第T10周：数据增强-海口c网

第T10周：数据增强

article/2025/7/18 10:35:00

import matplotlib.pyplot as plt
import numpy as np
#隐藏警告
import warnings
warnings.filterwarnings('ignore')from tensorflow.keras import layers
import tensorflow as tf
gpus = tf.config.list_physical_devices("GPU")if gpus:tf.config.experimental.set_memory_growth(gpus[0], True)  #设置GPU显存用量按需使用tf.config.set_visible_devices([gpus[0]],"GPU")# 打印显卡信息，确认GPU可用
print(gpus)

data_dir   = "./34-data/"
img_height = 224
img_width  = 224
batch_size = 32train_ds = tf.keras.preprocessing.image_dataset_from_directory(data_dir,validation_split=0.3,subset="training",seed=12,image_size=(img_height, img_width),batch_size=batch_size)

Found 600 files belonging to 2 classes.
Using 420 files for training.

val_ds = tf.keras.preprocessing.image_dataset_from_directory(data_dir,validation_split=0.3,subset="validation",seed=12,image_size=(img_height, img_width),batch_size=batch_size)

Found 600 files belonging to 2 classes.
Using 180 files for validation.

val_batches = tf.data.experimental.cardinality(val_ds)
test_ds     = val_ds.take(val_batches // 5)
val_ds      = val_ds.skip(val_batches // 5)print('Number of validation batches: %d' % tf.data.experimental.cardinality(val_ds))
print('Number of test batches: %d' % tf.data.experimental.cardinality(test_ds))

Number of validation batches: 5
Number of test batches: 1

class_names = train_ds.class_names
print(class_names)

['cat', 'dog']

AUTOTUNE = tf.data.AUTOTUNEdef 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)
test_ds  = test_ds.map(preprocess_image, num_parallel_calls=AUTOTUNE)train_ds = train_ds.cache().prefetch(buffer_size=AUTOTUNE)
val_ds   = val_ds.cache().prefetch(buffer_size=AUTOTUNE)

plt.figure(figsize=(15, 10))  # 图形的宽为15高为10for 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")

from tensorflow.keras import layers
from tensorflow.keras.models import Sequentialdata_augmentation = Sequential([layers.Rescaling(1./255),layers.RandomFlip("horizontal_and_vertical"),layers.RandomRotation(0.2),
])# Add the image to a batch.
image = tf.expand_dims(images[i], 0)plt.figure(figsize=(8, 8))
for i in range(9):augmented_image = data_augmentation(image)ax = plt.subplot(3, 3, i + 1)plt.imshow(augmented_image[0])plt.axis("off")

model = tf.keras.Sequential([data_augmentation,layers.Conv2D(16, 3, padding='same', activation='relu'),layers.MaxPooling2D(),
])model = tf.keras.Sequential([layers.Conv2D(16, 3, padding='same', activation='relu'),layers.MaxPooling2D(),layers.Conv2D(32, 3, padding='same', activation='relu'),layers.MaxPooling2D(),layers.Conv2D(64, 3, padding='same', activation='relu'),layers.MaxPooling2D(),layers.Flatten(),layers.Dense(128, activation='relu'),layers.Dense(len(class_names))
])model.compile(optimizer='adam',loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),metrics=['accuracy'])epochs=20
history = model.fit(train_ds,validation_data=val_ds,epochs=epochs
)

