딥러닝 cat and dog 문제

cat and dog

• 개와 고양이 사진을 분류하는 캐글 문제

import matplotlib.pyplot as plt
import os
import shutil
import zipfile
 
import tensorflow as tf
import numpy as np
import pandas as pd
 
#zip파일 경로 설정
data_zip_dir = '/Users/sok/Desktop/desktop/pytest/dogs-vs-cats'
train_zip_dir = os.path.join(data_zip_dir, 'train.zip')
test_zip_dir = os.path.join(data_zip_dir, 'test1.zip')
 
# 압축해제
with zipfile.ZipFile(train_zip_dir, 'r') as z:
    z.extractall()
with zipfile.ZipFile(test_zip_dir, 'r') as z:
    z.extractall()
 
train_dir = os.path.join(os.getcwd(), 'train')
test_dir = os.path.join(os.getcwd(), 'test')
 
# train용 폴더 생성
train_set_dir = os.path.join(train_dir, 'train_set')
os.mkdir(train_set_dir)
train_dog_dir = os.path.join(train_set_dir, 'dog')
os.mkdir(train_dog_dir)
train_cat_dir = os.path.join(train_set_dir, 'cat')
os.mkdir(train_cat_dir)
# valid용 폴더 생성
valid_set_dir = os.path.join(train_dir, 'valid_set')
os.mkdir(valid_set_dir)
valid_dog_dir = os.path.join(valid_set_dir, 'dog')
os.mkdir(valid_dog_dir)
valid_cat_dir = os.path.join(valid_set_dir, 'cat')
os.mkdir(valid_cat_dir)
# test용 폴더 생성
test_set_dir = os.path.join(train_dir, 'test_set')
os.mkdir(test_set_dir)
test_dog_dir = os.path.join(test_set_dir, 'dog')
os.mkdir(test_dog_dir)
test_cat_dir = os.path.join(test_set_dir, 'cat')
os.mkdir(test_cat_dir)
 
# image file name list 생성
dog_files = [f'dog.{i}.jpg' for i in range(12500)]
cat_files = [f'cat.{i}.jpg' for i in range(12500)]
 
# 각 폴더로 image 이동
for file in dog_files[:10000]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(train_dog_dir, file)
    shutil.move(src, dst)
    
for file in dog_files[10000:12000]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(valid_dog_dir, file)
    shutil.move(src, dst)
 
for file in dog_files[12000:12500]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(test_dog_dir, file)
    shutil.move(src, dst)
 
for file in cat_files[:10000]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(train_cat_dir, file)
    shutil.move(src, dst)
    
for file in cat_files[10000:12000]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(valid_cat_dir, file)
    shutil.move(src, dst)
 
for file in cat_files[12000:12500]:
    src = os.path.join(train_dir, file)
    dst = os.path.join(test_cat_dir, file)
    shutil.move(src, dst)

print(f'the number of train set : {len(os.listdir(train_dog_dir)) + len(os.listdir(train_cat_dir))}')
print(f'the number of validn set : {len(os.listdir(valid_dog_dir)) + len(os.listdir(valid_cat_dir))}')
print(f'the number of test set : {len(os.listdir(test_dog_dir)) + len(os.listdir(test_cat_dir))}')

#data generator
train_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255,
                                                      rotation_range=40, #사진 회전각도 범위; 0~180사이값
                                                      width_shift_range=0.2, #수평과 수직으로 평행 이동시킬 범위; 전체 너비와 높이에 대한 비율값
                                                       height_shift_range=0.2, 
                                                       shear_range=0.2, #사진을 3D 각도로 기울임
                                                       zoom_range=0.2, #사진을 확대할 범위
                                                       horizontal_flip=True, # 랜덤하게 이미지를 수평으로 뒤집음
                                                       fill_mode='nearest') # 회전이나 이동을 통해 빈 곳이 생기면 픽셀을 채우는 방법
valid_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255)
test_datagen = tf.keras.preprocessing.image.ImageDataGenerator(rescale=1./255)
 
train_generator = train_datagen.flow_from_directory(train_set_dir,
                                                   target_size=(150,150),
                                                   batch_size=32,
                                                   class_mode='binary')
valid_generator = valid_datagen.flow_from_directory(valid_set_dir,
                                                   target_size=(150,150),
                                                   batch_size=32,
                                                   class_mode='binary')
test_generator = test_datagen.flow_from_directory(test_set_dir,
                                                 target_size=(150,150),
                                                 batch_size=32,
                                                 class_mode='binary')
 
train_step = train_generator.n // 32
valid_step = valid_generator.n // 32
test_step = test_generator.n // 32

#CNN model 구현
model = tf.keras.models.Sequential([
    tf.keras.layers.Input(shape=(150,150,3)),
    tf.keras.layers.Conv2D(filters=32, kernel_size=(3,3), strides=1, padding='same', activation='relu'),
    tf.keras.layers.MaxPooling2D((2, 2)),
    tf.keras.layers.Conv2D(filters=64, kernel_size=(3,3), strides=1, padding='same', activation='relu'),
    tf.keras.layers.MaxPooling2D((2, 2)),
    tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3), strides=1, padding='same', activation='relu'),
    tf.keras.layers.MaxPooling2D((2, 2)),
    tf.keras.layers.Conv2D(filters=128, kernel_size=(3,3), strides=1, padding='same', activation='relu'),
    tf.keras.layers.MaxPooling2D((2, 2)),
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dropout(0.5),
    tf.keras.layers.Dense(1024, activation='relu'),
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Dense(512, activation='relu'),
    tf.keras.layers.Dropout(0.1),
    tf.keras.layers.Dense(1, activation='sigmoid')
])
 
model.summary()
# 신경망 모델 학습
model.compile(optimizer=tf.keras.optimizers.Adam(1e-3),
             loss='binary_crossentropy',
              metrics=['acc'])
              
model.fit_generator(train_generator,
                   steps_per_epoch=train_step,
                   epochs=10,
                   validation_data=valid_generator,
                   validation_steps=valid_step)

• C-P-C-P-C-P-C-P-FC