{"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","collapsed":true,"jupyter":{"outputs_hidden":true},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nfrom sklearn.model_selection import StratifiedKFold\nfrom imblearn.over_sampling import SMOTE\nfrom tensorflow.keras.preprocessing.image import ImageDataGenerator\nimport matplotlib.pyplot as plt\nimport os\nimport numpy as np","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df = pd.read_csv(\"../input/pp-merge/nodups_train.csv\")\ndisplay(df)\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(df['labels'].value_counts())","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# # label split\n\nlabel2id = {\n    'complex' : 0, \n    'frog_eye_leaf_spot': 1, \n    'powdery_mildew' : 2, \n    'rust' : 3 , \n    'scab' : 4,\n    'healthy' : 5\n}\n\ndisplay(df.head())\ndf[\"labels\"] = df[\"labels\"].map(lambda x : [i for i in x.split(\" \")])\ndisplay(df.head())\ndf[\"labels\"] = df[\"labels\"].map(lambda x : [label2id[i] for i in x])\ndisplay(df.head())","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# # label 하나씩으로 나누기\ndf_new = pd.DataFrame()\n\nfor image_data, label_data in df.values:\n    if len(label_data) > 1 :\n        for label in label_data:\n            tmp_list= [image_data, label]\n            df_new=df_new.append(pd.Series(tmp_list, index= df.columns), ignore_index=True)\n    else:\n        tmp_list= [image_data, label_data[0]]\n        df_new=df_new.append(pd.Series(tmp_list, index= df.columns), ignore_index=True)\n        \ndf_new = df_new.astype({'labels':'int32'})         \ndisplay(df_new)            \n\n# df_new.to_csv('../input/pp-merge/one_label.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# label 하나씩인 df\n# df_new = pd.read_csv('../input/pp-merge/one_label.csv', index=False)\n\ndisplay(df_new) \ndisplay(df_new['labels'].value_counts())","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# label별로 균일하게 data split\nsfk = StratifiedKFold(5)\nfor train_idx, valid_idx in sfk.split(df_new['image'], df_new['labels']):\n    df_train = df_new.iloc[train_idx]\n    df_valid = df_new.iloc[valid_idx]\n    break\n    \nprint(f\"train size: {len(df_train)}\")\nprint(f\"valid size: {len(df_valid)}\")\n\n# df_train.to_csv('./data/plant-pathology-2021-fgvc8/merge/train_data.csv', index=False)\n# df_valid.to_csv('./data/plant-pathology-2021-fgvc8/merge/validation_data.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df_train['labels'].unique()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# # Imagegenerator\n\n# train_datagen = ImageDataGenerator(rescale= 1/255)\n# validation_datagen = ImageDataGenerator(rescale= 1/255)\n\n# df_train = df_train.astype({'labels':'str'})\n# df_valid = df_valid.astype({'labels':'str'})\n\n# train_generator = train_datagen.flow_from_dataframe(\n#     df_train,\n#     directory = '../input/plant-pathology-2021-fgvc8/train_images/',\n#     x_col = 'image',\n#     y_col = 'labels',\n#     target_size = (150,150),\n#     color_mode = 'rgb',\n#     class_mode = 'sparse',\n#     batch_size = 20\n# )\n\n# validation_generator = validation_datagen.flow_from_dataframe(\n#     df_valid,\n#     directory = '../input/plant-pathology-2021-fgvc8/train_images/',\n#     x_col = 'image',\n#     y_col = 'labels',\n#     target_size = (150,150),\n#     color_mode = 'rgb',\n#     class_mode = 'sparse',\n#     batch_size = 20\n# )\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"train_datagen = ImageDataGenerator(rescale=1/255,\n                                   rotation_range=20,\n                                   width_shift_range=0.1,\n                                   height_shift_range=0.1,\n                                   zoom_range=0.1,\n                                   horizontal_flip=True,\n                                   vertical_flip=True)\nvalidation_datagen = ImageDataGenerator(rescale= 1/255)\ndf_train = df_train.astype({'labels':'str'})\ndf_valid = df_valid.astype({'labels':'str'})\ntrain_generator = train_datagen.flow_from_dataframe(\n    df_train,\n    directory = '../input/plant-pathology-2021-fgvc8/train_images',\n    x_col = 'image',\n    y_col = 'labels',\n    target_size = (150,150),\n    color_mode = 'rgb',\n    class_mode = 'categorical',\n    batch_size = 20\n)\nvalidation_generator = validation_datagen.flow_from_dataframe(\n    df_valid,\n    directory = '../input/plant-pathology-2021-fgvc8/train_images',\n    x_col = 'image',\n    y_col = 'labels',\n    target_size = (150,150),\n    color_mode = 'rgb',\n    class_mode = 'categorical',\n    batch_size = 20\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport tensorflow as tf\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Conv2D, Flatten, Dense\nfrom tensorflow.keras.layers import MaxPooling2D, Dropout\nfrom tensorflow.keras.optimizers import Adam\nimport matplotlib.pyplot as plt\nfrom sklearn.preprocessing import MinMaxScaler\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import