采用CNN实现图像的分类，图像选取了猫狗大战数据集的1万张图像（猫狗各5千）。模型采用自定义的CNN网络，版本是TensorFlow 2.0以上的版本。通过本篇文章，你可以学到图像分类常用的手段，包括：

1、图像增强

2、训练集和验证集切分

3、使用ModelCheckpoint保存最优模型

4、使用ReduceLROnPlateau调整学习率。

5、打印loss结果生成jpg图片。

网络详解

训练部分

1、导入依赖

 import os
 import numpy as np
 from tensorflow import keras
 from tensorflow.keras.optimizers import Adam
 from tensorflow.keras.models import Sequential
 from tensorflow.keras.layers import Dense, Dropout,BatchNormalization,Flatten
 from tensorflow.keras.layers import Conv2D, MaxPooling2D,GlobalAveragePooling2D
 import cv2
 from tensorflow.keras.preprocessing.image import img_to_array
 from sklearn.model_selection import train_test_split
 from tensorflow.python.keras import Input
 from tensorflow.python.keras.callbacks import ModelCheckpoint, ReduceLROnPlateau
 from tensorflow.python.keras.layers import PReLU, Activation
 from tensorflow.python.keras.models import Model

2、设置全局参数

norm_size=100#输入到网络的图像尺寸，单位是像素。

datapath='train'#图片的根目录

EPOCHS =100#训练的epoch个数

INIT_LR = 1e-3#初始学习率

labelList=[]#标签

dicClass={'cat':0,'dog':1}#类别

labelnum=2#类别个数

batch_size = 4

3、加载数据

def loadImageData():

    imageList = []

    listImage=os.listdir(datapath)#获取所有的图像

    for img in listImage:#遍历图像

        labelName=dicClass[img.split('.')[0]]#获取label对应的数字

        print(labelName)

        labelList.append(labelName)

        dataImgPath=os.path.join(datapath,img)

        print(dataImgPath)

        image = cv2.imdecode(np.fromfile(dataImgPath, dtype=np.uint8), -1)

        # load the image, pre-process it, and store it in the data list

        image = cv2.resize(image, (norm_size, norm_size), interpolation=cv2.INTER_LANCZOS4)

        image = img_to_array(image)

        imageList.append(image)

    imageList = np.array(imageList, dtype="int") / 255.0#归一化图像

    return imageList

print("开始加载数据")

imageArr=loadImageData()

labelList = np.array(labelList)

print("加载数据完成")

print(labelList)

4、定义模型

def bn_prelu(x):

    x = BatchNormalization(epsilon=1e-5)(x)

    x = PReLU()(x)

    return x

def build_model(out_dims, input_shape=(norm_size, norm_size, 3)):

    inputs_dim = Input(input_shape)

    x = Conv2D(32, (3, 3), strides=(2, 2), padding='same')(inputs_dim)

    x = bn_prelu(x)

    x = Conv2D(32, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = MaxPooling2D(pool_size=(2, 2))(x)

    x = Conv2D(64, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = Conv2D(64, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = MaxPooling2D(pool_size=(2, 2))(x)

    x = Conv2D(128, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = Conv2D(128, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = MaxPooling2D(pool_size=(2, 2))(x)

    x = Conv2D(256, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = Conv2D(256, (3, 3), strides=(1, 1), padding='same')(x)

    x = bn_prelu(x)

    x = GlobalAveragePooling2D()(x)

    dp_1 = Dropout(0.5)(x)

    fc2 = Dense(out_dims)(dp_1)

    fc2 = Activation('softmax')(fc2) #此处注意，为sigmoid函数

    model = Model(inputs=inputs_dim, outputs=fc2)

    return model

model=build_model(labelnum)#生成模型

optimizer = Adam(lr=INIT_LR)#加入优化器，设置优化器的学习率。

model.compile(optimizer =optimizer, loss='sparse_categorical_crossentropy', metrics=['accuracy'])

5、切割训练集和验证集

trainX,valX,trainY,valY = train_test_split(imageArr,labelList, test_size=0.3, random_state=42)

6、数据增强

from tensorflow.keras.preprocessing.image import ImageDataGenerator

train_datagen = ImageDataGenerator(featurewise_center=True,

    featurewise_std_normalization=True,

    rotation_range=20,

    width_shift_range=0.2,

    height_shift_range=0.2,

    horizontal_flip=True)

val_datagen = ImageDataGenerator()     #验证集不做图片增强

train_generator = train_datagen.flow(trainX,trainY,batch_size=batch_size,shuffle=True)

val_generator = val_datagen.flow(valX,valY,batch_size=batch_size,shuffle=True)