Epoch 1/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 7s 221ms/step - accuracy: 0.5332 - loss: 1.4747 - val_accuracy: 0.6149 - val_loss: 0.6562
Epoch 2/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 186ms/step - accuracy: 0.7401 - loss: 0.6087 - val_accuracy: 0.8243 - val_loss: 0.4336
Epoch 3/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 189ms/step - accuracy: 0.8792 - loss: 0.3209 - val_accuracy: 0.8311 - val_loss: 0.4042
Epoch 4/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 185ms/step - accuracy: 0.9527 - loss: 0.1470 - val_accuracy: 0.8851 - val_loss: 0.2969
Epoch 5/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 183ms/step - accuracy: 0.9618 - loss: 0.0960 - val_accuracy: 0.8919 - val_loss: 0.2645
Epoch 6/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 177ms/step - accuracy: 0.9918 - loss: 0.0439 - val_accuracy: 0.8919 - val_loss: 0.2542
Epoch 7/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 2s 177ms/step - accuracy: 1.0000 - loss: 0.0131 - val_accuracy: 0.8986 - val_loss: 0.3860
Epoch 8/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 192ms/step - accuracy: 1.0000 - loss: 0.0125 - val_accuracy: 0.8784 - val_loss: 0.6349
Epoch 9/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 187ms/step - accuracy: 1.0000 - loss: 0.0189 - val_accuracy: 0.8446 - val_loss: 0.6983
Epoch 10/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 179ms/step - accuracy: 0.9861 - loss: 0.0564 - val_accuracy: 0.8176 - val_loss: 0.6384
Epoch 11/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 2s 178ms/step - accuracy: 0.9861 - loss: 0.0658 - val_accuracy: 0.8581 - val_loss: 0.6085
Epoch 12/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 184ms/step - accuracy: 0.9987 - loss: 0.0110 - val_accuracy: 0.8716 - val_loss: 0.6948
Epoch 13/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 179ms/step - accuracy: 0.9956 - loss: 0.0157 - val_accuracy: 0.9189 - val_loss: 0.2913
Epoch 14/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 2s 177ms/step - accuracy: 0.9948 - loss: 0.0129 - val_accuracy: 0.8851 - val_loss: 0.3093
Epoch 15/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 2s 176ms/step - accuracy: 0.9982 - loss: 0.0141 - val_accuracy: 0.8716 - val_loss: 0.3558
Epoch 16/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 181ms/step - accuracy: 0.9963 - loss: 0.0087 - val_accuracy: 0.8784 - val_loss: 0.4718
Epoch 17/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 178ms/step - accuracy: 0.9946 - loss: 0.0170 - val_accuracy: 0.8986 - val_loss: 0.3190
Epoch 18/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 185ms/step - accuracy: 1.0000 - loss: 0.0042 - val_accuracy: 0.9257 - val_loss: 0.3658
Epoch 19/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 189ms/step - accuracy: 1.0000 - loss: 8.3204e-04 - val_accuracy: 0.9122 - val_loss: 0.3734
Epoch 20/20
14/14 ━━━━━━━━━━━━━━━━━━━━ 3s 191ms/step - accuracy: 1.0000 - loss: 5.7922e-04 - val_accuracy: 0.9054 - val_loss: 0.3889

loss, acc = model.evaluate(test_ds)
print("Accuracy", acc)

1/1 ━━━━━━━━━━━━━━━━━━━━ 0s 135ms/step - accuracy: 0.9062 - loss: 0.3615
Accuracy 0.90625

import random
# 这是大家可以自由发挥的一个地方
def aug_img(image):seed = (random.randint(0,9), 0)# 随机改变图像对比度stateless_random_brightness = tf.image.stateless_random_contrast(image, lower=0.1, upper=1.0, seed=seed)return stateless_random_brightness

image = tf.expand_dims(images[3]*255, 0)
print("Min and max pixel values:", image.numpy().min(), image.numpy().max())

Min and max pixel values: 0.0 255.0

plt.figure(figsize=(8, 8))
for i in range(9):augmented_image = aug_img(image)ax = plt.subplot(3, 3, i + 1)plt.imshow(augmented_image[0].numpy().astype("uint8"))plt.axis("off")

收获：由于tf版本较新在数据增强处进行代码修改代码如下：

from tensorflow.keras import layers
from tensorflow.keras.models import Sequential

data_augmentation = Sequential([
layers.Rescaling(1./255),
layers.RandomFlip("horizontal_and_vertical"),
layers.RandomRotation(0.2),
]) 学会了根据所导入库的版本对代码进行修改