{"cells":[{"metadata":{},"cell_type":"markdown","source":"# Cassava Leaf Disease Train Phase\n \nReferences, see also them:\n\n[Getting Started: TPUs + Cassava Leaf Disease](https://www.kaggle.com/jessemostipak/getting-started-tpus-cassava-leaf-disease)\n\n[CutMix and MixUp on GPU/TPU](https://www.kaggle.com/cdeotte/cutmix-and-mixup-on-gpu-tpu)\n\n[Getting started with 100+ flowers on TPU](https://www.kaggle.com/mgornergoogle/getting-started-with-100-flowers-on-tpu)\n\n[Triple Stratified KFold with TFRecords](https://www.kaggle.com/cdeotte/triple-stratified-kfold-with-tfrecords)"},{"metadata":{},"cell_type":"markdown","source":"# Set up environment"},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"import math, re, os, gc\nimport tensorflow as tf\nimport tensorflow.keras.backend as K\nimport numpy as np\nimport pandas as pd\nimport matplotlib.pyplot as plt\nfrom kaggle_datasets import KaggleDatasets\nfrom tensorflow import keras\nfrom functools import partial\nfrom sklearn.model_selection import train_test_split\nfrom sklearn.model_selection import KFold\nprint(\"Tensorflow version \" + tf.__version__)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Detect TPU"},{"metadata":{"_uuid":"d629ff2d2480ee46fbb7e2d37f6b5fab8052498a","_cell_guid":"79c7e3d0-c299-4dcb-8224-4455121ee9b0","trusted":true},"cell_type":"code","source":"try:\n    tpu = tf.distribute.cluster_resolver.TPUClusterResolver()\n    print('Device:', tpu.master())\n    tf.config.experimental_connect_to_cluster(tpu)\n    tf.tpu.experimental.initialize_tpu_system(tpu)\n    strategy = tf.distribute.experimental.TPUStrategy(tpu)\nexcept:\n    strategy = tf.distribute.get_strategy()\nprint('Number of replicas:', strategy.num_replicas_in_sync)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Set up variables"},{"metadata":{"trusted":true},"cell_type":"code","source":"AUTOTUNE = tf.data.experimental.AUTOTUNE\nGCS_PATH = KaggleDatasets().get_gcs_path('cassava-leaf-disease-classification')\nGCS_PATH_STRATIFICATED = KaggleDatasets().get_gcs_path('cassava-recreate-stratificated-tfrecords')\nBATCH_SIZE = 16 * strategy.num_replicas_in_sync\nAUG_BATCH = BATCH_SIZE\nIMAGE_SIZE = [512, 512]\nDIM = IMAGE_SIZE[0]\nCLASSES = ['0', '1', '2', '3', '4']\nEPOCHS = 30\nFOLDS = 5\nPHASE = 'train'\ndebug = True","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Define Augmentation"},{"metadata":{"trusted":true},"cell_type":"code","source":"ROT_ = 180.0\nSHR_ = 2.0\nHZOOM_ = 8.0\nWZOOM_ = 8.0\nHSHIFT_ = 8.0\nWSHIFT_ = 8.0","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_mat(rotation, shear, height_zoom, width_zoom, height_shift, width_shift):\n    # returns 3x3 transformmatrix which transforms indicies\n        \n    # CONVERT DEGREES TO RADIANS\n    rotation = math.pi * rotation / 180.\n    shear    = math.pi * shear    / 180.\n\n    def get_3x3_mat(lst):\n        return tf.reshape(tf.concat([lst],axis=0), [3,3])\n    \n    # ROTATION MATRIX\n    c1   = tf.math.cos(rotation)\n    s1   = tf.math.sin(rotation)\n    one  = tf.constant([1],dtype='float32')\n    zero = tf.constant([0],dtype='float32')\n    \n    rotation_matrix = get_3x3_mat([c1,   s1,   zero, \n                                   -s1,  c1,   zero, \n                                   zero, zero, one])    \n    # SHEAR MATRIX\n    c2 = tf.math.cos(shear)\n    s2 = tf.math.sin(shear)    \n    \n    shear_matrix = get_3x3_mat([one,  s2,   zero, \n                                zero, c2,   zero, \n                                zero, zero, one])        \n    # ZOOM MATRIX\n    zoom_matrix = get_3x3_mat([one/height_zoom, zero,           zero, \n                               zero,            one/width_zoom, zero, \n                               zero,            zero,           one])    \n    # SHIFT MATRIX\n    shift_matrix = get_3x3_mat([one,  zero, height_shift, \n                                zero, one,  width_shift, \n                                zero, zero, one])\n    \n    return