- 🍨 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
设置GPU
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")
导入数据
import os,PIL,pathlib
import matplotlib.pyplot as plt
import numpy as np
from tensorflow import keras
from tensorflow.keras import layers,models
data_dir = r"C:\Users\11054\Desktop\kLearning\p1_data"
data_dir = pathlib.Path(data_dir)
image_count = len(list(data_dir.glob('*/*.jpg')))
print("图片总数为：",image_count)
图片总数为： 1125
roses = list(data_dir.glob('sunrise/*.jpg'))
PIL.Image.open(str(roses[0]))
加载数据
batch_size = 32
img_height = 180
img_width = 180
"""
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)
Found 1125 files belonging to 4 classes.
Using 900 files for training.
"""
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)
Found 1125 files belonging to 4 classes.
Using 225 files for validation.
class_names = train_ds.class_names
print(class_names)
['cloudy', 'rain', 'shine', 'sunrise']
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")
检查数据
for image_batch, labels_batch in train_ds:
print(image_batch.shape)
print(labels_batch.shape)
break
(32, 180, 180, 3)
(32,)
配置数据集
AUTOTUNE = tf.data.AUTOTUNE
train_ds = train_ds.cache().shuffle(1000).prefetch(buffer_size=AUTOTUNE)
val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)
构建网络
num_classes = 4
"""
关于卷积核的计算不懂的可以参考文章：https://blog.csdn.net/qq_38251616/article/details/114278995
layers.Dropout(0.4) 作用是防止过拟合，提高模型的泛化能力。
在上一篇文章花朵识别中，训练准确率与验证准确率相差巨大就是由于模型过拟合导致的
关于Dropout层的更多介绍可以参考文章：https://mtyjkh.blog.csdn.net/article/details/115826689
"""
model = models.Sequential([
layers.Conv2D(16, (3, 3), activation='relu', input_shape=(img_height, img_width, 3)), # 卷积层1，卷积核3*3
layers.AveragePooling2D((2, 2)), # 池化层1，2*2采样
layers.Conv2D(32, (3, 3), activation='relu'), # 卷积层2，卷积核3*3
layers.AveragePooling2D((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.Dropout(0.3), # 让神经元以一定的概率停止工作，防止过拟合，提高模型的泛化能力。
layers.Flatten(), # Flatten层，连接卷积层与全连接层
layers.Dense(128, activation='relu'), # 全连接层，特征进一步提取
layers.Dense(num_classes) # 输出层，输出预期结果
])
model.summary() # 打印网络结构
C:\Users\11054\.conda\envs\py311\Lib\site-packages\keras\src\layers\convolutional\base_conv.py:107: UserWarning: Do not pass an `input_shape`/`input_dim` argument to a layer. When using Sequential models, prefer using an `Input(shape)` object as the first layer in the model instead.
super().__init__(activity_regularizer=activity_regularizer, **kwargs)
Model: "sequential"
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┳━━━━━━━━━━━━━━━━━┓ ┃ Layer (type) ┃ Output Shape ┃ Param # ┃ ┡━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━━━━━━━━━━━━━╇━━━━━━━━━━━━━━━━━┩ │ conv2d (Conv2D) │ (None, 178, 178, 16) │ 448 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ average_pooling2d (AveragePooling2D) │ (None, 89, 89, 16) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ conv2d_1 (Conv2D) │ (None, 87, 87, 32) │ 4,640 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ average_pooling2d_1 │ (None, 43, 43, 32) │ 0 │ │ (AveragePooling2D) │ │ │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ conv2d_2 (Conv2D) │ (None, 41, 41, 64) │ 18,496 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dropout (Dropout) │ (None, 41, 41, 64) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ flatten (Flatten) │ (None, 107584) │ 0 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense (Dense) │ (None, 128) │ 13,770,880 │ ├──────────────────────────────────────┼─────────────────────────────┼─────────────────┤ │ dense_1 (Dense) │ (None, 4) │ 516 │ └──────────────────────────────────────┴─────────────────────────────┴─────────────────┘
Total params: 13,794,980 (52.62 MB)
Trainable params: 13,794,980 (52.62 MB)
Non-trainable params: 0 (0.00 B)
编译
# 设置优化器
opt = tf.keras.optimizers.Adam(learning_rate=0.001)
model.compile(optimizer=opt,
loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
metrics=['accuracy'])
训练模型
epochs = 10
history = model.fit(
train_ds,
validation_data=val_ds,
epochs=epochs
)
Epoch 1/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m6s[0m 152ms/step - accuracy: 0.4130 - loss: 594.6376 - val_accuracy: 0.6444 - val_loss: 0.8201
Epoch 2/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 144ms/step - accuracy: 0.7569 - loss: 0.7675 - val_accuracy: 0.6889 - val_loss: 0.7538
Epoch 3/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 142ms/step - accuracy: 0.9237 - loss: 0.3134 - val_accuracy: 0.7733 - val_loss: 0.6527
Epoch 4/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 142ms/step - accuracy: 0.9228 - loss: 0.2068 - val_accuracy: 0.7644 - val_loss: 0.8386
Epoch 5/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 143ms/step - accuracy: 0.9287 - loss: 0.2245 - val_accuracy: 0.8044 - val_loss: 0.6802
Epoch 6/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 141ms/step - accuracy: 0.9833 - loss: 0.0794 - val_accuracy: 0.8000 - val_loss: 0.7839
Epoch 7/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 143ms/step - accuracy: 0.9932 - loss: 0.0422 - val_accuracy: 0.7778 - val_loss: 0.9634
Epoch 8/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 142ms/step - accuracy: 0.9816 - loss: 0.1061 - val_accuracy: 0.8089 - val_loss: 0.9127
Epoch 9/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 143ms/step - accuracy: 0.9967 - loss: 0.0190 - val_accuracy: 0.8044 - val_loss: 1.1308
Epoch 10/10
[1m29/29[0m [32m━━━━━━━━━━━━━━━━━━━━[0m[37m[0m [1m4s[0m 146ms/step - accuracy: 0.9934 - loss: 0.0234 - val_accuracy: 0.8178 - val_loss: 0.9809
模型评估
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()
个人小结
运行时遇到错误：
AttributeError: module ‘keras._tf_keras.keras.layers’ has no attribute ‘experimental’
将对应的内容 layers.experimental.preprocessing.Rescaling 删除后运行成功，Rescaling 层用于将输入图像的像素值从 [0, 255] 归一化到 [0, 1]，可以用如下操作代替归一化
import tensorflow as tf
from tensorflow.keras import layers
def mean_std_normalize(image):
return (image - 127.5) / 127.5
train_ds = train_ds.map(lambda x, y: (mean_std_normalize(x), y))
val_ds = val_ds.map(lambda x, y: (mean_std_normalize(x), y))
5118
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
