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

from tensorflow       import keras
from tensorflow.keras import layers,models
import numpy             as np
import matplotlib.pyplot as plt
import os,PIL,pathlib
import tensorflow as tf
import warnings as w
w.filterwarnings('ignore')data_dir = "./coffee/"
data_dir = pathlib.Path(data_dir)image_count = len(list(data_dir.glob('*/*.png')))print("图片总数为：",image_count)

图片总数为： 1200

二、数据预处理

1. 加载数据

使用image_dataset_from_directory方法将磁盘中的数据加载到tf.data.Dataset中

batch_size = 32
img_height = 224
img_width = 224

train_ds = tf.keras.preprocessing.image_dataset_from_directory(data_dir,validation_split=0.2,subset="training",seed=123,image_size=(img_height, img_width),batch_size=batch_size)

Found 1200 files belonging to 4 classes.
Using 960 files for training.

val_ds = tf.keras.preprocessing.image_dataset_from_directory(data_dir,validation_split=0.2,subset="validation",seed=123,image_size=(img_height, img_width),batch_size=batch_size)

Found 1200 files belonging to 4 classes.
Using 240 files for validation.

class_names = train_ds.class_names
print(class_names)

['Dark', 'Green', 'Light', 'Medium']

2.数据可视化 

plt.figure(figsize=(10, 4))  # 图形的宽为10高为5for images, labels in train_ds.take(1):for i in range(10):ax = plt.subplot(2, 5, i + 1)  plt.imshow(images[i].numpy().astype("uint8"))plt.title(class_names[labels[i]])plt.axis("off")

for image_batch, labels_batch in train_ds:print(image_batch.shape)print(labels_batch.shape)break

for image_batch, labels_batch in train_ds:print(image_batch.shape)print(labels_batch.shape)break

(32, 224, 224, 3)
(32,)

3. 配置数据集

• shuffle() ：打乱数据，关于此函数的详细介绍可以参考：https://zhuanlan.zhihu.com/p/42417456

• prefetch() ：预取数据，加速运行，其详细介绍可以参考我前两篇文章，里面都有讲解。

• cache() ：将数据集缓存到内存当中，加速运行

AUTOTUNE = tf.data.AUTOTUNEtrain_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds   = val_ds.cache().prefetch(buffer_size=AUTOTUNE)

 并且将数据归一化

normalization_layer = layers.experimental.preprocessing.Rescaling(1./255)train_ds = train_ds.map(lambda x, y: (normalization_layer(x), y))
val_ds   = val_ds.map(lambda x, y: (normalization_layer(x), y))image_batch, labels_batch = next(iter(val_ds))
first_image = image_batch[0]# 查看归一化后的数据
print(np.min(first_image), np.max(first_image))

 0.0 1.0

三、构建VGG-16网络

1.VGG优缺点分析：

• VGG优点

VGG的结构非常简洁，整个网络都使用了同样大小的卷积核尺寸（3x3）和最大池化尺寸（2x2）。

• VGG缺点

1)训练时间过长，调参难度大。2)需要的存储容量大，不利于部署。例如存储VGG-16权重值文件的大小为500多MB，不利于安装到嵌入式系统中。

2.网络结构图

结构说明：

• 13个卷积层（Convolutional Layer），分别用blockX_convX表示

• 3个全连接层（Fully connected Layer），分别用fcX与predictions表示

• 5个池化层（Pool layer），分别用blockX_pool表示

VGG-16包含了16个隐藏层（13个卷积层和3个全连接层），故称为VGG-16

 

model = tf.keras.applications.VGG16(weights='imagenet')
model.summary()

