文章目录
深度学习Week13——利用TensorFlow实现咖啡豆识别
一、前言
二、我的环境
三、前期工作
1、配置环境
2、导入数据
四、数据预处理
1、加载数据
2、可视化数据
3、检查数据
4、配置数据集
五、构建VGG-16模型
六、编译模型
七、训练模型
八、预测与评估
1、Accuracy图
九、拓展
一、前言
- 🍨 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
本篇内容分为两个部分,前面部分是学习K同学给的算法知识点以及复现,后半部分是自己的拓展与未解决的问题
二、我的环境
- 电脑系统:Windows 10
- 语言环境:Python 3.8.0
- 编译器:Pycharm2023.2.3
深度学习环境:TensorFlow
显卡及显存:RTX 3060 8G
三、前期工作
1、导入库并配置环境
from tensorflow import keras
from tensorflow.keras import layers,models
import os, PIL, pathlib
import matplotlib.pyplot as plt
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")
gpus
这一步与pytorch第一步类似,我们在写神经网络程序前无论是选择pytorch还是tensorflow都应该配置好gpu环境(如果有gpu的话)
2、 导入数据
导入所有好莱坞明星照片数据,依次分别为训练集图片(train_images)、训练集标签(train_labels)、测试集图片(test_images)、测试集标签(test_labels),数据集来源于K同学啊的网盘:数据集
data_dir = "E:\Deep_Learning\Data\Week7"
data_dir = pathlib.Path(data_dir)
image_count = len(list(data_dir.glob('*/*.jpg')))
print("图片总数为:",image_count)
#查看第一张图片:
roses = list(data_dir.glob('Dark/*.png'))
PIL.Image.open(str(roses[0]))
图片总数为: 1800
四、数据预处理
1、加载数据
batch_size = 32
img_height = 224
img_width = 224
使用image_dataset_from_directory
方法将磁盘中的数据加载到tf.data.Dataset
中
tf.keras.preprocessing.image_dataset_from_directory()
会将文件夹中的数据加载到tf.data.Dataset中,且加载的同时会打乱数据。
- class_names
- validation_split: 0和1之间的可选浮点数,可保留一部分数据用于验证。
- subset: training或validation之一。仅在设置validation_split时使用。
- seed: 用于shuffle和转换的可选随机种子。
- batch_size: 数据批次的大小。默认值:32
- image_size: 从磁盘读取数据后将其重新调整大小。默认:(256,256)。由于管道处理的图像批次必须具有相同的大小,因此该参数必须提供。
"""
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=123,
image_size=(img_height, img_width),
batch_size=batch_size)
输出:
Found 1200 files belonging to 4 classes.
Using 960 files for training.
- 验证集并没有参与训练过程梯度下降过程的,狭义上来讲是没有参与模型的参数训练更新的。
- 但是广义上来讲,验证集存在的意义确实参与了一个“人工调参”的过程,我们根据每一个epoch训练之后模型在valid data上的表现来决定是否需要训练进行early stop,或者根据这个过程模型的性能变化来调整模型的超参数,如学习率,batch_size等等。
- 因此,我们也可以认为,验证集也参与了训练,但是并没有使得模型去overfit验证集
"""
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=123,
image_size=(img_height, img_width),
batch_size=batch_size)
输出:
Found 1200 files belonging to 4 classes.
Using 240 files for validation.
我们可以通过class_names输出数据集的标签。标签将按字母顺序对应于目录名称。
class_names = train_ds.class_names
print(class_names)
[‘Dark’, ‘Green’, ‘Light’, ‘Medium’]
2、数据可视化
# 查看前20个图片
plt.figure(figsize=(20, 10))
for images, labels in train_ds.take(1):
for i in range(20):
ax = plt.subplot(5, 10, i + 1)
plt.imshow(images[i].numpy().astype("uint8"))
plt.title(class_names[labels[i]])
plt.axis("off")
3、再次检查数据
for image_batch, labels_batch in train_ds:
print(image_batch.shape)
print(labels_batch.shape)
break
(32, 224, 224, 3)
(32,)
Image_batch
是形状的张量(32,224,224,3)。这是一批形状224x224x3的32张图片(最后一维指的是彩色通道RGB。
Label_batch
是形状(32,)的张量,这些标签对应32张图片
4、配置数据集
shuffle():
打乱数据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
如果不使用prefetch()
,CPU 和 GPU/TPU 在大部分时间都处于空闲状态:
使用prefetch()
可显著减少空闲时间:
AUTOTUNE = tf.data.AUTOTUNE
train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)
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
五 、构建VGG-16模型
- 输入层:
输入层负责接收原始数据,将数据传递到网络中的第一层。 - 卷积层:
卷积层使用卷积核对输入数据进行滤波操作,以提取图像中的特征。 - 池化层:
池化层用于对卷积层的输出进行下采样,以减少数据的维度和计算量。 - 全连接层:
全连接层起到“特征提取器”的作用,将前面层的特征表示映射到输出层。 - 输出层:
输出层负责输出模型的预测结果。
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() # 打印模型
Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels.h5
553467096/553467096 [==============================] - 124s 0us/step
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: 138357544 (527.79 MB)
Trainable params: 138357544 (527.79 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
六、编译模型
具体函数解释参考第八周博客或者K同学啊的博客!
