{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"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 in \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 \"../input/\" directory.\n# For example, running this (by clicking run or pressing Shift+Enter) will list the files in the input directory\n\nimport os\nprint(os.listdir(\"../input\"))\n\n# Any results you write to the current directory are saved as output.","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"**Import Packages**"},{"metadata":{"trusted":true},"cell_type":"code","source":"from keras.applications.xception import Xception\nfrom keras.models import Model\nfrom keras.layers import Dense\nfrom keras.layers import Flatten\nfrom keras.optimizers import SGD,Adam,Adadelta\nfrom keras.callbacks import History \nfrom keras.preprocessing.image import ImageDataGenerator\nfrom keras.preprocessing.image import load_img\nfrom keras.preprocessing.image import img_to_array\nfrom keras.models import load_model\nfrom keras.models import model_from_json\nimport pandas as pd\nimport numpy as np\nimport matplotlib.pyplot as plt\nimport os","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Pre-Process Photos into Directories"},{"metadata":{"trusted":true},"cell_type":"code","source":"os.mkdir('dataset_dogs_vs_cats/')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Copy files\ncp -avr ../input/train/train/ /kaggle/working/dataset_dogs_vs_cats","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# organize dataset into a useful structure\nfrom os import makedirs\nfrom os import listdir\nfrom shutil import copyfile\nfrom random import seed\nfrom random import random\n# create directories\ndataset_home = '/kaggle/working/dataset_dogs_vs_cats/'\nsubdirs = ['train/', 'test/']\nfor subdir in subdirs:\n    # create label subdirectories\n    labeldirs = ['dogs/', 'cats/']\n    for labldir in labeldirs:\n        newdir = dataset_home + subdir + labldir\n        makedirs(newdir, exist_ok=True)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"os.listdir(\"/kaggle/working/dataset_dogs_vs_cats/\")","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#rm -r '/kaggle/working/dataset_dogs_vs_cats/'\n\n# seed random number generator\nseed(1)\n# define ratio of pictures to use for validation\nval_ratio = 0.25\n# copy training dataset images into subdirectories\nsrc_directory = '/kaggle/working/dataset_dogs_vs_cats/'\nfor file in listdir(src_directory):\n    src = src_directory + file\n    if random() < val_ratio:\n        dst_dir = 'test/'     \n    else:\n        dst_dir = 'train/'\n        \n    if file.startswith('cat'):\n        dst = dataset_home + dst_dir + 'cats/'  + file\n        copyfile(src, dst)\n    elif file.startswith('dog'):\n        dst = dataset_home + dst_dir + 'dogs/'  + file\n        copyfile(src, dst)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#Define train and test path\ntrain_path = '/kaggle/working/dataset_dogs_vs_cats/train/'\ntest_path  = '/kaggle/working/dataset_dogs_vs_cats/test/'","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Plot Some images"},{"metadata":{"trusted":true},"cell_type":"code","source":"# plot dog photos from the dogs vs cats dataset\nplt.figure(figsize=(18,10))\n# define location of dataset\nfolder = train_path+'dogs/'\n# plot first few images\nfor i in range(9):\n    # define subplot\n    pyplot.subplot(330 + 1 + i)\n    # define filename\n    filename = random.choice(os.listdir(folder))\n    file = folder + filename\n    # load image pixels\n    image = imread(file)\n    # plot raw pixel data\n    pyplot.imshow(image)\n# show the figure\npyplot.show()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# plot dog photos from the dogs vs cats dataset\nplt.figure(figsize=(18,10))\n# define location of dataset\nfolder = train_path+'cats/'\n# plot first few images\nfor i in range(9):\n    # define subplot\n    pyplot.subplot(330 + 1 + i)\n    # define filename\n    filename = random.choice(os.listdir(folder))\n    file = folder + filename\n    # load image pixels\n    image = imread(file)\n    # plot raw pixel data\n    pyplot.imshow(image)\n# show the figure\npyplot.show()","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Create a CNN Model based on Transfer Learning from Xception model\n[https://keras.io/applications/#xception](https://keras.io/applications/#xception)"},{"metadata":{},"cell_type":"markdown","source":"#### 1 - CNN Model - Optimizer:SGD Without Regularization"},{"metadata":{"trusted":true},"cell_type":"code","source":"# define cnn model\ndef Model_CNN_SGD_No_Regularization():\n    # load model\n    model = Xception(include_top=False,weights='imagenet', input_shape=(299, 299, 3))\n    # mark loaded layers as not trainable\n    for layer in model.layers:\n        layer.trainable = False\n    # add new classifier layers\n    flat1 = Flatten()(model.layers[-1].output)\n    class1 = Dense(128, activation='relu', kernel_initializer='he_uniform')(flat1)\n    output = Dense(1, activation='sigmoid')(class1)\n    # define new model\n    model = Model(inputs=model.inputs, outputs=output)\n    # compile model\n    opt = SGD(lr=0.001, momentum=0.9)\n    model.compile(optimizer=opt, loss='binary_crossentropy', metrics=['accuracy'])\n    return model","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"#### 2 - CNN Model - Optimizer:Adam