Model: "vgg16"
_________________________________________________________________Layer (type)                Output Shape              Param #   
=================================================================input_1 (InputLayer)        [(None, 224, 224, 3)]     0         block1_conv1 (Conv2D)       (None, 224, 224, 64)      1792      block1_conv2 (Conv2D)       (None, 224, 224, 64)      36928     block1_pool (MaxPooling2D)  (None, 112, 112, 64)      0         block2_conv1 (Conv2D)       (None, 112, 112, 128)     73856     block2_conv2 (Conv2D)       (None, 112, 112, 128)     147584    block2_pool (MaxPooling2D)  (None, 56, 56, 128)       0         block3_conv1 (Conv2D)       (None, 56, 56, 256)       295168    block3_conv2 (Conv2D)       (None, 56, 56, 256)       590080    block3_conv3 (Conv2D)       (None, 56, 56, 256)       590080    block3_pool (MaxPooling2D)  (None, 28, 28, 256)       0         block4_conv1 (Conv2D)       (None, 28, 28, 512)       1180160   block4_conv2 (Conv2D)       (None, 28, 28, 512)       2359808   block4_conv3 (Conv2D)       (None, 28, 28, 512)       2359808   block4_pool (MaxPooling2D)  (None, 14, 14, 512)       0         block5_conv1 (Conv2D)       (None, 14, 14, 512)       2359808   block5_conv2 (Conv2D)       (None, 14, 14, 512)       2359808   block5_conv3 (Conv2D)       (None, 14, 14, 512)       2359808   block5_pool (MaxPooling2D)  (None, 7, 7, 512)         0         flatten (Flatten)           (None, 25088)             0         fc1 (Dense)                 (None, 4096)              102764544 fc2 (Dense)                 (None, 4096)              16781312  predictions (Dense)         (None, 1000)              4097000   =================================================================
Total params: 138,357,544
Trainable params: 138,357,544
Non-trainable params: 0
_________________________________________________________________

四、编译

在准备对模型进行训练之前，还需要再对其进行一些设置。以下内容是在模型的编译步骤中添加的：

• 损失函数（loss）：用于衡量模型在训练期间的准确率。

• 优化器（optimizer）：决定模型如何根据其看到的数据和自身的损失函数进行更新。

• 指标（metrics）：用于监控训练和测试步骤。以下示例使用了准确率，即被正确分类的图像的比率。

# 设置初始学习率
initial_learning_rate = 1e-4lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(initial_learning_rate, decay_steps=30,      # 敲黑板！！！这里是指 steps，不是指epochsdecay_rate=0.92,     # lr经过一次衰减就会变成 decay_rate*lrstaircase=True)# 设置优化器
opt = tf.keras.optimizers.Adam(learning_rate=initial_learning_rate)model.compile(optimizer=opt,loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),metrics=['accuracy'])

五、训练模型 

epochs = 20history = model.fit(train_ds,validation_data=val_ds,epochs=epochs
)

