{"cells":[{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import numpy as np # linear algebra\nimport pandas as pd # data processing, CSV file I/O (e.g. pd.read_csv)\nimport os\nprint(os.listdir(\"../input\"))","execution_count":null,"outputs":[]},{"metadata":{"_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","trusted":true},"cell_type":"code","source":"from tqdm import tqdm_notebook\nfrom PIL import Image, ImageFilter, ImageChops, ImageOps","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!rm -r ../train\n!rm -r ../test","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"!mkdir ../train\n!mkdir ../test\n!mkdir ../train/0\n!mkdir ../train/1\n\n\n!mkdir ../test/0\n!mkdir ../test/1","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"data = pd.read_csv('/kaggle/input/train_labels.csv')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"data.head()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"one = data.query(\"label == 1\")['id'].values\none_train = one[:len(one) * 4 //5]\none_test = one[len(one) * 4 //5:]\n\nzero = data.query(\"label == 0\")['id'].values\nzero_train = zero[:len(zero) * 4 //5]\nzero_test = zero[len(zero) * 4 //5:]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"ORIGINAL_SIZE = 96\nCROP_SIZE = 32\n\ndef process(path, mode, label):\n\n    \n    colorImage = Image.open(\"../input/train/%s.tif\" % (path))\n    \n    \n    start_crop = (ORIGINAL_SIZE - CROP_SIZE) // 2\n    end_crop = start_crop + CROP_SIZE\n    colorImage = colorImage.crop((start_crop, start_crop, end_crop, end_crop))\n\n    \n    for I in range(4):\n        colorImage = colorImage.transpose(Image.ROTATE_90)\n        colorImage.save(\"../%s/%d/%s_%d.png\" % (mode, label, path, I), \"PNG\")\n        \n    colorImage = colorImage.transpose(Image.FLIP_LEFT_RIGHT)\n\n    for I in range(4):\n        colorImage = colorImage.transpose(Image.ROTATE_90)\n        colorImage.save(\"../%s/%d/%s_%d.png\" % (mode, label, path, I+4), \"PNG\")\n    \n    colorImage = colorImage.transpose(Image.FLIP_TOP_BOTTOM)\n    \n    for I in range(4):\n        colorImage = colorImage.transpose(Image.ROTATE_90)\n        colorImage.save(\"../%s/%d/%s_%d.png\" % (mode, label, path, I+8), \"PNG\")  \n\n    colorImage = colorImage.transpose(Image.FLIP_LEFT_RIGHT)\n    \n    for I in range(4):\n        colorImage = colorImage.transpose(Image.ROTATE_90)\n        colorImage.save(\"../%s/%d/%s_%d.png\" % (mode, label, path, I+12), \"PNG\")  \n    \n    \n    '''    \n    colorImage = Image.open(\"../input/train/%s.tif\" % (path))\n    start_crop = (ORIGINAL_SIZE - CROP_SIZE) // 2\n    end_crop = start_crop + CROP_SIZE\n    colorImage = colorImage.crop((start_crop, start_crop, end_crop, end_crop))\n\n    colorImage.save(\"../%s/%d/%s.png\" % (mode, label, path), \"PNG\")\n    '''","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"for I in tqdm_notebook(one_train):\n    process(I, \"train\", 1)\nfor I in tqdm_notebook(one_test):\n    process(I, \"test\", 1)\nfor I in tqdm_notebook(zero_train):\n    process(I, \"train\", 0)\nfor I in tqdm_notebook(zero_test):\n    process(I, \"test\", 0)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from keras_applications.imagenet_utils import _obtain_input_shape\nfrom keras.models import Sequential\nfrom keras import backend as K\nfrom keras.layers import Input, Convolution2D, MaxPooling2D, Activation, concatenate, Dropout, warnings\nfrom keras.layers import GlobalAveragePooling2D, GlobalMaxPooling2D, Dense\nfrom keras.models import Model\nfrom keras.engine.topology import get_source_inputs\nfrom keras.utils import get_file\nfrom keras.utils import layer_utils\n\n\nsq1x1 = \"squeeze1x1\"\nexp1x1 = \"expand1x1\"\nexp3x3 = \"expand3x3\"\nrelu = \"relu_\"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def fire_module(x, fire_id, squeeze=16, expand=64):\n    s_id = 'fire' + str(fire_id) + '/'\n\n    if K.image_data_format() == 'channels_first':\n        channel_axis = 1\n    else:\n        channel_axis = 3\n    \n    x = Convolution2D(squeeze, (1, 1), padding='valid', name=s_id + sq1x1)(x)\n    x = Activation('relu', name=s_id + relu + sq1x1)(x)\n\n    left = Convolution2D(expand, (1, 1), padding='valid', name=s_id + exp1x1)(x)\n    left = Activation('relu', name=s_id + relu + exp1x1)(left)\n\n    right = Convolution2D(expand, (3, 3), padding='same', name=s_id + exp3x3)(x)\n    right = Activation('relu', name=s_id + relu + exp3x3)(right)\n\n    x = concatenate([left, right], axis=channel_axis, name=s_id + 'concat')\n    return x\n\n\n# Original SqueezeNet from paper.\n\ndef SqueezeNet():\n\n\n    inputs = Input(shape=(32,32,3))\n\n\n    x = Convolution2D(64, (3, 3), strides=(2, 2), padding='valid', name='conv1')(inputs)\n    x = Activation('relu', name='relu_conv1')(x)\n    x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool1')(x)\n\n    x = fire_module(x, fire_id=2, squeeze=16, expand=64)\n    x = fire_module(x, fire_id=3, squeeze=16, expand=64)\n    x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool3')(x)\n\n    x = fire_module(x, fire_id=4, squeeze=32, expand=128)\n    x = fire_module(x, fire_id=5, squeeze=32, expand=128)\n    x = MaxPooling2D(pool_size=(3, 3), strides=(2, 2), name='pool5')(x)\n\n    x = fire_module(x, fire_id=6, squeeze=48, expand=192)\n    x = fire_module(x, fire_id=7, squeeze=48, expand=192)\n    x = fire_module(x, fire_id=8, squeeze=64, expand=256)\n    x = fire_module(x, fire_id=9, squeeze=64, expand=256)\n    \n    x = GlobalAveragePooling2D()(x)\n    x = Dense(1, activation='sigmoid')(x)\n\n    \n    model = Model(inputs, x, name='squeezenet')\n\n    return model","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"from keras.preprocessing.image import ImageDataGenerator","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"datagen = ImageDataGenerator()\ntrain_it = datagen.flow_from_directory('../train/', class_mode='binary', batch_size=128, target_size=(32,32))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"test_it = datagen.flow_from_directory('../test/', class_mode='binary', batch_size=128, target_size=(32,32))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = SqueezeNet()\nmodel.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.fit_generator(train_it, steps_per_epoch=128, epochs=5, validation_data=test_it, validation_steps=128)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model.","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}