Without Regularization"},{"metadata":{"trusted":true},"cell_type":"code","source":"# define cnn model\ndef Model_CNN_ADAM_No_Regularization():\n    # load model\n    model = Xception(include_top=False,weights='imagenet', input_shape=(299, 299, 3))\n    # mark loaded layers as not trainable\n    for layer in model.layers:\n        layer.trainable = False\n    # add new classifier layers\n    flat1 = Flatten()(model.layers[-1].output)\n    class1 = Dense(128, activation='relu', kernel_initializer='he_uniform')(flat1)\n    output = Dense(1, activation='sigmoid')(class1)\n    # define new model\n    model = Model(inputs=model.inputs, outputs=output)\n    # compile model\n    opt = Adam(lr=0.001, decay=0.0)\n    model.compile(optimizer=opt, loss='binary_crossentropy', metrics=['accuracy'])\n    return model","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"#### 3 - CNN Model - Optimizer:Adadelta Without Regularization"},{"metadata":{"trusted":true},"cell_type":"code","source":"# define cnn model\ndef Model_CNN_ADADELTA_No_Regularization():\n    # load model\n    model = Xception(include_top=False,weights='imagenet', input_shape=(299, 299, 3))\n    # mark loaded layers as not trainable\n    for layer in model.layers:\n        layer.trainable = False\n    # add new classifier layers\n    flat1 = Flatten()(model.layers[-1].output)\n    class1 = Dense(128, activation='relu', kernel_initializer='he_uniform')(flat1)\n    output = Dense(1, activation='sigmoid')(class1)\n    # define new model\n    model = Model(inputs=model.inputs, outputs=output)\n    # compile model\n    opt = Adadelta(lr=0.001, decay=0.0)\n    model.compile(optimizer=opt, loss='binary_crossentropy', metrics=['accuracy'])\n    return model","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Prepare the datasets to the Fit process"},{"metadata":{"trusted":true},"cell_type":"code","source":"from keras.preprocessing.image import ImageDataGenerator, load_img\n\n\n#For this case, we'll use Data Augmentation\ntraining_datagen = ImageDataGenerator(rescale=1./255\n                                      #,validation_split=0.1\n                                      ,data_format='channels_last'\n        ,shear_range=0.2,\n        zoom_range=0.2,\n        horizontal_flip=True,\n        featurewise_center=True)\n\n\ntest_datagen = ImageDataGenerator(rescale=1./255\n                                   #,validation_split=0.1\n                                  ,data_format='channels_last')\n\n\ntraining_set = training_datagen.flow_from_directory(\n        directory=train_path,\n        target_size=(299, 299),\n        batch_size=64,\n        #classes=['Dog','Cat'],\n        subset = \"training\",\n        #save_to_dir = os.path.join(dataset_path,'train'),\n        class_mode='binary')\n\ntest_set = test_datagen.flow_from_directory(\n       directory=test_path,\n        target_size=(299, 299),\n        batch_size=64,\n        #classes=['Dog','Cat'],\n        #subset = \"validation\",\n        #save_to_dir = os.path.join(dataset_path,'test'),\n        class_mode='binary')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### Let's train our models"},{"metadata":{"trusted":true},"cell_type":"code","source":"#Define Models\nModel_CNN_SGD_No_Regularization = Model_CNN_SGD_No_Regularization()\nModel_CNN_ADAM_No_Regularization = Model_CNN_ADAM_No_Regularization()\nModel_CNN_ADADELTA_No_Regularization = Model_CNN_ADADELTA_No_Regularization()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"history_1 = History()\nhistory_2 = History()\nhistory_3 = History()\nepochs = 10","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Model_CNN_SGD_No_Regularization.fit_generator(training_set,steps_per_epoch=len(training_set),epochs=epochs,validation_data=test_set,validation_steps=len(test_set),callbacks=[history_1])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Model_CNN_ADAM_No_Regularization.fit_generator(training_set,steps_per_epoch=len(training_set),epochs=epochs,validation_data=test_set,validation_steps=len(test_set),callbacks=[history_2])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"Model_CNN_ADAM_No_Regularization.fit_generator(training_set,steps_per_epoch=len(training_set),epochs=epochs,validation_data=test_set,validation_steps=len(test_set),callbacks=[history_2])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"df = pd.DataFrame()\ndf['class'] = [os.listdir(train_path)[j][:3] for j in range(len(train_path))]\ndf['filepath'] = [os.listdir(train_path)[j][:3] for j in range(len(train_path))]\n#Rename files\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def rename_image_files(path):\n\"\"\"This fucntion will be used to rename images from the train dataset\"\"\"\n    i=0\n    os.listdir(path)\n    for filename in os.listdir(train_path): \n        dst = filename[4:]+str(\"_\") + str(i) + \".jpg\"\n        src =path + filename \n        dst =path+ dst \n\n        # rename() function will \n        # rename all the files \n        os.rename(src, dst) \n        i += 1\n    return None","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"df","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"display_name":"Python 3","language":"python","name":"python3"},"language_info":{"name":"python","version":"3.6.4","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"}},"nbformat":4,"nbformat_minor":1}