import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Dense, Conv2D, Flatten, Dropout, MaxPooling2D
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import os, sys
import numpy as np
import matplotlib.pyplot as plt
print(tf.__version__)
print(np.__version__)
print(sys.version)
2.10.0 1.21.5 3.9.12 (main, Apr 4 2022, 05:22:27) [MSC v.1916 64 bit (AMD64)]
dat_URL = 'https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip'
path_to_zip = tf.keras.utils.get_file('cats_and_dogs.zip',
origin=dat_URL, extract=True)
PATH = os.path.join(os.path.dirname(path_to_zip), 'cats_and_dogs_filtered')
print(PATH)
C:\Users\totofriend\.keras\datasets\cats_and_dogs_filtered
## 경로 지정
train_dir = os.path.join(PATH, 'train')
val_dir = os.path.join(PATH, 'validation')
train_cats_dir = os.path.join(train_dir, 'cats') # directory with our training cat pictures
train_dogs_dir = os.path.join(train_dir, 'dogs') # directory with our training dog pictures
val_cats_dir = os.path.join(val_dir, 'cats') # directory with our validation cat pictures
val_dogs_dir = os.path.join(val_dir, 'dogs') # directory with our validation dog pictures
## 경로에 이미지 데이터의 개수
num_cats_tr = len(os.listdir(train_cats_dir))
num_dogs_tr = len(os.listdir(train_dogs_dir))
num_cats_val = len(os.listdir(val_cats_dir))
num_dogs_val = len(os.listdir(val_dogs_dir))
total_train = num_cats_tr + num_dogs_tr
total_val = num_cats_val + num_dogs_val
print('total training cat images:', num_cats_tr)
print('total training dog images:', num_dogs_tr)
print("--")
print('total validation cat images:', num_cats_val)
print('total validation dog images:', num_dogs_val)
print("--")
print("Total training images:", total_train)
print("Total validation images:", total_val)
total training cat images: 1000 total training dog images: 1000 -- total validation cat images: 500 total validation dog images: 500 -- Total training images: 2000 Total validation images: 1000
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
# 'dog.838.jpg', 'dog.967.jpg', 'dog.126.jpg', 'dog.102.jpg', 'dog.563.jpg', 'dog.214.jpg']
plt.figure(figsize=(10, 10))
img_idx = [838, 967, 126, 102, 563, 214]
cnt = 1
for i in img_idx:
imgpath = PATH + "/train/dogs/dog." + str(i) + ".jpg"
img = mpimg.imread(imgpath)
print(img.shape)
plt.subplot(3,2,cnt)
plt.imshow(img)
cnt += 1
(458, 499, 3) (499, 375, 3) (240, 319, 3) (373, 499, 3) (339, 305, 3) (335, 448, 3)
batch_size = 20
epochs = 15
IMG_HEIGHT = 150
IMG_WIDTH = 150
# 학습용 데이터에 대한 제너레이터(Generator)
train_img_generator = ImageDataGenerator(rescale=1./255)
# 검증용 데이터에 대한 제너레이터(Generator)
val_img_generator = ImageDataGenerator(rescale=1./255)
train_data_gen = train_img_generator.flow_from_directory(batch_size=batch_size,
directory=train_dir,
shuffle=True,
target_size=(IMG_HEIGHT, IMG_WIDTH),
class_mode='binary')
val_data_gen = val_img_generator.flow_from_directory(batch_size=batch_size,
directory=val_dir,
target_size=(IMG_HEIGHT, IMG_WIDTH),
class_mode='binary')
Found 2000 images belonging to 2 classes. Found 1000 images belonging to 2 classes.
model = Sequential([
Conv2D(16, 3, padding='same', activation='relu', input_shape=(IMG_HEIGHT, IMG_WIDTH ,3)),
MaxPooling2D(),
Conv2D(32, 3, padding='same', activation='relu'),
MaxPooling2D(),
Conv2D(64, 3, padding='same', activation='relu'),
MaxPooling2D(),
Flatten(),
Dense(512, activation='relu'),
Dense(1, activation='sigmoid')
])
model.compile(optimizer='adam',
loss='binary_crossentropy',
metrics=['accuracy'])
model.summary()
Model: "sequential"
_________________________________________________________________
Layer (type) Output Shape Param #
=================================================================
conv2d (Conv2D) (None, 150, 150, 16) 448
max_pooling2d (MaxPooling2D (None, 75, 75, 16) 0
)
conv2d_1 (Conv2D) (None, 75, 75, 32) 4640
max_pooling2d_1 (MaxPooling (None, 37, 37, 32) 0
2D)
conv2d_2 (Conv2D) (None, 37, 37, 64) 18496
max_pooling2d_2 (MaxPooling (None, 18, 18, 64) 0
2D)
flatten (Flatten) (None, 20736) 0
dense (Dense) (None, 512) 10617344
dense_1 (Dense) (None, 1) 513
=================================================================
Total params: 10,641,441
Trainable params: 10,641,441
Non-trainable params: 0
_________________________________________________________________
print("학습용 데이터 : ", total_train)
print("검증용 데이터 : ", total_val)
print("1 epoch(train) : ", total_train // batch_size)
print("1 epoch(val) : ", total_val // batch_size)
print("batch size : ", batch_size)
print("epochs = ", epochs)
