cnn_2.py

import tensorflow as tf
from tensorflow import keras

import matplotlib.pyplot as plt

import numpy as np

import os
import zipfile

import random

import shutil

# 다운로드 받은 이미지 압축 파일 해제
# print(os.listdir("./drive/MyDrive"))
local_zip = './train.zip'

zip_ref = zipfile.ZipFile(local_zip, 'r')
zip_ref.extractall('./dev/train')
zip_ref.close()

local_zip = './test.zip'

zip_ref = zipfile.ZipFile(local_zip, 'r')
zip_ref.extractall('./dev/test')
zip_ref.close()


# 훈련 이미지 디렉토리 설정

base_dir = './dev'

train_dir = os.path.join(base_dir, 'train/train/')
test_dir = os.path.join(base_dir, 'test/test/0/')

train_val = {}

try:
    if not os.path.exists(base_dir + "/" + "val"):
        os.makedirs(base_dir + "/" + "val")
except OSError:
    print ('Error: Creating directory. ' +  base_dir + "/" + "val")

    
for folder in os.listdir(train_dir):
    
    try:
        if not os.path.exists(base_dir + "/" + "val/" + folder):
            os.makedirs(base_dir + "/" + "val/" + folder)
    except OSError:
        print ('Error: Creating directory. ' +  base_dir + "/" + "val/" + folder)
    
    for f_name in os.listdir(train_dir + folder)[int(len(os.listdir(train_dir + folder)) * 0.8):]:
        new_path = shutil.move(train_dir + folder + "/" + f_name, "./dev/val/" + folder)

# train / val 분리 확인
for folder in os.listdir(train_dir):
    print(len(os.listdir(base_dir + "/" + "val/" + folder)))
    print(len(os.listdir(base_dir + "/" + "train/train/" + folder)))
    print("-----")
print(os.listdir(base_dir + "/" + "val"))
# os.rmdir(base_dir + "/" + "val/" + '.ipynb_checkpoints')


# train/test 경로에 따른 이미지 파일 확인
# training image 폴더명 : 파일명 형태로 정리
train_file = {}

test_file = os.listdir(test_dir)
for folder in os.listdir(train_dir):
    train_file[folder] = os.listdir(train_dir + folder)
print(len(os.listdir(test_dir)))
print(os.listdir(train_dir))

def to_one_hot(labels, dimension=7):
    result = np.zeros((len(labels), dimension))
    for i, label in enumerate(labels):
        result[i, label] = 1.
    return result

get_ipython().run_line_magic('matplotlib', 'inline')
import matplotlib.image as mpimg

# 이미지를 matplotlib를 사용해서 4 * 4 형태의 격자로 출력 예정
nrows = 4
ncols = 4

fig = plt.gcf()
fig.set_size_inches(ncols * 4, nrows * 4)

for folder_name in os.listdir(train_dir):
    for i, img_path in enumerate(train_file[folder_name][:16]):
        sp = plt.subplot(nrows, ncols, i + 1)
        sp.axis('Off')
        img = mpimg.imread(train_dir + folder_name + "/" + img_path)
        plt.imshow(img)

plt.show()

# validation image show
for i, img_path in enumerate(test_file[:16]):
    sp = plt.subplot(nrows, ncols, i + 1)
    sp.axis('Off')
    img = mpimg.imread(test_dir + img_path)
    plt.imshow(img)

plt.show()

## CNN 다중 분류 이미지 sample test

import keras_preprocessing
from keras_preprocessing import image
from keras_preprocessing.image import ImageDataGenerator

training_datagen = ImageDataGenerator(
      rescale = 1./255,
	  rotation_range=40,
      width_shift_range=0.2,
      height_shift_range=0.2,
      shear_range=0.2,
      zoom_range=0.2,
      horizontal_flip=True,
      fill_mode='nearest')

validation_datagen = ImageDataGenerator(
    rescale = 1./255,
	  rotation_range=40,
      width_shift_range=0.2,
      height_shift_range=0.2,
      shear_range=0.2,
      zoom_range=0.2,
      horizontal_flip=True,
      fill_mode='nearest')

test_datagen = ImageDataGenerator(rescale = 1./255)

train_generator = training_datagen.flow_from_directory(
	train_dir,
	target_size=(227,227),
	class_mode='categorical',
    batch_size=48
)

val_generator = validation_datagen.flow_from_directory(
	base_dir + "/" + "val/",
	target_size=(227,227),
	class_mode='categorical', 
    batch_size=48
)

test_generator = test_datagen.flow_from_directory(
	'./dev/test/test',
	target_size=(227,227),
	class_mode='categorical',
  batch_size=48
)

model = tf.keras.models.Sequential([

    tf.keras.layers.Conv2D(64, (3,3), activation='relu', input_shape=(227, 227, 3)),
    tf.keras.layers.MaxPooling2D(2, 2),

    tf.keras.layers.Conv2D(64, (3,3), activation='relu'),
    tf.keras.layers.MaxPooling2D(2,2),
    
    tf.keras.layers.Flatten(),
    tf.keras.layers.Dropout(0.2),

    tf.keras.layers.Dense(512, activation='relu'),
    
    tf.keras.layers.BatchNormalization(),
    tf.keras.layers.Dropout(0.2),
    
    tf.keras.layers.Dense(256, activation='relu'),
    
    tf.keras.layers.BatchNormalization(),
    tf.keras.layers.Dropout(0.2),
    
    tf.keras.layers.Dense(128, activation='relu'), 
    
    tf.keras.layers.Dense(7, activation='softmax')
])


model.summary()

model.compile(loss = 'categorical_crossentropy', 
              optimizer=keras.optimizers.Adam(learning_rate=1e-4), 
              metrics=['accuracy']) 
# optimizer changed => rmsprop into adam
# adam을 사용했을 때 훈련 정확도가 더 높아졌다.
# 이유가 뭘까?


history = model.fit(train_generator, epochs=20, validation_data = val_generator)

prediction = model.predict(test_generator)
prediction

evaluate = model.evaluate(test_generator)
evaluate