{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"# This Python 3 environment comes with many helpful analytics libraries installed\n# It is defined by the kaggle/python Docker image: https://github.com/kaggle/docker-python\n# For example, here's several helpful packages to load\n\nimport numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\n\n# Input data files are available in the read-only \"../input/\" directory\n# For example, running this (by clicking run or pressing Shift+Enter) will list all files under the input directory\n\nimport os\nfor dirname, _, filenames in os.walk('/kaggle/input'):\n    for filename in filenames:\n        print(os.path.join(dirname, filename))\n\n# You can write up to 20GB to the current directory (/kaggle/working/) that gets preserved as output when you create a version using \"Save & Run All\" \n# You can also write temporary files to /kaggle/temp/, but they won't be saved outside of the current session","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import tensorflow as tf\nfrom tensorflow.keras.utils import to_categorical\nfrom sklearn.model_selection import train_test_split\nfrom PIL import Image\nimport glob\nimport matplotlib.pyplot as plt\nimport random\nimport matplotlib.image as mpimg\nimport shutil","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# zipファイルを展開","metadata":{}},{"cell_type":"code","source":"base_dir = \"../input/dogs-vs-cats\"\ntrain_dir = os.path.join(base_dir, \"train.zip\")\ntest_dir = os.path.join(base_dir, \"test1.zip\")\n\nimport zipfile\nwith zipfile.ZipFile(train_dir,\"r\") as z:\n    z.extractall()\n\nwith zipfile.ZipFile(test_dir,\"r\") as z:\n    z.extractall()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"os.listdir(\"/kaggle/working/\") #dir確認","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"os.getcwd()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"passを作る","metadata":{}},{"cell_type":"code","source":"train_path = '/kaggle/working/train/' #trainデータ\ntest1_path = '/kaggle/working/test1/' #testデータ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 猫と犬のディレクトリを作り分ける","metadata":{}},{"cell_type":"code","source":"\"\"\"\nos.mkdir(train_path + 'cats')\ncat_dir = ('/kaggle/working/train/cats/')\nos.mkdir(train_path + 'dogs')\ndog_dir = ('/kaggle/working/train/dogs/')\n\ndirs = [each for each in os.listdir(train_path)]\n\nfor file in dirs:\n    folder = file.split(\".\")[0]\n    if folder == 'cat':\n        shutil.move(train_path+file, cat_dir)\n    elif folder == 'dog':\n        shutil.move(train_path+file, dog_dir)\n\"\"\"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 画像の前処理","metadata":{}},{"cell_type":"code","source":"\"\"\"\nfolder = ['cats','dogs']\nimage_size = 224 #インプット画像のサイズ\nX = []\nY = []\nfor index, name in enumerate(folder): #ディレクトリの種類、画像ファイル名を繰返し取得する\n    dir = train_path + name\n    files = glob.glob(dir + \"/*.jpg\") #ファイル名の取得\n    for i, file in enumerate(files): #各画像へ↓の処理を繰返し実施する\n        image = Image.open(file) #画像の読み込み\n        image = image.convert(\"RGB\") #RGBに変換\n        image = image.resize((image_size, image_size)) #サイズ変更\n        data = np.asarray(image) #画像データを配列へ変更\n        X.append(data) #画像データを繰返し追加\n        Y.append(index) #cat:0, dog:1の値を繰返し追加\nX = np.array(X)\nY = np.array(Y)\n\"\"\"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\"\"\"\n#正解ラベルの形式を変換\nY = to_categorical(Y, 2)\nprint(Y)\n\"\"\"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"filenames = os.listdir(train_path) #dog123~~~,cat1234~~など\ncategories = [] #リスト初期化\nfor filename in filenames: #filenames(dog123~~~,cat1234~~)を一つずつfilenameに入れてforループ\n    category = filename.split('.')[0]#filenameは(dog.4784.jpg)この形なのでsplitを用いることにより[dog,4784,jpg]に分けられるこれの[0]をとることで[dog,cat]を受け取れる\n    if category == 'dog':\n        categories.append(1)#dog = 1,cat = 0をcnnでの判定で用いる。\n    else:\n        categories.append(0)\ndata = pd.DataFrame({'filename' : filenames , 'category' : categories})#dataframe型でdataに代入する(filename = dog.4784.jpg,category = 1(dog))\ndata.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data['category'].value_counts()#確認(dogとcatの数)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data['category1'] = data['category'].apply(lambda x: 'cat' if x== 0 else 'dog')#datafram型にcategory1(dog,cat)を追加する\ndata.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"data.drop(['category'] , inplace = True , axis = 1)#category(1,0)を削除する\ndata.rename(columns  = {'category1' : 'category'} , inplace =True)#category1(dog,cat)をcategoryに名前を変える\ndata.