0420 AWS_cat_dog_small_CNN_구현 및 문제점 해결

1. 일부 이미지 분리 (총 4000개)

import os, shutil

original_dataset_dir = './data/cat_dog/train'

## directory 생성 ##

base_dir = 'data/cat_dog_small'
os.mkdir(base_dir)

train_dir = os.path.join(base_dir,'train')
os.mkdir(train_dir)
validation_dir = os.path.join(base_dir,'validation')
os.mkdir(validation_dir)
test_dir = os.path.join(base_dir,'test')
os.mkdir(test_dir)

train_cats_dir = os.path.join(train_dir,'cats')
os.mkdir(train_cats_dir)
train_dogs_dir = os.path.join(train_dir,'dogs')
os.mkdir(train_dogs_dir)

validation_cats_dir = os.path.join(validation_dir,'cats')
os.mkdir(validation_cats_dir)
validation_dogs_dir = os.path.join(validation_dir,'dogs')
os.mkdir(validation_dogs_dir)

test_cats_dir = os.path.join(test_dir,'cats')
os.mkdir(test_cats_dir)
test_dogs_dir = os.path.join(test_dir,'dogs')
os.mkdir(test_dogs_dir)

## file 복사 ##

fnames = ['cat.{}.jpg'.format(i) for i in range(1000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(train_cats_dir, fname)
    shutil.copyfile(src,dst)

fnames = ['cat.{}.jpg'.format(i) for i in range(1000,1500)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(validation_cats_dir, fname)
    shutil.copyfile(src,dst)
    
fnames = ['cat.{}.jpg'.format(i) for i in range(1500,2000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(test_cats_dir, fname)
    shutil.copyfile(src,dst)
    

fnames = ['dog.{}.jpg'.format(i) for i in range(1000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(train_dogs_dir, fname)
    shutil.copyfile(src,dst)

fnames = ['dog.{}.jpg'.format(i) for i in range(1000,1500)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(validation_dogs_dir, fname)
    shutil.copyfile(src,dst)
    
fnames = ['dog.{}.jpg'.format(i) for i in range(1500,2000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir,fname)
    dst = os.path.join(test_dogs_dir, fname)
    shutil.copyfile(src,dst) 

2. ImageDataGenerator

import os
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import matplotlib.pyplot as plt

train_dir = './data/cat_dog_small/train'    # cats folder, dogs folder
valid_dir = './data/cat_dog_small/validation'

# ImageDataGenerator를 생성해요!
train_datagen = ImageDataGenerator(rescale=1/255)
validation_datagen = ImageDataGenerator(rescale=1/255)

train_generator = train_datagen.flow_from_directory(
    train_dir,    # target directory
    classes=['cats', 'dogs'],
    target_size=(150,150),
    batch_size=20,
    class_mode='binary'
)

validation_generator = validation_datagen.flow_from_directory(
    valid_dir,    # target directory
    classes=['cats', 'dogs'],
    target_size=(150,150),
    batch_size=20,
    class_mode='binary'
)

3. 모델 생성

import numpy as np
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D
from tensorflow.keras.layers import Flatten, Dense, Dropout
from tensorflow.keras.optimizers import Adam, RMSprop

model = Sequential()

model.add(Conv2D(filters=32,
                 kernel_size=(3,3),
                 activation='relu',
                 input_shape=(150,150,3)))

model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Conv2D(filters=64,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(Conv2D(filters=128,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Conv2D(filters=128,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(MaxPooling2D(pool_size=(2,2)))

#### feature extraction ####

model.add(Flatten())

model.add(Dense(units=256,
                activation='relu'))

model.add(Dense(units=1,
                activation='sigmoid'))

print(model.summary())

4. 모델 학습 및 저장

model.compile(optimizer=Adam(learning_rate=1e-4),
              loss='binary_crossentropy',
              metrics=['accuracy'])

history = model.fit(train_generator,    # 2000개 이미지 , 20개씩 뽑아내.. 100번
                   steps_per_epoch=100,
                   epochs=30,
                   validation_data=validation_generator,  # 1000개 이미지 , 20개
                   validation_steps=50)

model.save('./data/cats_dogs_small_cnn_model.h5')

5. 그래프

import matplotlib.pyplot as plt

train_acc = history.history['accuracy']
valid_acc = history.history['val_accuracy']

