{"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","execution":{"iopub.status.busy":"2021-07-30T05:33:48.512904Z","iopub.execute_input":"2021-07-30T05:33:48.513611Z","iopub.status.idle":"2021-07-30T05:33:48.52791Z","shell.execute_reply.started":"2021-07-30T05:33:48.513503Z","shell.execute_reply":"2021-07-30T05:33:48.526831Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#This program uses Resnet50 fine tuning to classify cats and dogs.\n#Standard library\nimport copy\nimport re\nimport random\nimport glob\nimport os\nimport zipfile\nimport shutil\n\n#Related third party\nfrom sklearn.model_selection import train_test_split\nimport numpy as np\nfrom PIL import Image\nfrom sklearn.preprocessing import LabelEncoder\nimport cv2\nfrom keras.utils import np_utils\nfrom matplotlib import pyplot as plt\nimport pandas as pd\n\nfrom keras.applications.resnet50 import ResNet50\nfrom keras.models import Model\nimport tensorflow as tf\n\nimport keras\nfrom tensorflow import keras\nfrom keras.layers import Conv2D, MaxPooling2D,Input\nfrom keras.layers import Dense, Dropout, Flatten, Activation,GlobalAveragePooling2D\nfrom keras.models import Sequential\nfrom keras.callbacks import EarlyStopping, ReduceLROnPlateau, ModelCheckpoint\nfrom keras.optimizers import Adam,SGD,RMSprop","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:33:48.529247Z","iopub.execute_input":"2021-07-30T05:33:48.529694Z","iopub.status.idle":"2021-07-30T05:33:56.289687Z","shell.execute_reply.started":"2021-07-30T05:33:48.529653Z","shell.execute_reply":"2021-07-30T05:33:56.288518Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#unzip\n#Skip this cell if it's already unziped.\nzipfile.ZipFile('../input/dogs-vs-cats-redux-kernels-edition/test.zip').extractall()\nzipfile.ZipFile('../input/dogs-vs-cats-redux-kernels-edition/train.zip').extractall()","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:33:56.291965Z","iopub.execute_input":"2021-07-30T05:33:56.292284Z","iopub.status.idle":"2021-07-30T05:34:21.883128Z","shell.execute_reply.started":"2021-07-30T05:33:56.292244Z","shell.execute_reply":"2021-07-30T05:34:21.88184Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#make dir\n#Skip this cell if you have already created the directory\nos.makedirs(\"dataset\")\nos.makedirs(\"dataset/train\")\nos.makedirs(\"dataset/validation\")\nos.makedirs(\"dataset/train/cats\")\nos.makedirs(\"dataset/train/dogs\")\nos.makedirs(\"dataset/validation/cats\")\nos.makedirs(\"dataset/validation/dogs\")","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:34:21.884912Z","iopub.execute_input":"2021-07-30T05:34:21.88525Z","iopub.status.idle":"2021-07-30T05:34:21.8928Z","shell.execute_reply.started":"2021-07-30T05:34:21.885219Z","shell.execute_reply":"2021-07-30T05:34:21.891595Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#function definition\n\ndef square(img):\n    h,w,c = img.shape\n    long_edge = max(h,w)\n    top = 0\n    bottom = 0\n    left = 0\n    right = 0\n    if h < long_edge:\n        diff_h = long_edge - h\n        top = diff_h // 2\n        bottom = diff_h - top\n    elif w < long_edge:\n        diff_w = long_edge - w\n        left = diff_w // 2\n        right = diff_w - left\n    else:\n        pass\n    img = cv2.copyMakeBorder(img, top, bottom, left, right, cv2.BORDER_CONSTANT, value=[0, 0, 0])\n    return img","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:34:21.894619Z","iopub.execute_input":"2021-07-30T05:34:21.895378Z","iopub.status.idle":"2021-07-30T05:34:21.905211Z","shell.execute_reply.started":"2021-07-30T05:34:21.895328Z","shell.execute_reply":"2021-07-30T05:34:21.90429Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#pre processing\n#If you have already transformed and moved the image, skip this step.\ndogs = glob.glob(os.path.join(\"train\",\"dog.*.jpg\"))\ncats = glob.glob(os.path.join(\"train\",\"cat.