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_df , val_df = train_test_split(data , test_size = 0.2 , random_state = 42)#traindataをtrainとval(確認用)に分けるsizeはtrainが80%valが20%,seed値(random_state)を設定することで毎回同じデータに分けることが可能\ntrain_df = train_df.reset_index(drop = True)#splitをすることでindexがランダムになるので初期化する\nval_df = val_df.reset_index(drop = True)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\"\"\"\n#学習用データとテストデータへ分割\nX_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.20)\n\"\"\"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"\"\"\"\ndel X #メモリ削減\n\"\"\"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# ジェネレータの作成","metadata":{}},{"cell_type":"code","source":"image_size = 224#画像サイズ一辺\nImage_size = (image_size,image_size)#縦,横(224,224)\nbatch_size = 128#まとまり\nepochs = 10#iteration","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from tensorflow.keras.preprocessing.image import ImageDataGenerator\n\n# 画像データの拡張と前処理の設定\ntrain_datagen = ImageDataGenerator(\n    rescale=1.0/255,#0~1に正規化\n    rotation_range=45,#回転,-45~45\n    shear_range = 0.1,#シアー強度\n    zoom_range = 0.2 , #ズーム\n    horizontal_flip = True , #左右反転\n    width_shift_range = 0.1 , #水平シフト\n    height_shift_range = 0.1 #垂直シフト\n)\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# ジェネレータの作成\ntrain_generator = train_datagen.flow_from_dataframe(\n    train_df, #dataframe\n    train_path, #trainデータへのパス('/kaggle/working/train/')\n    x_col='filename',#特徴量\n    y_col='category',#category(0,1)\n    target_size=Image_size,#size\n    class_mode='categorical',#返すラベルの配列の形を決定する\n    batch_size=batch_size\n)  ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"validation_datagen = ImageDataGenerator(rescale = 1./255)\nval_generator = validation_datagen.flow_from_dataframe(\n    val_df, \n    train_path, \n    x_col='filename',\n    y_col='category',\n    target_size=Image_size,\n    class_mode='categorical',\n    batch_size=batch_size\n)  ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# モデルの作成","metadata":{}},{"cell_type":"code","source":"\n#CNNモデル（ResNet50）のライブラリをインポート\nfrom tensorflow.keras.applications.resnet50 import ResNet50\nfrom tensorflow.keras.models import Sequential, Model\nfrom tensorflow.keras.layers import Input, Flatten, Dense, Conv2D , MaxPooling2D , Dropout, Activation , BatchNormalization\n\n#input画像の定義\ninput_tensor = tf.keras.Input(shape=(image_size, image_size, 3))\n#ResNet50モデルの呼出し。ここでweights=\"imagenet\"とすることで学習済の重みを使うことができる\n#include_top=Falseとすることで出力層だけを今回の学習に合わせた形で書き換えます\nResNet50 = tf.keras.applications.ResNet50(include_top=False, weights=\"imagenet\", \n                                          input_tensor=input_tensor,classes=2,)\n#出力層の書換え。今回は2種類の画像分類のモデルを作成。\ntop_model = Sequential()\ntop_model.add(Flatten(input_shape = ResNet50.output_shape[1:]))\ntop_model.add(Dense(2, activation = 'sigmoid'))\nmodel = Model(inputs=ResNet50.inputs, outputs=top_model(ResNet50.outputs))","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from tensorflow.keras.optimizers import Adam \nadam = Adam(learning_rate=0.0001)\nmodel.compile(optimizer= adam, loss='categorical_crossentropy', metrics=['accuracy'])","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model.summary()","metadata":{"collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 学習","metadata":{}},{"cell_type":"code","source":"from keras.callbacks import EarlyStopping, ReduceLROnPlateau\nearlystop = EarlyStopping(patience = 10)\nlearning_rate_reduction = ReduceLROnPlateau(monitor='val_acc', \n                                            patience=2, \n                                            verbose=1, \n                                            factor=0.5, \n                                            min_lr=0.00001)\ncallbacks = [earlystop , learning_rate_reduction]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# モデルの学習\nhistory = model.fit_generator(\n    train_generator , \n    epochs = epochs , \n    validation_data=val_generator,\n    validation_steps=val_df.shape[0]//batch_size,\n    steps_per_epoch=train_df.shape[0]//batch_size,\n    callbacks=callbacks\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# 保存","metadata":{}},{"cell_type":"code","source":"model.save_weights('model.1')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"markdown","source":"# モデル→予測","metadata":{}},{"cell_type":"code","source":"test_filenames = os.listdir(test1_path)\ntest = pd.DataFrame({'filename' : test_filenames})\nnb_samples = test.shape[0]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_gen = ImageDataGenerator(rescale=1./255)\ntest_generator = test_gen.flow_from_dataframe(\n    test , \n    test1_path ,\n    x_col='filename',\n    y_col=None,\n    class_mode=None,\n    target_size=(image_size,image_size),\n    batch_size=batch_size,\n    shuffle=False\n    )","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"predict = model.predict_generator(test_generator , steps = np.ceil(nb_samples/batch_size))\ntest['category'] = np.argmax(predict  , axis= -1)\ntest.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"submission_df = test.copy()\nsubmission_df['id'] = submission_df['filename'].str.split('.').str[0]\nsubmission_df['label'] = submission_df['category']\nsubmission_df.head()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"submission_df.drop(['filename' , 'category'] , axis = 1 , inplace = True)\nsubmission_df.to_csv('submission.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]}]}