- 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
- 🍖 原作者:K同学啊 | 接辅导、项目定制
一、我的环境
1.语言环境:Python 3.9
2.编译器:Pycharm
3.深度学习环境:TensorFlow 2.10.0
二、GPU设置
若使用的是cpu则可忽略
import tensorflow as tf
gpus = tf.config.list_physical_devices("GPU")
if gpus:
gpu0 = gpus[0] #如果有多个GPU,仅使用第0个GPU
tf.config.experimental.set_memory_growth(gpu0, True) #设置GPU显存用量按需使用
tf.config.set_visible_devices([gpu0],"GPU")
三**、导入数据**
python
import matplotlib.pyplot as plt
# 支持中文
plt.rcParams['font.sans-serif'] = ['SimHei'] # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False # 用来正常显示负号
import os,PIL,pathlib
#隐藏警告
import warnings
warnings.filterwarnings('ignore')
data_dir = "./data"
data_dir = pathlib.Path(data_dir)
image_count = len(list(data_dir.glob('*/*')))
print("图片总数为:",image_count)
#图片总数为:3400
四**、数据预处理**
python
batch_size = 8
img_height = 224
img_width = 224
"""
关于image_dataset_from_directory()的详细介绍可以参考文章:https://mtyjkh.blog.csdn.net/article/details/117018789
"""
train_ds = tf.keras.preprocessing.image_dataset_from_directory(
data_dir,
validation_split=0.2,
subset="training",
seed=12,
image_size=(img_height, img_width),
batch_size=batch_size)
"""
关于image_dataset_from_directory()的详细介绍可以参考文章:https://mtyjkh.blog.csdn.net/article/details/117018789
"""
val_ds = tf.keras.preprocessing.image_dataset_from_directory(
data_dir,
validation_split=0.2,
subset="validation",
seed=12,
image_size=(img_height, img_width),
batch_size=batch_size)
class_names = train_ds.class_names
print(class_names)
运行结果:
python
['cat', 'dog']
五、可视化图片
python
plt.figure(figsize=(15, 10)) # 图形的宽为15高为10
for images, labels in train_ds.take(1):
for i in range(8):
ax = plt.subplot(5, 8, i + 1)
plt.imshow(images[i])
plt.title(class_names[labels[i]])
plt.axis("off")
plt.show()
运行结果:
再次检查数据:
for image_batch, labels_batch in train_ds:
print(image_batch.shape)
print(labels_batch.shape)
break
运行结果:
(32, 224, 224, 3)
(32,)
六、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
- cache():将数据集缓存到内存当中,加速运行
python
AUTOTUNE = tf.data.AUTOTUNE
def preprocess_image(image,label):
return (image/255.0,label)
# 归一化处理
train_ds = train_ds.map(preprocess_image, num_parallel_calls=AUTOTUNE)
val_ds = val_ds.map(preprocess_image, num_parallel_calls=AUTOTUNE)
train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)
七、自建模型
python
from tensorflow.keras import layers, models, Input
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dense, Flatten, Dropout
def VGG16(nb_classes, input_shape):
input_tensor = Input(shape=input_shape)
# 1st block
x = Conv2D(64, (3,3), activation='relu', padding='same',name='block1_conv1')(input_tensor)
x = Conv2D(64, (3,3), activation='relu', padding='same',name='block1_conv2')(x)
x = MaxPooling2D((2,2), strides=(2,2), name = 'block1_pool')(x)
# 2nd block
x = Conv2D(128, (3,3), activation='relu', padding='same',name='block2_conv1')(x)
x = Conv2D(128, (3,3), activation='relu', padding='same',name='block2_conv2')(x)
x = MaxPooling2D((2,2), strides=(2,2), name = 'block2_pool')(x)
# 3rd block
x = Conv2D(256, (3,3), activation='relu', padding='same',name='block3_conv1')(x)
x = Conv2D(256, (3,3), activation='relu', padding='same',name='block3_conv2')(x)
x = Conv2D(256, (3,3), activation='relu', padding='same',name='block3_conv3')(x)
x = MaxPooling2D((2,2), strides=(2,2), name = 'block3_pool')(x)
# 4th block
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block4_conv1')(x)
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block4_conv2')(x)
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block4_conv3')(x)
x = MaxPooling2D((2,2), strides=(2,2), name = 'block4_pool')(x)
# 5th block
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block5_conv1')(x)
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block5_conv2')(x)