1.设置动态学习率
# 设置初始学习率
initial_learning_rate = 1e-4
lr_schedule = tf.keras.optimizers.schedules.ExponentialDecay(
initial_learning_rate,
decay_steps=30, # 敲黑板!!!这里是指 steps,不是指epochs
decay_rate=0.92, # lr经过一次衰减就会变成 decay_rate*lr
staircase=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'])
七、训练模型
网络越来越复杂,对算力要求也更高,使用GPU来跑都慢
epochs = 20
history = model.fit(train_ds,
validation_data=val_ds,
epochs=epochs,
callbacks=[checkpointer, earlystopper])
八、预测
1、Accuracy图与Loss图
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()
结果:
九、拓展
自己构建VGG-16模型
首先,导入了必要的模块:layers
, models
, tensorflow.keras
库的模块,用于构建神经网络模型。
其次,定义函数VGG16(nb_classes, input_shape)
,接受两个参数:类别:nb_classes
和形状 input_shape
表示输入数据的形状。
通过Input
函数创建了一个输入张量input_tensor
,形状为input_shape
按照VGG16的结构,定义卷积层和池化层:
使用 ReLU 作为激活函数,让卷积层分布在不同的block
中,每个block
中有若干个卷积层,每个block
之间有一个池化层。每个卷积层后面跟着一个池化层,通过MaxPooling2D
函数实现。
在所有卷积和池化层之后,添加了全连接层:
通过Flatten层将卷积层输出的特征图展平成一维向量。
然后通过两个Dense层定义包含4096个神经元的全连接层,每个层包含4096个神经元,并使用 ReLU 作为激活函数。
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(len(class_names), (img_width, img_height, 3))
model.summary()
输出:
Model: "model"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
input_2 (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, 4) 16388
=================================================================
Total params: 134276932 (512.23 MB)
Trainable params: 134276932 (512.23 MB)
Non-trainable params: 0 (0.00 Byte)
_________________________________________________________________
再进行训练后的结果为:
Epoch 1/20
30/30 [] - 350s 12s/step - loss: 1.2732 - accuracy: 0.3396 - val_loss: 1.0613 - val_accuracy: 0.5042
Epoch 2/20
30/30 [] - 345s 12s/step - loss: 0.7579 - accuracy: 0.6385 - val_loss: 0.7859 - val_accuracy: 0.5875
Epoch 3/20
30/30 [] - 345s 12s/step - loss: 0.5615 - accuracy: 0.7312 - val_loss: 0.5375 - val_accuracy: 0.7708
Epoch 4/20
30/30 [] - 346s 12s/step - loss: 0.4541 - accuracy: 0.8177 - val_loss: 0.2172 - val_accuracy: 0.9208
Epoch 5/20
30/30 [] - 346s 12s/step - loss: 0.2854 - accuracy: 0.8990 - val_loss: 0.2677 - val_accuracy: 0.8875
Epoch 6/20
30/30 [] - 347s 12s/step - loss: 0.1397 - accuracy: 0.9615 - val_loss: 0.1139 - val_accuracy: 0.9500
Epoch 7/20
30/30 [] - 349s 12s/step - loss: 0.0786 - accuracy: 0.9760 - val_loss: 0.1361 - val_accuracy: 0.9542
Epoch 8/20
30/30 [] - 347s 12s/step - loss: 0.1082 - accuracy: 0.9604 - val_loss: 0.1560 - val_accuracy: 0.9458
Epoch 9/20
30/30 [] - 346s 12s/step - loss: 0.1542 - accuracy: 0.9427 - val_loss: 0.0899 - val_accuracy: 0.9583
Epoch 10/20
30/30 [] - 345s 12s/step - loss: 0.0674 - accuracy: 0.9771 - val_loss: 0.1207 - val_accuracy: 0.9500
Epoch 11/20
30/30 [] - 345s 12s/step - loss: 0.0906 - accuracy: 0.9729 - val_loss: 0.0790 - val_accuracy: 0.9708
Epoch 12/20
30/30 [] - 346s 12s/step - loss: 0.1257 - accuracy: 0.9531 - val_loss: 0.1498 - val_accuracy: 0.9500
Epoch 13/20
30/30 [] - 350s 12s/step - loss: 0.0890 - accuracy: 0.9698 - val_loss: 0.0828 - val_accuracy: 0.9625
Epoch 14/20
30/30 [] - 346s 12s/step - loss: 0.0833 - accuracy: 0.9708 - val_loss: 0.0794 - val_accuracy: 0.9792
Epoch 15/20
30/30 [] - 345s 12s/step - loss: 0.0708 - accuracy: 0.9833 - val_loss: 0.0811 - val_accuracy: 0.9708
Epoch 16/20
30/30 [] - 345s 12s/step - loss: 0.0901 - accuracy: 0.9646 - val_loss: 0.2803 - val_accuracy: 0.8958
Epoch 17/20
30/30 [] - 345s 12s/step - loss: 0.1145 - accuracy: 0.9740 - val_loss: 0.1020 - val_accuracy: 0.9667
Epoch 18/20
30/30 [] - 346s 12s/step - loss: 0.1251 - accuracy: 0.9656 - val_loss: 0.0977 - val_accuracy: 0.9625
Epoch 19/20
30/30 [] - 346s 12s/step - loss: 0.0904 - accuracy: 0.9688 - val_loss: 0.2341 - val_accuracy: 0.9083
Epoch 20/20
30/30 [] - 358s 12s/step - loss: 0.1015 - accuracy: 0.9677 - val_loss: 0.0894 - val_accuracy: 0.9708