Tensorflow深度学习案例2：cifar10彩色图片识别

本文为为🔗365天深度学习训练营内部文章

原作者：K同学啊

一、前期工作

1. 导入数据

import tensorflow as tf
from tensorflow.keras import datasets, layers, models
import matplotlib.pyplot as plt

(train_images, train_labels), (test_images, test_labels) = datasets.cifar10.load_data()

A local file was found, but it seems to be incomplete or outdated because the auto file hash does not match the original value of 6d958be074577803d12ecdefd02955f39262c83c16fe9348329d7fe0b5c001ce so we will re-download the data.
Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz
170498071/170498071 [==============================] - 33s 0us/step

2.归一化 

# 将像素的值标准化至0到1的区间内。
train_images, test_images = train_images / 255.0, test_images / 255.0

train_images.shape,test_images.shape,train_labels.shape,test_labels.shape

((50000, 32, 32, 3), (10000, 32, 32, 3), (50000, 1), (10000, 1))

 3.可视化

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

plt.figure(figsize=(20,10))
for i in range(20):
    plt.subplot(5,10,i+1)
    plt.xticks([])
    plt.yticks([])
    plt.grid(False)
    plt.imshow(train_images[i], cmap=plt.cm.binary)
    plt.xlabel(class_names[train_labels[i][0]])
plt.show()

 

二 构建CNN网络 

⭐关于卷积层：可参考【知识储备】部分

⭐池化层

池化层对提取到的特征信息进行降维，一方面使特征图变小，简化网络计算复杂度；另一方面进行特征压缩，提取主要特征，增加平移不变性，减少过拟合风险。但其实池化更多程度上是一种计算性能的一个妥协，强硬地压缩特征的同时也损失了一部分信息，所以现在的网络比较少用池化层或者使用优化后的如SoftPool。

池化层包括最大池化层（MaxPooling）和平均池化层（AveragePooling），均值池化对背景保留更好，最大池化对纹理提取更好）。同卷积计算，池化层计算窗口内的平均值或者最大值。例如通过一个 2*2 的最大池化层，其计算方式如下：

 

 构建模型的结构图

 

model = models.Sequential([
    layers.Conv2D(32, (3, 3), activation='relu', input_shape=(32, 32, 3)), #卷积层1，卷积核3*3
    layers.MaxPooling2D((2, 2)),                   #池化层1，2*2采样
    layers.Conv2D(64, (3, 3), activation='relu'),  #卷积层2，卷积核3*3
    layers.MaxPooling2D((2, 2)),                   #池化层2，2*2采样
    layers.Conv2D(64, (3, 3), activation='relu'),  #卷积层3，卷积核3*3
    
    layers.Flatten(),                      #Flatten层，连接卷积层与全连接层
    layers.Dense(64, activation='relu'),   #全连接层，特征进一步提取
    layers.Dense(10)                       #输出层，输出预期结果
])

model.summary()  # 打印网络结构

Model: "sequential"
_________________________________________________________________
 Layer (type)                Output Shape              Param #   
=================================================================
 conv2d (Conv2D)             (None, 30, 30, 32)        896       
                                                                 
 max_pooling2d (MaxPooling2D  (None, 15, 15, 32)       0         
 )                                                               
                                                                 
 conv2d_1 (Conv2D)           (None, 13, 13, 64)        18496     
                                                                 
 max_pooling2d_1 (MaxPooling  (None, 6, 6, 64)         0         
 2D)                                                             
                                                                 
 conv2d_2 (Conv2D)           (None, 4, 4, 64)          36928     
                                                                 
 flatten (Flatten)           (None, 1024)              0         
                                                                 
 dense (Dense)               (None, 64)                65600     
                                                                 
 dense_1 (Dense)             (None, 10)                650       
                                                                 
=================================================================
Total params: 122,570
Trainable params: 122,570
Non-trainable params: 0
_________________________________________________________________

三 编译 

选择Adam优化器

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

四 训练模型

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

Epoch 1/10
1563/1563 [==============================] - 24s 15ms/step - loss: 1.5113 - accuracy: 0.4479 - val_loss: 1.2449 - val_accuracy: 0.5568
Epoch 2/10
1563/1563 [==============================] - 23s 15ms/step - loss: 1.1505 - accuracy: 0.5936 - val_loss: 1.1101 - val_accuracy: 0.6066
Epoch 3/10
1563/1563 [==============================] - 23s 15ms/step - loss: 1.0184 - accuracy: 0.6427 - val_loss: 0.9937 - val_accuracy: 0.6517
Epoch 4/10
1563/1563 [==============================] - 24s 15ms/step - loss: 0.9248 - accuracy: 0.6750 - val_loss: 0.9507 - val_accuracy: 0.6729
Epoch 5/10
1563/1563 [==============================] - 24s 15ms/step - loss: 0.8459 - accuracy: 0.7045 - val_loss: 0.9667 - val_accuracy: 0.6656
Epoch 6/10
1563/1563 [==============================] - 23s 15ms/step - loss: 0.7860 - accuracy: 0.7245 - val_loss: 0.8740 - val_accuracy: 0.6970
Epoch 7/10
1563/1563 [==============================] - 23s 14ms/step - loss: 0.7386 - accuracy: 0.7406 - val_loss: 0.8711 - val_accuracy: 0.7007
Epoch 8/10
1563/1563 [==============================] - 23s 14ms/step - loss: 0.6854 - accuracy: 0.7579 - val_loss: 0.8796 - val_accuracy: 0.7035
Epoch 9/10
1563/1563 [==============================] - 22s 14ms/step - loss: 0.6427 - accuracy: 0.7753 - val_loss: 0.8673 - val_accuracy: 0.7087
Epoch 10/10
1563/1563 [==============================] - 22s 14ms/step - loss: 0.6053 - accuracy: 0.7870 - val_loss: 0.8549 - val_accuracy: 0.7096

 五 预测

plt.imshow(test_images[1])
import numpy as np

pre = model.predict(test_images)
print(class_names[np.argmax(pre[1])])

 

 六 模型评估

import matplotlib.pyplot as plt

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')
plt.show()

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

print(test_acc)

0.7095999717712402

总结：

与上文手写数字识别架构、逻辑、思路、语句近似相同

导入内置数据集，归一化数据拆分训练集，构建模型训练并评估