【深度学习实战—4】：基于AlexNet的手写数字识别（附Keras实现）

✨博客主页：王乐予🎈
✨年轻人要：Living for the moment（活在当下）！💪
🏆推荐专栏：【图像处理】【千锤百炼Python】【深度学习】【排序算法】

在【深度学习实战—3】：基于LetNet-5的手写数字识别（附Keras实现）中，我们使用LeNet-5实现了手写数字的识别；

在【深度学习经典网络架构—2】：AlexNet（附Keras实现）中我们讲了AlexNet经典网络；

今天带来的是使用AlexNet经典网络进行手写数字识别！文末附完整程序！

很多地方，如数据集获取，网络训练等都和【深度学习实战—3】：基于LetNet-5的手写数字识别（附Keras实现）差不多，有不同的就是网络的搭建，

😺一、网络搭建

因为论文中的alexnet的输入是（227, 227, 3），所以我们在搭建网络的时候，要将输入改为（28, 28, 1）。

def alexnet():
    model = Sequential()

    model.add(Conv2D(96, (11, 11), strides=(4, 4), input_shape=(28, 28, 1), padding='same', activation='relu',
                     kernel_initializer='uniform'))
    model.add(MaxPooling2D(pool_size=(3, 3), strides=(2, 2)))
    model.add(BatchNormalization())

    model.add(Conv2D(256, (5, 5), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))
    model.add(MaxPooling2D(pool_size=(3, 3), strides=(2, 2)))
    model.add(BatchNormalization())

    model.add(Conv2D(384, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Conv2D(384, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Conv2D(256, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Flatten())
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(10, activation='softmax'))
    print(model.summary())

    return model

😺二、训练过程

由于程序很多地方与LeNet-5手写数字识别类似，因此这里只放训练过程图，文末查看完整程序！

😺三、测试模型

Test loss: 0.08178182691335678
Test accuracy: 0.9837999939918518
前20张图片预测结果： [7 2 1 0 4 1 4 9 5 9 0 6 9 0 1 5 9 7 3 4]

😺四、预测结果可视化

😺五、建立混淆矩阵

😺附录：完整程序

from keras.datasets import mnist
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import seaborn as sns
from sklearn.metrics import confusion_matrix
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Input, Dropout, BatchNormalization
from keras.models import Model, Sequential
from keras.utils import np_utils


"""
数据集获取
"""
def get_mnist_data():

    (x_train_original, y_train_original), (x_test_original, y_test_original) = mnist.load_data()

    # 从训练集中分配验证集
    x_val = x_train_original[50000:]
    y_val = y_train_original[50000:]
    x_train = x_train_original[:50000]
    y_train = y_train_original[:50000]

    # 将图像转换为四维矩阵(nums,rows,cols,channels), 这里把数据从unint类型转化为float32类型, 提高训练精度。
    x_train = x_train.reshape(x_train.shape[0], 28, 28, 1).astype('float32')
    x_val = x_val.reshape(x_val.shape[0], 28, 28, 1).astype('float32')
    x_test = x_test_original.reshape(x_test_original.shape[0], 28, 28, 1).astype('float32')

    # 原始图像的像素灰度值为0-255，为了提高模型的训练精度，通常将数值归一化映射到0-1。
    x_train = x_train / 255
    x_val = x_val / 255
    x_test = x_test / 255

    # 图像标签一共有10个类别即0-9，这里将其转化为独热编码（One-hot）向量
    y_train = np_utils.to_categorical(y_train)
    y_val = np_utils.to_categorical(y_val)
    y_test = np_utils.to_categorical(y_test_original)

    return x_train, y_train, x_val, y_val, x_test, y_test


"""
定义alexnet网络模型
"""
def alexnet():
    model = Sequential()

    model.add(Conv2D(96, (11, 11), strides=(4, 4), input_shape=(28, 28, 1), padding='same', activation='relu',
                     kernel_initializer='uniform'))
    model.add(MaxPooling2D(pool_size=(3, 3), strides=(2, 2)))
    model.add(BatchNormalization())

    model.add(Conv2D(256, (5, 5), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))
    model.add(MaxPooling2D(pool_size=(3, 3), strides=(2, 2)))
    model.add(BatchNormalization())

    model.add(Conv2D(384, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Conv2D(384, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Conv2D(256, (3, 3), strides=(1, 1), padding='same', activation='relu', kernel_initializer='uniform'))

    model.add(Flatten())
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(4096, activation='relu'))
    model.add(Dropout(0.5))
    model.add(Dense(10, activation='softmax'))
    print(model.summary())

    return model

"""
编译网络并训练
"""
x_train, y_train, x_val, y_val, x_test, y_test = get_mnist_data()
model = alexnet()

# 编译网络（定义损失函数、优化器、评估指标）
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])

# 开始网络训练（定义训练数据与验证数据、定义训练代数，定义训练批大小）
train_history = model.fit(x_train, y_train, validation_data=(x_val, y_val), epochs=10, batch_size=32, verbose=2)

# 模型保存
model.save('alexnet_mnist.h5')

# 定义训练过程可视化函数（训练集损失、验证集损失、训练集精度、验证集精度）
def show_train_history(train_history, train, validation):
    plt.plot(train_history.history[train])
    plt.plot(train_history.history[validation])
    plt.title('Train History')
    plt.ylabel(train)
    plt.xlabel('Epoch')
    plt.legend(['train', 'validation'], loc='best')
    plt.show()

show_train_history(train_history, 'accuracy', 'val_accuracy')
show_train_history(train_history, 'loss', 'val_loss')

# 输出网络在测试集上的损失与精度
score = model.evaluate(x_test, y_test)
print('Test loss:', score[0])
print('Test accuracy:', score[1])

# 测试集结果预测
predictions = model.predict(x_test)
predictions = np.argmax(predictions, axis=1)
print('前20张图片预测结果：', predictions[:20])

# 预测结果图像可视化
(x_train_original, y_train_original), (x_test_original, y_test_original) = mnist.load_data()
def mnist_visualize_multiple_predict(start, end, length, width):

    for i in range(start, end):
        plt.subplot(length, width, 1 + i)
        plt.imshow(x_test_original[i], cmap=plt.get_cmap('gray'))
        title_true = 'true=' + str(y_test_original[i])
        # title_prediction = ',' + 'prediction' + str(model.predict_classes(np.expand_dims(x_test[i], axis=0)))
        title_prediction = ',' + 'prediction' + str(predictions[i])
        title = title_true + title_prediction
        plt.title(title)
        plt.xticks([])
        plt.yticks([])
    plt.show()

mnist_visualize_multiple_predict(start=0, end=9, length=3, width=3)

# 混淆矩阵
cm = confusion_matrix(y_test_original, predictions)
cm = pd.DataFrame(cm)
class_names = ['0', '1', '2', '3', '4', '5', '6', '7', '8', '9']

def plot_confusion_matrix(cm):
    plt.figure(figsize=(10, 10))
    sns.heatmap(cm, cmap='Oranges', linecolor='black', linewidth=1, annot=True, fmt='', xticklabels=class_names, yticklabels=class_names)
    plt.xlabel("Predicted")
    plt.ylabel("Actual")
    plt.title("Confusion Matrix")
    plt.show()

plot_confusion_matrix(cm)