Epoch 1/20
30/30 [==============================] - 346s 11s/step - loss: 1.7546 - accuracy: 0.2625 - val_loss: 1.4646 - val_accuracy: 0.2125
Epoch 2/20
30/30 [==============================] - 352s 12s/step - loss: 1.3637 - accuracy: 0.3104 - val_loss: 1.0428 - val_accuracy: 0.4583
Epoch 3/20
30/30 [==============================] - 338s 11s/step - loss: 0.7237 - accuracy: 0.6458 - val_loss: 0.4818 - val_accuracy: 0.7833
Epoch 4/20
30/30 [==============================] - 336s 11s/step - loss: 0.3633 - accuracy: 0.8479 - val_loss: 1.1034 - val_accuracy: 0.6167
Epoch 5/20
30/30 [==============================] - 340s 11s/step - loss: 0.2880 - accuracy: 0.8927 - val_loss: 0.1480 - val_accuracy: 0.9500
Epoch 6/20
30/30 [==============================] - 338s 11s/step - loss: 0.1802 - accuracy: 0.9333 - val_loss: 0.4709 - val_accuracy: 0.8458
Epoch 7/20
30/30 [==============================] - 334s 11s/step - loss: 0.1468 - accuracy: 0.9490 - val_loss: 0.0214 - val_accuracy: 1.0000
Epoch 8/20
30/30 [==============================] - 339s 11s/step - loss: 0.0174 - accuracy: 0.9969 - val_loss: 0.0196 - val_accuracy: 0.9875
Epoch 9/20
30/30 [==============================] - 329s 11s/step - loss: 0.0399 - accuracy: 0.9875 - val_loss: 0.2539 - val_accuracy: 0.9292
Epoch 10/20
30/30 [==============================] - 330s 11s/step - loss: 0.2606 - accuracy: 0.9073 - val_loss: 0.0737 - val_accuracy: 0.9917
Epoch 11/20
30/30 [==============================] - 334s 11s/step - loss: 0.0610 - accuracy: 0.9812 - val_loss: 0.0070 - val_accuracy: 1.0000
Epoch 12/20
30/30 [==============================] - 341s 11s/step - loss: 0.0296 - accuracy: 0.9917 - val_loss: 0.0256 - val_accuracy: 0.9875
Epoch 13/20
30/30 [==============================] - 335s 11s/step - loss: 0.0252 - accuracy: 0.9917 - val_loss: 0.0431 - val_accuracy: 0.9833
Epoch 14/20
30/30 [==============================] - 345s 12s/step - loss: 0.0058 - accuracy: 0.9979 - val_loss: 0.0088 - val_accuracy: 0.9958
Epoch 15/20
30/30 [==============================] - 557s 19s/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 0.0144 - val_accuracy: 0.9917
Epoch 16/20
30/30 [==============================] - 340s 11s/step - loss: 3.6823e-04 - accuracy: 1.0000 - val_loss: 0.0052 - val_accuracy: 0.9958
Epoch 17/20
30/30 [==============================] - 347s 12s/step - loss: 5.9116e-05 - accuracy: 1.0000 - val_loss: 0.0064 - val_accuracy: 0.9958
Epoch 18/20
30/30 [==============================] - 347s 12s/step - loss: 2.5309e-05 - accuracy: 1.0000 - val_loss: 0.0048 - val_accuracy: 0.9958
Epoch 19/20
30/30 [==============================] - 350s 12s/step - loss: 1.0864e-05 - accuracy: 1.0000 - val_loss: 0.0033 - val_accuracy: 1.0000
Epoch 20/20
30/30 [==============================] - 341s 11s/step - loss: 6.0013e-06 - accuracy: 1.0000 - val_loss: 0.0045 - val_accuracy: 0.9958

 六 可视化结果

acc = history.history['accuracy']
val_acc = history.history['val_accuracy']loss = history.history['loss']
val_loss = history.history['val_loss']epochs_range = range(epochs)plt.figure(figsize=(12, 4))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, acc, label='Training Accuracy')
plt.plot(epochs_range, val_acc, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')plt.subplot(1, 2, 2)
plt.plot(epochs_range, loss, label='Training Loss')
plt.plot(epochs_range, val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()

预测图片 

import numpy as np# 采用加载的模型（new_model）来看预测结果
plt.figure(figsize=(18, 3))  # 图形的宽为18高为5
plt.suptitle("预测结果展示")for images, labels in val_ds.take(1):for i in range(8):ax = plt.subplot(1,8, i + 1)  # 显示图片plt.imshow(images[i].numpy())# 需要给图片增加一个维度img_array = tf.expand_dims(images[i], 0) # 使用模型预测图片中的人物predictions = model.predict(img_array)plt.title(class_names[np.argmax(predictions)])plt.axis("off")

1/1 [==============================] - 0s 279ms/step
1/1 [==============================] - 0s 110ms/step
1/1 [==============================] - 0s 118ms/step
1/1 [==============================] - 0s 109ms/step
1/1 [==============================] - 0s 110ms/step
1/1 [==============================] - 0s 104ms/step
1/1 [==============================] - 0s 111ms/step
1/1 [==============================] - 0s 115ms/step