x = Conv2D(512, (3,3), activation='relu', padding='same',name='block5_conv3')(x)
x = MaxPooling2D((2,2), strides=(2,2), name = 'block5_pool')(x)
# full connection
x = Flatten()(x)
x = Dense(4096, activation='relu', name='fc1')(x)
x = Dense(4096, activation='relu', name='fc2')(x)
output_tensor = Dense(nb_classes, activation='softmax', name='predictions')(x)
model = Model(input_tensor, output_tensor)
return model
model=VGG16(1000, (img_width, img_height, 3))
model.summary()
运行结果:
python
_________________________________________________________________
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
_________________________________________________________________
八、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。
python
model.compile(optimizer="adam",
loss ='sparse_categorical_crossentropy',
metrics =['accuracy'])
九、训练模型
python
epochs = 20
history = model.fit(
train_ds,
validation_data=val_ds,
epochs=epochs
)
from tqdm import tqdm
import tensorflow.keras.backend as K
epochs = 10
lr = 1e-4
# 记录训练数据,方便后面的分析
history_train_loss = []
history_train_accuracy = []
history_val_loss = []
history_val_accuracy = []
for epoch in range(epochs):
train_total = len(train_ds)
val_total = len(val_ds)
"""
total:预期的迭代数目
ncols:控制进度条宽度
mininterval:进度更新最小间隔,以秒为单位(默认值:0.1)
"""
with tqdm(total=train_total, desc=f'Epoch {epoch + 1}/{epochs}',mininterval=1,ncols=100) as pbar:
lr = lr*0.92
K.set_value(model.optimizer.lr, lr)
for image,label in train_ds:
"""
训练模型,简单理解train_on_batch就是:它是比model.fit()更高级的一个用法
想详细了解 train_on_batch 的同学,
可以看看我的这篇文章:https://www.yuque.com/mingtian-fkmxf/hv4lcq/ztt4gy
"""
history = model.train_on_batch(image,label)
train_loss = history[0]
train_accuracy = history[1]
pbar.set_postfix({"loss": "%.4f"%train_loss,
"accuracy":"%.4f"%train_accuracy,
"lr": K.get_value(model.optimizer.lr)})
pbar.update(1)
history_train_loss.append(train_loss)
history_train_accuracy.append(train_accuracy)
print('开始验证!')
with tqdm(total=val_total, desc=f'Epoch {epoch + 1}/{epochs}',mininterval=0.3,ncols=100) as pbar:
for image,label in val_ds:
history = model.test_on_batch(image,label)
val_loss = history[0]
val_accuracy = history[1]
pbar.set_postfix({"loss": "%.4f"%val_loss,
"accuracy":"%.4f"%val_accuracy})
pbar.update(1)
history_val_loss.append(val_loss)
history_val_accuracy.append(val_accuracy)
print('结束验证!')
print("验证loss为:%.4f"%val_loss)
print("验证准确率为:%.4f"%val_accuracy)
运行结果:
python
Epoch 1/10: 100%|█| 340/340 [01:12<00:00, 4.72it/s, train_loss=0.5849, train_acc=0.6250, lr=9.2e-5]
开始验证!
Epoch 1/10: 100%|██████████████████| 85/85 [00:06<00:00, 12.56it/s, val_loss=0.5191, val_acc=0.6250]
Epoch 2/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.5880
验证准确率为:0.5221
Epoch 2/10: 100%|█| 340/340 [01:04<00:00, 5.30it/s, train_loss=0.0058, train_acc=1.0000, lr=8.46e-5
Epoch 2/10: 0%| | 0/85 [00:00<?, ?it/s]开始验证!
Epoch 2/10: 100%|██████████████████| 85/85 [00:06<00:00, 12.27it/s, val_loss=0.0123, val_acc=1.0000]
Epoch 3/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.3964
验证准确率为:0.9074
Epoch 3/10: 100%|█| 340/340 [01:21<00:00, 4.15it/s, train_loss=0.0024, train_acc=1.0000, lr=7.79e-5
开始验证!
Epoch 3/10: 100%|██████████████████| 85/85 [00:07<00:00, 11.98it/s, val_loss=0.0075, val_acc=1.0000]
Epoch 4/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.0712
验证准确率为:0.9676
Epoch 4/10: 100%|█| 340/340 [01:04<00:00, 5.28it/s, train_loss=0.0010, train_acc=1.0000, lr=7.16e-5
Epoch 4/10: 0%| | 0/85 [00:00<?, ?it/s]开始验证!
Epoch 4/10: 100%|██████████████████| 85/85 [00:07<00:00, 12.11it/s, val_loss=0.0009, val_acc=1.0000]
结束验证!
验证loss为:0.0746
验证准确率为:0.9706
Epoch 5/10: 100%|█| 340/340 [01:03<00:00, 5.38it/s, train_loss=0.0034, train_acc=1.0000, lr=6.59e-5
开始验证!
