猫狗数据集95%的准确率，tensorflow实现

猫狗数据集下载DATA_CAT | Kaggle

有25000张图片

导入相关的库

import warnings
import matplotlib.pyplot as plt
import numpy as np
import os
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import layers
warnings.filterwarnings("ignore")

导入数据

PATH = '/kaggle/input/data-cat'

train_dir = os.path.join(PATH, 'train')
validation_dir = os.path.join(PATH, 'val')

BATCH_SIZE = 32
IMG_SIZE = (160, 160)

train_ds = tf.keras.utils.image_dataset_from_directory(train_dir,
                                                        shuffle=True,
                                                        batch_size=BATCH_SIZE,
                                                        image_size=IMG_SIZE)

val_ds = tf.keras.utils.image_dataset_from_directory(validation_dir,
                                                     shuffle=True,
                                                     batch_size=BATCH_SIZE,
                                                     image_size=IMG_SIZE)

数据增强

data_augmentation = keras.Sequential(
    [
        layers.RandomFlip("horizontal"),
        layers.RandomRotation(0.1),
    ]
)

建立模型

inputs = keras.Input(shape=(160, 160, 3))
x = data_augmentation(inputs)
x = layers.Rescaling(1./255)(x)

x = layers.Conv2D(filters=32, kernel_size=3, padding='same', activation="relu")(x)
x = layers.MaxPooling2D(pool_size=2)(x)

x = layers.Conv2D(filters=64, kernel_size=3, padding='same', activation="relu")(x)
x = layers.MaxPooling2D(pool_size=2)(x)

x = layers.Conv2D(filters=128, kernel_size=3, padding='same', activation="relu")(x)
x = layers.MaxPooling2D(pool_size=2)(x)

x = layers.Conv2D(filters=256, kernel_size=3, padding='same', activation="relu")(x)
x = layers.MaxPooling2D(pool_size=2)(x)

x = layers.Conv2D(filters=256, kernel_size=3, padding='same', activation="relu")(x)
x = layers.MaxPooling2D(pool_size=2)(x)

x = layers.GlobalAveragePooling2D()(x)
x = layers.Dropout(0.5)(x)
outputs = layers.Dense(1, activation="sigmoid")(x)
model = keras.Model(inputs=inputs, outputs=outputs)
keras.utils.plot_model(model, show_shapes=True, to_file='model.png')

 训练模型

initial_epochs = 50
callbacks = [
    keras.callbacks.ModelCheckpoint("save_at_{epoch}.h5"),
]
model.compile(
    optimizer=keras.optimizers.Adam(1e-3),
    loss="binary_crossentropy",
    metrics=["accuracy"],
)
history = model.fit(
    train_ds, epochs=initial_epochs, callbacks=callbacks, validation_data=val_ds,
)

Epoch 1/50

2022-09-24 05:12:23.783195: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:185] None of the MLIR Optimization Passes are enabled (registered 2)
2022-09-24 05:12:25.882535: I tensorflow/stream_executor/cuda/cuda_dnn.cc:369] Loaded cuDNN version 8005

625/625 [==============================] - 93s 135ms/step - loss: 0.6731 - accuracy: 0.5711 - val_loss: 0.6144 - val_accuracy: 0.6706
Epoch 2/50
625/625 [==============================] - 33s 53ms/step - loss: 0.6203 - accuracy: 0.6612 - val_loss: 0.6539 - val_accuracy: 0.6168
Epoch 3/50
625/625 [==============================] - 34s 53ms/step - loss: 0.5829 - accuracy: 0.6982 - val_loss: 0.5381 - val_accuracy: 0.7266
Epoch 4/50
625/625 [==============================] - 34s 53ms/step - loss: 0.5478 - accuracy: 0.7246 - val_loss: 0.4862 - val_accuracy: 0.7676
Epoch 5/50
625/625 [==============================] - 34s 53ms/step - loss: 0.5065 - accuracy: 0.7563 - val_loss: 0.4445 - val_accuracy: 0.7980
Epoch 6/50
625/625 [==============================] - 33s 53ms/step - loss: 0.4601 - accuracy: 0.7865 - val_loss: 0.4467 - val_accuracy: 0.7922
Epoch 7/50
625/625 [==============================] - 33s 53ms/step - loss: 0.4237 - accuracy: 0.8073 - val_loss: 0.3804 - val_accuracy: 0.8352

···

Epoch 45/50
625/625 [==============================] - 33s 53ms/step - loss: 0.1145 - accuracy: 0.9541 - val_loss: 0.1895 - val_accuracy: 0.9440
Epoch 46/50
625/625 [==============================] - 33s 52ms/step - loss: 0.1170 - accuracy: 0.9546 - val_loss: 0.1676 - val_accuracy: 0.9402
Epoch 47/50
625/625 [==============================] - 33s 52ms/step - loss: 0.1092 - accuracy: 0.9567 - val_loss: 0.1700 - val_accuracy: 0.9414
Epoch 48/50
625/625 [==============================] - 33s 52ms/step - loss: 0.1123 - accuracy: 0.9555 - val_loss: 0.1478 - val_accuracy: 0.9502
Epoch 49/50
625/625 [==============================] - 33s 53ms/step - loss: 0.1080 - accuracy: 0.9584 - val_loss: 0.1675 - val_accuracy: 0.9472
Epoch 50/50
625/625 [==============================] - 33s 52ms/step - loss: 0.1045 - accuracy: 0.9575 - val_loss: 0.1585 - val_accuracy: 0.9500

acc = history.history['accuracy']
val_acc = history.history['val_accuracy']

loss = history.history['loss']
val_loss = history.history['val_loss']

plt.figure(figsize=(8, 8))
plt.subplot(2, 1, 1)
plt.plot(acc, "bo", label='Training Accuracy')
plt.plot(val_acc, "b", label='Validation Accuracy')
plt.legend(loc='lower right')
plt.ylabel('Accuracy')
plt.ylim([min(plt.ylim()),1])
plt.title('Training and Validation Accuracy')

plt.subplot(2, 1, 2)
plt.plot(loss, "bo", label='Training Loss')
plt.plot(val_loss, "b", label='Validation Loss')
plt.legend(loc='upper right')
plt.ylabel('Cross Entropy')
plt.ylim([0,1.0])
plt.title('Training and Validation Loss')
plt.xlabel('epoch')
plt.savefig("A.png", dpi=900)
plt.show()