classification_report\n\n\nmodel = Sequential()\n\nmodel.add(Conv2D(filters=64,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu',\n                 \n                 input_shape=(150,150,3)))\n\nmodel.add(Conv2D(filters=64,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=128,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=128,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=512,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=512,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=512,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\n\nmodel.add(Flatten())\nmodel.add(Dense(units=512,\n                activation='relu'))\n\nmodel.add(Dropout(rate=0.5))\n\nmodel.add(Dense(units=128,\n                activation='relu'))\nmodel.add(Dropout(rate=0.5))\n\nmodel.add(Dense(units=6,\n                activation='softmax'))\n\nprint(model.summary())\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"METRIC = \"val_f1_score\"\n\ndef create_callbacks(metric = METRIC):\n    \n    cpk_path = './best_model.h5'\n    \n    checkpoint = tf.keras.callbacks.ModelCheckpoint(\n        filepath=cpk_path,\n        monitor= metric,\n        mode='max',\n        save_best_only=True,\n        verbose=1,\n    )\n\n    reducelr = tf.keras.callbacks.ReduceLROnPlateau(\n        monitor= metric,\n        mode='max',\n        factor=0.2,\n        patience=3,\n        verbose=1\n    )\n\n    earlystop = tf.keras.callbacks.EarlyStopping(\n        monitor= metric,\n        mode='max',\n        patience=10, \n        verbose=1\n    )\n    \n    callbacks = [checkpoint, reducelr, earlystop]         \n    \n    return callbacks","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# # First HP\n# from tensorflow.keras.optimizers import RMSprop,Adam\n# import tensorflow_addons as tfa\n\n# epochs = 30\n# batch_size = 50\n# optimizer = Adam(lr = 0.0001)\n# model.compile(optimizer = optimizer,\n#              loss = 'CategoricalCrossentropy',\n#              metrics = ['accuracy'])\n# callbacks = create_callbacks()\n# history = model.fit(train_generator,\n#                     epochs = epochs,\n#                     steps_per_epoch=100,\n#                     validation_data = validation_generator,\n#                     verbose=1, \n#                     callbacks = callbacks)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# model.save('../input/plant-pathology-2021-fgvc8/cnn.h5')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# figure, axis = plt.subplots(2, 1, figsize=(15,15))\n# axis.ravel()\n# axis[0].plot(history.history['accuracy'],label='Training Data')\n# axis[0].plot(history.history['val_accuracy'], label='Validation Data')\n# axis[0].set(xlabel='Epochs',ylabel='Accuracy', title='Accuracy vs Epochs')\n# axis[0].legend(loc=\"upper left\")\n\n# axis[1].plot(history.history['loss'], label='Training Data')\n# axis[1].plot(history.history['val_loss'], label='Validation Data')\n# axis[1].set(xlabel='Epochs',ylabel='Loss', title='Categorical Crossentropy Loss vs Epochs')\n# axis[1].legend(loc=\"upper left\")\n\n# plt.show()","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import pandas as pd\nfrom sklearn.model_selection import StratifiedKFold\nfrom imblearn.over_sampling import SMOTE\nfrom tensorflow.keras.preprocessing.image import ImageDataGenerator\nimport matplotlib.pyplot as plt\nimport os\nimport numpy as np","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"smote_train = pd.read_csv(\"../input/pp-merge/train_data.csv\")\ndisplay(smote_train)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# # label별로 균일하게 data split\n# sfk = StratifiedKFold(5)\n# for train_idx, valid_idx in sfk.split(df_new['image'], df_new['labels']):\n#     df_train = df_new.iloc[train_idx]\n#     df_valid = df_new.iloc[valid_idx]\n#     break\n    \n# print(f\"train size: {len(df_train)}\")\n# print(f\"valid size: {len(df_valid)}\")\n\n# # df_train.to_csv('./data/plant-pathology-2021-fgvc8/merge/train_data.csv', index=False)\n# # df_valid.to_csv('./data/plant-pathology-2021-fgvc8/merge/validation_data.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y_train = smote_train['labels'].values","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_index = list(smote_train.columns.values)[1:]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_train = smote_train[x_index].values","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"smote_valid = pd.read_csv(\"../input/pp-merge/validation_data.csv\")","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"y_valid = smote_valid['labels'].values","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_val_index = list(smote_valid.columns.values)[1:]","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_valid=smote_valid[x_val_index].values","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import numpy as np\nimport pandas as pd\nimport tensorflow as tf\nfrom tensorflow.keras.models import