基于VGG-19的猫十二分类

1.数据详情

cat_12数据集包含3个部分，训练集cat_12_train,测试集cat_test,以及存储图片名称及标签的train_list.txt

2.数据预处理

train_list.txt文件存储了文件名称和标签，文件格式如下，我们定义了prepare_image函数分别将文件路径与标签存储于X_train与y_train中,并打乱数据

 加载数据时，使用keras.image模块进行数据的加载，并进行数据归一化操作

from keras.preprocessing import image
import random
# 分离路径与标签
def prepare_image(file_path):
    X_train = []
    y_train = []

    with open(file_path) as f:
        context = f.readlines()
    random.shuffle(context)

    for str in context:
        str = str.strip('\n').split('\t')

        X_train.append('./cat_12/' + str[0])
        y_train.append(str[1])

    return X_train, y_train


# 数据归一化
def preprocess_image(img):
    img = image.load_img(img, target_size=(224, 224))
    img = image.img_to_array(img)
    img = img / 255.0
    return img

3.构建VGG 19并进行训练

（1）构建VGG_block模块

一个模块就代表VGG网络中的卷积组

class Vgg_Block(keras.layers.Layer):
    def __init__(self, units, filters, **kwargs):
        super().__init__(**kwargs)
        self.main_layer = []
        for i in range(units):
            self.main_layer.append(keras.layers.Conv2D(filters=filters, kernel_size=(3, 2),
                                                       padding="same", strides=(1, 1),
                                                       activation="relu"))
        self.main_layer.append(keras.layers.MaxPool2D(pool_size=(2, 2)))

    def call(self, inputs):
        Z = inputs
        for layer in self.main_layer:
            Z = layer(Z)
        return Z

（2）构建VGG 19模型

model = keras.models.Sequential()
model.add(keras.layers.Input(shape=(224, 224, 3)))
model.add(Vgg_Block(2, 64))
model.add(Vgg_Block(2, 128))
model.add(Vgg_Block(4, 256))
model.add(Vgg_Block(4, 512))
model.add(Vgg_Block(4, 512))
model.add(keras.layers.Flatten())
model.add(keras.layers.Dense(4096, activation="relu"))
model.add(keras.layers.Dropout(0.5))
model.add(keras.layers.Dense(4096, activation="relu"))
model.add(keras.layers.Dropout(0.5))
model.add(keras.layers.Dense(12, activation="softmax"))

（3）.模型训练完整代码如下

from dataset import prepare_image, preprocess_image
import tensorflow as tf
from tqdm import tqdm
import numpy as np
from matplotlib import pyplot as plt

X_train, y_train = prepare_image('./cat_12/train_list.txt')
train_images = []

for i in tqdm(X_train):
    train_image = preprocess_image(i)
    train_images.append(train_image)

from tensorflow import keras

train_images = np.array(train_images)
print(train_images.shape)
y_train = keras.utils.to_categorical(y_train, 12)


class Vgg_Block(keras.layers.Layer):
    def __init__(self, units, filters, **kwargs):
        super().__init__(**kwargs)
        self.main_layer = []
        for i in range(units):
            self.main_layer.append(keras.layers.Conv2D(filters=filters, kernel_size=(3, 2),
                                                       padding="same", strides=(1, 1),
                                                       activation="relu"))
        self.main_layer.append(keras.layers.MaxPool2D(pool_size=(2, 2)))

    def call(self, inputs):
        Z = inputs
        for layer in self.main_layer:
            Z = layer(Z)
        return Z


model = keras.models.Sequential()
model.add(keras.layers.Input(shape=(224, 224, 3)))
model.add(Vgg_Block(2, 64))
model.add(Vgg_Block(2, 128))
model.add(Vgg_Block(4, 256))
model.add(Vgg_Block(4, 512))
model.add(Vgg_Block(4, 512))
model.add(keras.layers.Flatten())
model.add(keras.layers.Dense(4096, activation="relu"))
model.add(keras.layers.Dropout(0.5))
model.add(keras.layers.Dense(4096, activation="relu"))
model.add(keras.layers.Dropout(0.5))
model.add(keras.layers.Dense(12, activation="softmax"))

# keras.utils.plot_model(model=model, to_file='AlexNet.png', show_shapes=True)

model.compile(loss="categorical_crossentropy", optimizer="sgd", metrics=["accuracy"])
history = model.fit(train_images, y_train, epochs=20, batch_size=16, validation_split=0.2)

model.save('the_AlexNet_model.h5')


def show_training_history(train_history, train, val):
    plt.plot(train_history[train], linestyle='-', color='b')
    plt.plot(train_history[val], linestyle='--', color='r')
    plt.xlabel('Epoch', fontsize=12)
    plt.ylabel('train', fontsize=12)
    plt.legend(['train', 'validation'], loc='lower right')
    plt.show()


show_training_history(history.history, 'loss', 'val_loss')
show_training_history(history.history, 'acc', 'val_acc')