cifar10图像分类--卷积神经网络

链接：https://pan.baidu.com/s/1Nxty7ntGSUUA18oa_-ORSQ
提取码：in9m
cifar10数据集百度云链接

从本地加载图片

import os
import numpy as np
import pickle as p

def load_CIFAR_batch(filename):
    with open(filename,'rb')as f:
        data_dict = p.load(f,encoding='bytes')
        images = data_dict[b'data']
        labels =data_dict[b'labels']
        
        images = images.reshape(10000,3,32,32)
        images = images.transpose(0,2,3,1)
        labels = np.array(labels)
        return images,labels

def load_CIFAR_data(data_dir):
    images_train=[]
    labels_train=[]
    for i in range(5):
        f = os.path.join(data_dir,'data_batch_%d'%(i+1))
        print('loading',f)
        images_batch,label_batch = load_CIFAR_batch(f)
        
        images_train.append(images_batch)
        labels_train.append(label_batch)
        
        Xtrain = np.concatenate(images_train)
        Ytrain = np.concatenate(labels_train)
        del images_batch,label_batch
    Xtest,Ytest = load_CIFAR_batch(os.path.join(data_dir,'test_batch'))
    print('finished loadding CIFAR-10 data')
    return Xtrain,Ytrain,Xtest,Ytest
data_dir = 'D:\cat_dog\cifar-10-python\cifar-10-batches-py'
train_images, train_labels, test_images, test_labels = load_CIFAR_data(data_dir)

train_images, test_images = train_images / 255.0, test_images / 255.0
import tensorflow as tf
from tensorflow.keras import datasets, layers, models
import matplotlib.pyplot as plt

展示图片

class_names = ['airplane', 'automobile', 'bird', 'cat', 'deer','dog', 'frog', 'horse', 'ship', 'truck']

plt.figure(figsize=(10,10))
for i in range(25):
    plt.subplot(5,5,i+1)
    plt.xticks([])
    plt.yticks([])
    plt.grid(False)
    plt.imshow(train_images[i], cmap=plt.cm.binary)
    # The CIFAR labels happen to be arrays, 
    # which is why you need the extra index
    plt.xlabel(class_names[train_labels[i]])
plt.show()

建个模型

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
# 显示模型信息
model.summary()

model.add(layers.Flatten()) # 3维 转为 1维
model.add(layers.Dense(64, activation='relu'))  # **函数relu
model.add(layers.Dense(10, activation='softmax'))  # **函数softmax CIFAR有10个类别输出，所以softmax这里参数设置为10
# 再看看模型情况
model.summary()

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

checkpoint_path = "training_1/cp.ckpt"
checkpoint_dir = os.path.dirname(checkpoint_path)

# 创建一个保存模型权重的回调
cp_callback = tf.keras.callbacks.ModelCheckpoint(filepath=checkpoint_path,
                                                 save_weights_only=True,
                                                 verbose=2)

history = model.fit(train_images, train_labels, epochs=10, 
                    validation_data=(test_images, test_labels),
                   callbacks=[cp_callback])

结果看起来还不错

绘制精确度

plt.plot(history.history['accuracy'], label='accuracy')
plt.plot(history.history['val_accuracy'], label = 'val_accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.ylim([0.5, 1])
plt.legend(loc='lower right')

test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)

不加卷积和池化的神经网络

结果是比较差的。

拓展

用CNN分类minist

可以取得更高的精度，加载完数据后需要给最后一维增加一个通道。

增加卷积和池化层

model = models.Sequential()
model.add(layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (3, 3), activation='relu'))
model.add(layers.MaxPooling2D((2, 2)))
model.add(layers.Conv2D(64, (2, 2), activation='relu'))

model.add(layers.Flatten()) # 3维 转为 1维
model.add(layers.Dense(64, activation='relu'))  # **函数relu
model.add(layers.Dense(10, activation='softmax'))  # **函数softmax CIFAR有10个类别输出，所以softmax这里参数设置为10
# 再看看模型情况
model.summary()

模型的参数变得更少了
那么结果如何呢

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

history = model.fit(train_images, train_labels, epochs=10, 
                    validation_data=(test_images, test_labels),
                   )

准确度没有提升多少

plt.plot(history.history['accuracy'], label='accuracy')
plt.plot(history.history['val_accuracy'], label = 'val_accuracy')
plt.xlabel('Epoch')
plt.ylabel('Accuracy')
plt.ylim([0.5, 1])
plt.legend(loc='lower right')

test_loss, test_acc = model.evaluate(test_images,  test_labels, verbose=2)