7、设置callback函数

checkpointer = ModelCheckpoint(filepath='weights_best_simple_model.hdf5',

                            monitor='val_accuracy',verbose=1, save_best_only=True, mode='max')



reduce = ReduceLROnPlateau(monitor='val_accuracy',patience=10,

                                            verbose=1,

                                            factor=0.5,

                                            min_lr=1e-6)

8、训练并保存模型

history = model.fit_generator(train_generator,

       steps_per_epoch=trainX.shape[0]/batch_size,

       validation_data = val_generator,

       epochs=EPOCHS,

       validation_steps=valX.shape[0]/batch_size,

       callbacks=[checkpointer,reduce],

       verbose=1,shuffle=True)

model.save('my_model_.h5')

9、保存训练历史数据

import os

loss_trend_graph_path = r"WW_loss.jpg"

acc_trend_graph_path = r"WW_acc.jpg"

import matplotlib.pyplot as plt

print("Now,we start drawing the loss and acc trends graph...")

# summarize history for accuracy

fig = plt.figure(1)

plt.plot(history.history["accuracy"])

plt.plot(history.history["val_accuracy"])

plt.title("Model accuracy")

plt.ylabel("accuracy")

plt.xlabel("epoch")

plt.legend(["train", "test"], loc="upper left")

plt.savefig(acc_trend_graph_path)

plt.close(1)

# summarize history for loss

fig = plt.figure(2)

plt.plot(history.history["loss"])

plt.plot(history.history["val_loss"])

plt.title("Model loss")

plt.ylabel("loss")

plt.xlabel("epoch")

plt.legend(["train", "test"], loc="upper left")

plt.savefig(loss_trend_graph_path)

plt.close(2)

print("We are done, everything seems OK...")

# #windows系统设置10关机

os.system("shutdown -s -t 10")

img

img

完整代码： import os
import numpy as np
from tensorflow import keras
from tensorflow.keras.optimizers import Adam
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Dropout,BatchNormalization,Flatten
from tensorflow.keras.layers import Conv2D, MaxPooling2D,GlobalAveragePooling2D
import cv2
from tensorflow.keras.preprocessing.image import img_to_array
from sklearn.model_selection import train_test_split
from tensorflow.python.keras import Input
from tensorflow.python.keras.callbacks import ModelCheckpoint, ReduceLROnPlateau
from tensorflow.python.keras.layers import PReLU, Activation
from tensorflow.python.keras.models import Model

norm_size=100
datapath='train'
EPOCHS =100
INIT_LR = 1e-3
labelList=[]
dicClass={'cat':0,'dog':1}
labelnum=2
batch_size = 4
def loadImageData():
imageList = []
listImage=os.listdir(datapath)
for img in listImage:
labelName=dicClass[img.split('.')[0]]
print(labelName)
labelList.append(labelName)
dataImgPath=os.path.join(datapath,img)
print(dataImgPath)
image = cv2.imdecode(np.fromfile(dataImgPath, dtype=np.uint8), -1)
# load the image, pre-process it, and store it in the data list
image = cv2.resize(image, (norm_size, norm_size), interpolation=cv2.INTER_LANCZOS4)
image = img_to_array(image)
imageList.append(image)
imageList = np.array(imageList, dtype="int") / 255.0
return imageList
print("开始加载数据")
imageArr=loadImageData()
labelList = np.array(labelList)
print("加载数据完成")
print(labelList)
def bn_prelu(x):
x = BatchNormalization(epsilon=1e-5)(x)
x = PReLU()(x)
return x