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

figure = plt.figure()
ax1 = figure.add_subplot(1,2,1)
ax2 = figure.add_subplot(1,2,2)

ax1.plot(train_acc, color='r', label='train accuracy')
ax1.plot(valid_acc, color='b', label='valid accuracy')
ax1.legend()

ax2.plot(train_loss, color='r', label='train loss')
ax2.plot(valid_loss, color='b', label='valid loss')
ax2.legend()

plt.tight_layout()
plt.show()

1. Image Augmentation

1. 이미지 증식 방법 (이미지 파일 지정)

from tensorflow.keras.preprocessing import image
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import matplotlib.pyplot as plt

# train_datagen = ImageDataGenerator(rescale=1/255)

# 이미지 증식하는 ImageDataGenerator
datagen = ImageDataGenerator(rotation_range=20,    # 최대 20도내에서 rotation
                             width_sahift_range=0.1,  # 옆으로 10%내로 움직임
                             height_shift_range=0.1, # 위아래로 10%내로 움직임
                             zoom_range=0.1,         # 10%내로 확대, 축소
                             horizontal_flip=True,  # 좌우반전 허용
                             vertical_flip=True,    # 상하반전 허용
                             fill_mode='nearest')
# fill_mode='nearest'
# 이미지를 rotation하면 원본 이미지가 손상될 수 있음
# 근처에 있는 값을 이용해서 pixel이 깨지는 것을 보정

img = image.load_img('./data/cat_dog_small/train/cats/cat.3.jpg',
                     target_size=(150,150))

# keras 제공 함수 : image.img_to_array()
x = image.img_to_array(img) # image data에서 pixel data(ndarray) 추출
print(type(x), x.shape)     # (150, 150, 3)

x = x.reshape((1,) + x.shape) # 4차원으로 변경
print(x.shape)                # (1, 150, 150, 3)

figure = plt.figure()
ax = []

for i in range(20):
    ax.append(figure.add_subplot(4,5,i+1))

    
# generator를 이용해야 하는데 끝도 없이 나오면 안 되므로 설정
idx = 0
# datagen.flow() : generator로부터 data를 당겨 와라
# 하나의 이미지(x)에 증식된 하나의 데이터(batch_size=1)만 가져온다.
for batch in datagen.flow(x, batch_size=1):
    # pixel data를 다시 image data로 변경해서 ax[idx]에 찍어서 subplot 생성
    ax[idx].imshow(image.array_to_img(batch[0]))
    idx += 1
    if idx == 20:
        break

plt.tight_layout()
plt.show()

2. 이미지 증식을 이용한 CNN

4000개의 이미지 학습

import os
from tensorflow.keras.preprocessing.image import ImageDataGenerator
import matplotlib.pyplot as plt

train_dir = './data/cat_dog_small/train'    # cats folder, dogs folder
valid_dir = './data/cat_dog_small/validation'

# 이미지 증식해서 ImageDataGenerator를 생성해요!
train_datagen = ImageDataGenerator(rescale=1/255,
                                   rotation_range=30,
                                   width_shift_range=0.1,
                                   height_shift_range=0.1,
                                   zoom_range=0.2,
                                   horizontal_flip=True,                             
                                   fill_mode='nearest')

validation_datagen = ImageDataGenerator(rescale=1/255) # 증식 x

train_generator = train_datagen.flow_from_directory(
    train_dir,    # target directory
    classes=['cats', 'dogs'],
    target_size=(150,150),
    batch_size=20,
    class_mode='binary'
)

validation_generator = validation_datagen.flow_from_directory(
    valid_dir,                   # target directory
    classes=['cats', 'dogs'],
    target_size=(150,150),
    batch_size=20,
    class_mode='binary'
)

import numpy as np
import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D
from tensorflow.keras.layers import Flatten, Dense, Dropout
from tensorflow.keras.optimizers import Adam, RMSprop

model = Sequential()

model.add(Conv2D(filters=32,
                 kernel_size=(3,3),
                 activation='relu',
                 input_shape=(150,150,3)))

model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Conv2D(filters=64,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(Conv2D(filters=128,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(MaxPooling2D(pool_size=(2,2)))

model.add(Conv2D(filters=128,
                 kernel_size=(3,3),
                 activation='relu'))

model.add(MaxPooling2D(pool_size=(2,2)))

#### feature extraction ####

model.add(Flatten())

model.add(Dense(units=256,
                activation='relu'))

model.add(Dense(units=1,
                activation='sigmoid'))

print(model.summary())

model.compile(optimizer=Adam(learning_rate=1e-4),
              loss='binary_crossentropy',
              metrics=['accuracy'])

history = model.fit(train_generator,    # 2000개 이미지, 20개씩 뽑아내
                   steps_per_epoch=100, # 100번
                   epochs=30,
                   validation_data=validation_generator, # 1000개 이미지, 20개씩
                   validation_steps=50) # 50번

model.save('./data/cats_dogs_small_cnn_model_augmentation.h5')

import matplotlib.pyplot as plt

train_acc = history.history['accuracy']
valid_acc = history.history['val_accuracy']