*.jpg\"))    \n\nfor file_dog in dogs:\n    input = cv2.imread(file_dog)\n    input = square(input)\n    input = cv2.resize(input,dsize=(224,224))\n    cv2.imwrite(file_dog,input)\n\nfor file_cat in cats:\n    input = cv2.imread(file_cat)\n    input = square(input)\n    input = cv2.resize(input,dsize=(224,224))\n    cv2.imwrite(file_cat,input)","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:34:21.906818Z","iopub.execute_input":"2021-07-30T05:34:21.907516Z","iopub.status.idle":"2021-07-30T05:35:56.961193Z","shell.execute_reply.started":"2021-07-30T05:34:21.907471Z","shell.execute_reply":"2021-07-30T05:35:56.959823Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#parameta definition\nimg_width,img_height = 224,224\ntrain_size = 20000\nvalidation_size = 5000\nbatch_size = 32\n\nepochs = 80\nlearning_rate = 1e-5\ntest_num = 12500\n\ndetect_class = ['dogs','cats']\nclass_size = len(detect_class)\n\ntrain_data_dir = './dataset/train'\nvalidation_data_dir = './dataset/validation'","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:35:56.963072Z","iopub.execute_input":"2021-07-30T05:35:56.96345Z","iopub.status.idle":"2021-07-30T05:35:56.971696Z","shell.execute_reply.started":"2021-07-30T05:35:56.963417Z","shell.execute_reply":"2021-07-30T05:35:56.970555Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from keras.preprocessing.image import ImageDataGenerator\n\n#normalization and image padding\ntrain_data = ImageDataGenerator(rescale = 1.0 / 255, zoom_range = 0.5, horizontal_flip = True,vertical_flip=True,rotation_range=45)\nvalidation_data = ImageDataGenerator(rescale = 1.0 / 255)\n\ntrain_generator = train_data.flow_from_directory(train_data_dir,target_size=(img_width,img_height),color_mode='rgb',classes = detect_class,class_mode='categorical',batch_size=batch_size)\nvalidation_generator = validation_data.flow_from_directory(validation_data_dir,target_size=(img_width,img_height),color_mode='rgb',classes=detect_class,class_mode = 'categorical',batch_size=batch_size)\n\n#layer parameta definition\ninput_tensor = Input(shape=(img_width,img_height,3))\nbase_model = ResNet50(include_top = False,weights='imagenet',input_tensor=input_tensor)\nbase_model.trainable = True\n\n#model definition\ntop_model = Sequential()\n\ntop_model.add(Flatten(input_shape=base_model.output_shape[1:]))\ntop_model.add(Dense(class_size,activation='softmax'))\n\nmodel = Model(inputs=base_model.input,outputs=top_model(base_model.output))\n\n#compiler definition\nmodel.compile(loss = 'categorical_crossentropy',optimizer=RMSprop(lr=learning_rate),metrics=['accuracy'])","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:35:56.974408Z","iopub.execute_input":"2021-07-30T05:35:56.974798Z","iopub.status.idle":"2021-07-30T05:36:19.105202Z","shell.execute_reply.started":"2021-07-30T05:35:56.974705Z","shell.execute_reply":"2021-07-30T05:36:19.101158Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#train\nhistory = model.fit_generator(train_generator,steps_per_epoch=train_size // batch_size,epochs=epochs,validation_data=validation_generator,validation_steps=validation_size//batch_size)","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:36:19.107467Z","iopub.status.idle":"2021-07-30T05:36:19.108116Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#Visualize accuracy and save parameta\n\nhistory.history.keys()\nfig = plt.figure()\nplt.plot(range(1, epochs+1), history.history['accuracy'], label=\"training\")\nplt.plot(range(1, epochs+1), history.history['val_accuracy'], label=\"validation\")\nplt.xlabel('Epochs')\nplt.ylabel('Accuracy')\nplt.legend()\nplt.show()\n\nfig.savefig(\"history.png\")\nmodel.save_weights(\"weights.h5\")","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:36:19.109115Z","iopub.status.idle":"2021-07-30T05:36:19.109645Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"#test and plot csv for submit\nfrom keras.preprocessing import image\nimport csv\n\nwith open(\"output.csv\",\"a\") as file:\n    print(\"id,label\",file = file)\n    for test_image_num in range(1,test_num + 1):\n        img = cv2.imread(\"test/\"+str(test_image_num)+\".jpg\")\n        img = square(img)\n        img = cv2.resize(img,dsize=(224,224))\n        trans_img = image.img_to_array(img)\n        trans_img = np.expand_dims(trans_img, axis=0)\n        trans_img = trans_img / 255\n        pred = model.predict(trans_img)[0]\n        print(pred)\n        print(str(test_image_num) + \",\" + str(pred[0]),file=file)","metadata":{"execution":{"iopub.status.busy":"2021-07-30T05:36:19.110527Z","iopub.status.idle":"2021-07-30T05:36:19.111083Z"},"trusted":true},"execution_count":null,"outputs":[]}]}