Epoch 5/10: 100%|██████████████████| 85/85 [00:07<00:00, 11.04it/s, val_loss=0.0029, val_acc=1.0000]
结束验证!
验证loss为:0.0350
验证准确率为:0.9897
Epoch 6/10: 100%|█| 340/340 [01:02<00:00, 5.43it/s, train_loss=0.0000, train_acc=1.0000, lr=6.06e-5
Epoch 6/10: 0%| | 0/85 [00:00<?, ?it/s]开始验证!
Epoch 6/10: 100%|██████████████████| 85/85 [00:07<00:00, 11.08it/s, val_loss=0.0009, val_acc=1.0000]
Epoch 7/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.0520
验证准确率为:0.9868
Epoch 7/10: 100%|█| 340/340 [01:21<00:00, 4.15it/s, train_loss=0.0219, train_acc=1.0000, lr=5.58e-5
开始验证!
Epoch 7/10: 100%|██████████████████| 85/85 [00:08<00:00, 10.19it/s, val_loss=0.0050, val_acc=1.0000]
Epoch 8/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.0280
验证准确率为:0.9941
Epoch 8/10: 100%|█| 340/340 [01:02<00:00, 5.43it/s, train_loss=0.0003, train_acc=1.0000, lr=5.13e-5
开始验证!
Epoch 8/10: 100%|██████████████████| 85/85 [00:07<00:00, 11.22it/s, val_loss=0.0013, val_acc=1.0000]
结束验证!
验证loss为:0.0374
验证准确率为:0.9868
Epoch 9/10: 100%|█| 340/340 [01:02<00:00, 5.44it/s, train_loss=0.0004, train_acc=1.0000, lr=4.72e-5
Epoch 9/10: 0%| | 0/85 [00:00<?, ?it/s]开始验证!
Epoch 9/10: 100%|██████████████████| 85/85 [00:07<00:00, 11.24it/s, val_loss=0.0002, val_acc=1.0000]
Epoch 10/10: 0%| | 0/340 [00:00<?, ?it/s]结束验证!
验证loss为:0.0995
验证准确率为:0.9750
Epoch 10/10: 100%|█| 340/340 [01:22<00:00, 4.15it/s, train_loss=0.0001, train_acc=1.0000, lr=4.34e-
Epoch 10/10: 0%| | 0/85 [00:00<?, ?it/s]开始验证!
Epoch 10/10: 100%|█████████████████| 85/85 [00:08<00:00, 10.36it/s, val_loss=0.0002, val_acc=1.0000]
结束验证!
验证loss为:0.0219
验证准确率为:0.9941
十、模型评估
python
epochs_range = range(epochs)
plt.figure(figsize=(12, 4))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, history_train_accuracy, label='Training Accuracy')
plt.plot(epochs_range, history_val_accuracy, label='Validation Accuracy')
plt.legend(loc='lower right')
plt.title('Training and Validation Accuracy')
plt.subplot(1, 2, 2)
plt.plot(epochs_range, history_train_loss, label='Training Loss')
plt.plot(epochs_range, history_val_loss, label='Validation Loss')
plt.legend(loc='upper right')
plt.title('Training and Validation Loss')
plt.show()
十一、预测
python
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")
运行结果:
根据代码中bug,修改后运行结果如下:
十二、总结
本周通过tenserflow框架创建VGG16网络模型进行猫狗识别:
VGG16模型结构:
VGG16共包含:
13个卷积层(Convolutional Layer),分别用conv3-XXX表示 (XXX为输出通道数,3代表kernel_size)
3个全连接层(Fully connected Layer),分别用FC-XXXX表示(XXX为输出神经元个数)
5个池化层(Pool layer),分别用maxpool表示。
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。
tqdm
是一个强大的工具,它简单易用,高度可定制,适合于各种循环任务,特别是在数据处理和机器学习领域中。通过使用tqdm
,开发者可以提供更好的用户体验,准确地展示程序的执行进度。
tqdm的基本特性如下所述:
- 易用性:tqdm的使用非常简单,通常只需在循环的迭代器上添加tqdm()。只需在 Python 循环中包裹你的迭代器,一行代码就能产生一个精美的进度条。
- 灵活性:兼容广泛的迭代环境,包括列表、文件、生成器等。它可以和 for 循环、pandas dataframe的 apply 函数以及 Python 的 map 函数等等配合使用。
- 高效性:对代码的运行效率影响极小。tqdm 使用了智能算法,即使在数据流非常快的情况下,也不会拖慢你的代码速度。
- 可定制性:允许用户自定义进度条的各种属性,如进度条长度、格式等。