학습용 데이터 : 2000 검증용 데이터 : 1000 1 epoch(train) : 100 1 epoch(val) : 50 batch size : 20 epochs = 15
%%time
### 모델 학습
history = model.fit( # model.fit_generator(
train_data_gen, # 학습용 데이터
steps_per_epoch=total_train // batch_size,
epochs=epochs,
validation_data=val_data_gen, # 검증 데이터
validation_steps=total_val // batch_size # 검증 데이터 스텝
)
Epoch 1/15 100/100 [==============================] - 26s 255ms/step - loss: 0.8062 - accuracy: 0.5260 - val_loss: 0.6593 - val_accuracy: 0.6440 Epoch 2/15 100/100 [==============================] - 24s 243ms/step - loss: 0.6501 - accuracy: 0.6355 - val_loss: 0.6540 - val_accuracy: 0.5810 Epoch 3/15 100/100 [==============================] - 24s 236ms/step - loss: 0.5868 - accuracy: 0.6965 - val_loss: 0.5784 - val_accuracy: 0.7140 Epoch 4/15 100/100 [==============================] - 24s 238ms/step - loss: 0.5056 - accuracy: 0.7480 - val_loss: 0.5702 - val_accuracy: 0.7220 Epoch 5/15 100/100 [==============================] - 24s 237ms/step - loss: 0.4187 - accuracy: 0.8090 - val_loss: 0.5661 - val_accuracy: 0.7270 Epoch 6/15 100/100 [==============================] - 24s 238ms/step - loss: 0.3423 - accuracy: 0.8535 - val_loss: 0.5924 - val_accuracy: 0.7320 Epoch 7/15 100/100 [==============================] - 24s 238ms/step - loss: 0.2424 - accuracy: 0.8995 - val_loss: 0.6806 - val_accuracy: 0.7370 Epoch 8/15 100/100 [==============================] - 24s 238ms/step - loss: 0.1627 - accuracy: 0.9395 - val_loss: 0.7745 - val_accuracy: 0.7340 Epoch 9/15 100/100 [==============================] - 24s 239ms/step - loss: 0.0973 - accuracy: 0.9685 - val_loss: 0.9576 - val_accuracy: 0.7330 Epoch 10/15 100/100 [==============================] - 24s 236ms/step - loss: 0.0772 - accuracy: 0.9760 - val_loss: 1.0749 - val_accuracy: 0.6850 Epoch 11/15 100/100 [==============================] - 24s 237ms/step - loss: 0.0486 - accuracy: 0.9840 - val_loss: 1.3200 - val_accuracy: 0.7110 Epoch 12/15 100/100 [==============================] - 24s 237ms/step - loss: 0.0303 - accuracy: 0.9920 - val_loss: 1.3180 - val_accuracy: 0.7090 Epoch 13/15 100/100 [==============================] - 24s 235ms/step - loss: 0.0105 - accuracy: 0.9990 - val_loss: 1.5725 - val_accuracy: 0.7240 Epoch 14/15 100/100 [==============================] - 24s 235ms/step - loss: 0.0023 - accuracy: 1.0000 - val_loss: 1.7049 - val_accuracy: 0.7280 Epoch 15/15 100/100 [==============================] - 24s 236ms/step - loss: 9.1471e-04 - accuracy: 1.0000 - val_loss: 1.7873 - val_accuracy: 0.7310 CPU times: total: 52min 19s Wall time: 5min 58s
model.evaluate(val_data_gen)
50/50 [==============================] - 3s 64ms/step - loss: 1.7873 - accuracy: 0.7310
[1.7873262166976929, 0.7310000061988831]
model.save('cats_and_dogs_small_1.h5')
### 결과 시각화
acc = history.history['accuracy']
loss = history.history['loss']
val_acc = history.history['val_accuracy']
val_loss = history.history['val_loss']
epochs_range = range(epochs)
plt.figure(figsize=(12, 8))
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()
from tensorflow.keras.models import load_model
model2 = load_model("cats_and_dogs_small_1.h5")
model2
<keras.engine.sequential.Sequential at 0x24895257c40>
epochs = 5
%%time
### 모델 학습
hist = model2.fit( # model.fit_generator(
train_data_gen, # 학습용 데이터
steps_per_epoch=total_train // batch_size,
epochs=5,
validation_data=val_data_gen, # 검증 데이터
validation_steps=total_val // batch_size # 검증 데이터 스텝
)
Epoch 1/5 100/100 [==============================] - 25s 245ms/step - loss: 0.0059 - accuracy: 0.9990 - val_loss: 1.5655 - val_accuracy: 0.7380 Epoch 2/5 100/100 [==============================] - 26s 261ms/step - loss: 0.0015 - accuracy: 1.0000 - val_loss: 1.6401 - val_accuracy: 0.7450 Epoch 3/5 100/100 [==============================] - 25s 248ms/step - loss: 4.0299e-04 - accuracy: 1.0000 - val_loss: 1.7490 - val_accuracy: 0.7450 Epoch 4/5 100/100 [==============================] - 24s 243ms/step - loss: 2.3251e-04 - accuracy: 1.0000 - val_loss: 1.7873 - val_accuracy: 0.7470 Epoch 5/5 100/100 [==============================] - 24s 245ms/step - loss: 1.6189e-04 - accuracy: 1.0000 - val_loss: 1.8173 - val_accuracy: 0.7480 CPU times: total: 17min 32s Wall time: 2min 4s
### 결과 시각화
acc = hist.history['accuracy']
loss = hist.history['loss']
val_acc = hist.history['val_accuracy']
val_loss = hist.history['val_loss']
epochs_range = range(epochs)
plt.figure(figsize=(12, 8))
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()