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

figure = plt.figure()
ax1 = figure.add_subplot(1,2,1)
ax2 = figure.add_subplot(1,2,2)

ax1.plot(train_acc, color='r', label='train accuracy')
ax1.plot(valid_acc, color='b', label='valid accuracy')
ax1.legend()

ax2.plot(train_loss, color='r', label='train loss')
ax2.plot(valid_loss, color='b', label='valid loss')
ax2.legend()

plt.tight_layout()
plt.show()

parameter를 조절하고 epoch 수를 늘리면 더 나은 그래프가 나올 것임

2. Transfer Learning (전이학습)

from tensorflow.keras.applications import VGG16

model_base = VGG16(weights='imagenet',  # 어떤 것으로 만들어진 전이학습인가
                   include_top=True,    # FC_Layer 포함
                   input_shape=(224,224,3))

print(model_base.summary())

1. activation map 생성

# 우리가 가지고 있는 데이터(개와 고양이)를 VGG16에 통과시켜서
# ndarray 형태로 저장

import os
import numpy as np
from tensorflow.keras.preprocessing.image import ImageDataGenerator

base_dir = './data/cat_dog_small'
train_dir = os.path.join(base_dir, 'train')      # ./data/cat_dog_small/train
valid_dir = os.path.join(base_dir, 'validation')
test_dir = os.path.join(base_dir, 'test')

datagen = ImageDataGenerator(rescale=1/255)

# (어디에서 데이터를 받아 들일 것인지, 데이터 몇 장인지)
# shape=(이미지 개수, feature extraction의 최종 형태 : 4,4,512 )
def extraction_feature(directory, sample_count):
    # 일단 0으로 채운다. => 나중에 데이터가 채워질 것임
    features = np.zeros(shape=(sample_count,4,4,512))
    labels = np.zeros(shape=(sample_count,)) # 값이 1개만 들어있는 튜플
    
    # flow : directory에서 값을 가져온다
    generator = datagen.flow_from_directory(
        directory,
        classes=['cats', 'dogs'],
        target_size=(150,150),
        batch_size=20,            # 20개씩 추출
        class_mode='binary'
    )
    

    i = 0
    
    for x_data_batch, t_data_batch in generator:
        # 전이학습을 통과해야 한다. => model_base
        # model_base.predict : FC Layer가 있으면 예측값이 나오는데 없으므로
        # feature extraction의 최종 형태(None, 4, 4, 512)가 나온다.
        # ㄴ 20개 이미지 특성추출
        feature_batch = model_base.predict(x_data_batch)
        features[i*20:(i+1)*20] = feature_batch   # [0:20] [20:40] ...
        labels[i*20:(i+1)*20] = t_data_batch
        
        i = i + 1
        
        if i*20 >= sample_count:
            break
            
    return features, labels
    
# train_features : train의 x_data (4차원)
# train_labels : train의 t_data (1차원)
# 튜플 형태로 뽑아낸다. (ndarray)
# 
train_features, train_labels = extraction_feature(train_dir, 2000)
valid_features, valid_labels = extraction_feature(valid_dir, 1000)
test_features, test_labels = extraction_feature(test_dir, 1000)

2. activation map을 이용해서 DNN 학습

train_x_data = np.reshape(train_features,(2000,4*4*512))  # 2차원
train_t_data = train_labels

valid_x_data = np.reshape(valid_features,(1000,4*4*512))  # 2차원
valid_t_data = valid_labels

test_x_data = np.reshape(test_features,(1000,4*4*512))  # 2차원
test_t_data = test_labels

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Flatten, Dense, Dropout
from tensorflow.keras.optimizers import Adam

model = Sequential()

model.add(Flatten(input_shape=(4*4*512,)))

model.add(Dense(units=256,
                activation='relu'))
model.add(Dropout(rate=0.5))

model.add(Dense(units=1,
                activation='sigmoid'))

model.compile(optimizer=Adam(learning_rate=1e-4),
              loss='binary_crossentropy',
              metrics=['accuracy'])

history = model.fit(train_x_data,
                    train_t_data,
                    epochs=30,
                    batch_size=20,
                    validation_data=(valid_x_data, valid_t_data))