def build_model(out_dims, input_shape=(norm_size, norm_size, 3)):
inputs_dim = Input(input_shape)
x = Conv2D(32, (3, 3), strides=(2, 2), padding='same')(inputs_dim)
x = bn_prelu(x)
x = Conv2D(32, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Conv2D(64, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = Conv2D(64, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Conv2D(128, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = Conv2D(128, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = MaxPooling2D(pool_size=(2, 2))(x)
x = Conv2D(256, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = Conv2D(256, (3, 3), strides=(1, 1), padding='same')(x)
x = bn_prelu(x)
x = GlobalAveragePooling2D()(x)
dp_1 = Dropout(0.5)(x)
fc2 = Dense(out_dims)(dp_1)
fc2 = Activation('softmax')(fc2) #此处注意，为sigmoid函数
model = Model(inputs=inputs_dim, outputs=fc2)
return model
model=build_model(labelnum)
optimizer = Adam(lr=INIT_LR)
model.compile(optimizer =optimizer, loss='sparse_categorical_crossentropy', metrics=['accuracy'])
trainX,valX,trainY,valY = train_test_split(imageArr,labelList, test_size=0.3, random_state=42)
from tensorflow.keras.preprocessing.image import ImageDataGenerator
train_datagen = ImageDataGenerator(featurewise_center=True,
featurewise_std_normalization=True,
rotation_range=20,
width_shift_range=0.2,
height_shift_range=0.2,
horizontal_flip=True)
val_datagen = ImageDataGenerator() #验证集不做图片增强

train_generator = train_datagen.flow(trainX,trainY,batch_size=batch_size,shuffle=True)
val_generator = val_datagen.flow(valX,valY,batch_size=batch_size,shuffle=True)
checkpointer = ModelCheckpoint(filepath='weights_best_simple_model.hdf5',
monitor='val_accuracy',verbose=1, save_best_only=True, mode='max')

reduce = ReduceLROnPlateau(monitor='val_accuracy',patience=10,
verbose=1,
factor=0.5,
min_lr=1e-6)
history = model.fit_generator(train_generator,
steps_per_epoch=trainX.shape[0]/batch_size,
validation_data = val_generator,
epochs=EPOCHS,
validation_steps=valX.shape[0]/batch_size,
callbacks=[checkpointer,reduce],
verbose=1,shuffle=True)
model.save('my_model_.h5')
print(history)
import os
loss_trend_graph_path = r"WW_loss.jpg"
acc_trend_graph_path = r"WW_acc.jpg"
import matplotlib.pyplot as plt
print("Now,we start drawing the loss and acc trends graph...")
# summarize history for accuracy
fig = plt.figure(1)
plt.plot(history.history["accuracy"])
plt.plot(history.history["val_accuracy"])
plt.title("Model accuracy")
plt.ylabel("accuracy")
plt.xlabel("epoch")
plt.legend(["train", "test"], loc="upper left")
plt.savefig(acc_trend_graph_path)
plt.close(1)
# summarize history for loss
fig = plt.figure(2)
plt.plot(history.history["loss"])
plt.plot(history.history["val_loss"])
plt.title("Model loss")
plt.ylabel("loss")
plt.xlabel("epoch")
plt.legend(["train", "test"], loc="upper left")
plt.savefig(loss_trend_graph_path)
plt.close(2)
print("We are done, everything seems OK...")
# #windows系统设置10关机
os.system("shutdown -s -t 10")
测试部分 1、导入依赖 import cv2
import numpy as np
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.models import load_model
import time
2、设置全局参数 norm_size=100
imagelist=[]
emotion_labels = {
0: 'cat',
1: 'dog'
}
3、加载模型 emotion_classifier=load_model(**"my_model_.h5"**)
t1=time.time()
4、处理图片 image = cv2.imdecode(np.fromfile(**'test/8.jpg'**, dtype=np.uint8), -1)
\# load the image, pre-process it, and store it in the data list
image = cv2.resize(image, (norm_size, norm_size), interpolation=cv2.INTER_LANCZOS4)
image = img_to_array(image)
imagelist.append(image)
imageList = np.array(imagelist, dtype=**"float"**) / 255.0
5、预测类别 pre=np.argmax(emotion_classifier.predict(imageList))
emotion = emotion_labels[pre]
t2=time.time()
print(emotion)
t3=t2-t1
print(t3)
完整代码 import cv2
import numpy as np
from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.models import load_model
import time
norm_size=100
imagelist=[]
emotion_labels = {
0: 'cat',
1: 'dog'
}
emotion_classifier=load_model("my_model_.h5")
t1=time.time()
image = cv2.imdecode(np.fromfile('test/8.jpg', dtype=np.uint8), -1)
# load the image, pre-process it, and store it in the data list
image = cv2.resize(image, (norm_size, norm_size), interpolation=cv2.INTER_LANCZOS4)
image = img_to_array(image)
imagelist.append(image)
imageList = np.array(imagelist, dtype="float") / 255.0
pre=np.argmax(emotion_classifier.predict(imageList))
emotion = emotion_labels[pre]
t2=time.time()
print(emotion)
t3=t2-t1
print(t3)