Sequential\nfrom tensorflow.keras.layers import Conv2D, Flatten, Dense\nfrom tensorflow.keras.layers import MaxPooling2D, Dropout\nfrom tensorflow.keras.optimizers import Adam\nimport matplotlib.pyplot as plt\nfrom sklearn.preprocessing import MinMaxScaler\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.metrics import classification_report\n\n\nmodel = Sequential()\n\nmodel.add(Conv2D(filters=64,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu',\n                 \n                 input_shape=(28,28,3)))\n\nmodel.add(Conv2D(filters=64,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(Conv2D(filters=64,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=128,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=128,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(Conv2D(filters=128,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=256,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\nmodel.add(MaxPooling2D(pool_size=(2,2)))\n\nmodel.add(Conv2D(filters=512,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\nmodel.add(Conv2D(filters=512,\n                 kernel_size=(3,3),\n                 padding='same',\n                 activation='relu'))\n\n\nmodel.add(Flatten())\nmodel.add(Dense(units=512,\n                activation='relu'))\nmodel.add(Dropout(rate=0.5))\n\nmodel.add(Dense(units=128,\n                activation='relu'))\nmodel.add(Dropout(rate=0.5))\n\nmodel.add(Dense(units=6,\n                activation='softmax'))\n\nprint(model.summary())\n","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"x_train","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model.compile(optimizer=Adam(learning_rate=1e-4),\n              loss='sparse_categorical_crossentropy',\n              metrics=['sparse_categorical_accuracy'])\n\n\nhistory = model.fit(x_train.reshape(-1,28,28,3),\n                    y_train,\n                    steps_per_epoch=250,\n                    epochs=50,\n                    batch_size=100,\n                    verbose=1\n                    )","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model.save('../kaggle/working/smote_cnn.h5')","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_df = pd.read_csv('../input/plant-pathology-2021-fgvc8/sample_submission.csv')\ntest_dir = '../input/plant-pathology-2021-fgvc8/test_images/'\n# test_df = pd.DataFrame()\n# test_df['image'] = os.listdir(test_dir)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"datagen = ImageDataGenerator(rescale=1/255)\n\ntest_generator = datagen.flow_from_dataframe(\n    test_df,\n    directory = '../input/plant-pathology-2021-fgvc8/test_images',\n    x_col = 'image',\n    y_col = 'labels',\n    target_size = (28,28),\n    color_mode = 'rgb',\n    class_mode = 'categorical',\n    batch_size = 20\n)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"prediction = model.predict(test_generator)\nprint(prediction)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_label_list = []\nfor i in prediction:\n    test_result = np.argmax(i)\n    \n    test_label_list.append(test_result)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"labels_new_dict = { 'complex' : 0, \n                    'frog_eye_leaf_spot': 1, \n                    'powdery_mildew' : 2, \n                    'rust' : 3 , \n                    'scab' : 4,\n                    'healthy' : 5}\n\nfor key, value in labels_new_dict.items():\n    for i in range(len(test_label_list)):\n        if test_label_list[i] == value:\n            test_label_list[i] = key\nprint(test_label_list) ","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(test_label_list)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"test_image_list = os.listdir('../input/plant-pathology-2021-fgvc8/test_images') \nfinal_df = pd.DataFrame()\nfinal_df['image'] = test_df['image']\nfinal_df['labels'] = test_label_list\ndisplay(final_df)\n# test_df.to_csv('/home/lab09/data/plant-pathology-2021-fgvc8/test_result.csv', index=False)\nfinal_df.to_csv('./submission.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# test_df.to_csv('/home/lab09/data/plant-pathology-2021-fgvc8/test_result.csv', index=False)\ntest_df.to_csv('./submission.csv', index=False)","metadata":{"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# First HP\nfrom tensorflow.keras.optimizers import RMSprop,Adam\nimport tensorflow_addons as tfa\n\nepochs = 30\nbatch_size = 50\noptimizer = Adam(lr = 0.001)\nmodel.compile(optimizer = optimizer,\n             loss = 'SparseCategoricalCrossentropy',\n             metrics = ['SparseCategoricalCrossentropy'])\ncallbacks = create_callbacks()\nhistory = model.fit(train_generator,\n                    epochs = epochs,\n                    steps_per_epoch=200,\n                    validation_data = validation_generator,\n                    verbose=1, \n                    callbacks = callbacks)","metadata":{"jupyter":{"source_hidden":true}},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}