K.dot(K.dot(rotation_matrix, shear_matrix), \n                 K.dot(zoom_matrix,     shift_matrix))\n\n\ndef transform_mat(image, DIM=IMAGE_SIZE[0]):    \n    # input image - is one image of size [dim,dim,3] not a batch of [b,dim,dim,3]\n    # output - image randomly rotated, sheared, zoomed, and shifted\n    XDIM = DIM%2 \n    \n    rot = ROT_ * tf.random.normal([1], dtype='float32')\n    shr = SHR_ * tf.random.normal([1], dtype='float32') \n    h_zoom = 1.0 + tf.random.normal([1], dtype='float32') / HZOOM_\n    w_zoom = 1.0 + tf.random.normal([1], dtype='float32') / WZOOM_\n    h_shift = HSHIFT_ * tf.random.normal([1], dtype='float32') \n    w_shift = WSHIFT_ * tf.random.normal([1], dtype='float32') \n\n    # GET TRANSFORMATION MATRIX\n    m = get_mat(rot,shr,h_zoom,w_zoom,h_shift,w_shift) \n\n    # LIST DESTINATION PIXEL INDICES\n    x   = tf.repeat(tf.range(DIM//2, -DIM//2,-1), DIM)\n    y   = tf.tile(tf.range(-DIM//2, DIM//2), [DIM])\n    z   = tf.ones([DIM*DIM], dtype='int32')\n    idx = tf.stack( [x,y,z] )\n    \n    # ROTATE DESTINATION PIXELS ONTO ORIGIN PIXELS\n    idx2 = K.dot(m, tf.cast(idx, dtype='float32'))\n    idx2 = K.cast(idx2, dtype='int32')\n    idx2 = K.clip(idx2, -DIM//2+XDIM+1, DIM//2)\n    \n    # FIND ORIGIN PIXEL VALUES           \n    idx3 = tf.stack([DIM//2-idx2[0,], DIM//2-1+idx2[1,]])\n    d    = tf.gather_nd(image, tf.transpose(idx3))\n        \n    return tf.reshape(d, [DIM, DIM,3])","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Decode the data"},{"metadata":{"trusted":true},"cell_type":"code","source":"# this code will convert our test image data to a float32 \ndef to_float32(image, label):\n    return tf.cast(image, tf.float32), label","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def decode_image(image):\n    image = tf.image.decode_jpeg(image, channels=3)\n    image = tf.cast(image, tf.float32) / 255.0\n    image = tf.reshape(image, [*IMAGE_SIZE, 3])\n    return image","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def read_tfrecord(example, labeled):\n    tfrecord_format = {\n        \"image\": tf.io.FixedLenFeature([], tf.string),\n        \"target\": tf.io.FixedLenFeature([], tf.int64)\n    } if labeled else {\n        \"image\": tf.io.FixedLenFeature([], tf.string),\n        \"image_name\": tf.io.FixedLenFeature([], tf.string)\n    }\n    example = tf.io.parse_single_example(example, tfrecord_format)\n    image = decode_image(example['image'])\n    if labeled:\n        label = tf.cast(example['target'], tf.int32)\n        return image, label\n    idnum = example['image_name']\n    return image, idnum","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def load_dataset(filenames, labeled=True, ordered=False):\n    ignore_order = tf.data.Options()\n    if not ordered:\n        ignore_order.experimental_deterministic = False # disable order, increase speed\n    dataset = tf.data.TFRecordDataset(filenames, num_parallel_reads=AUTOTUNE) # automatically interleaves reads from multiple files\n    dataset = dataset.with_options(ignore_order) # uses data as soon as it streams in, rather than in its original order\n    dataset = dataset.map(partial(read_tfrecord, labeled=labeled), num_parallel_calls=AUTOTUNE)\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"TRAINING_FILENAMES = tf.io.gfile.glob('../input/cassava-leaf-disease-classification/train_tfrecords/' + '*train*.tfrec')\nTEST_FILENAMES = tf.io.gfile.glob('../input/cassava-leaf-disease-classification/test_tfrecords/' + 'ld_test*.tfrec')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def count_data_items(filenames):\n    # the number of data items is written in the name of the .tfrec files, i.e. flowers00-230.tfrec = 230 data items\n    n = [int(re.compile(r\"-([0-9]*)\\.\").search(filename).group(1)) for filename in filenames]\n    return np.sum(n)\n\nNUM_TRAINING_IMAGES = int( count_data_items(TRAINING_FILENAMES) * (FOLDS-1.)/FOLDS )\nNUM_VALIDATION_IMAGES = int( count_data_items(TRAINING_FILENAMES) * (1./FOLDS) )\nNUM_TEST_IMAGES = count_data_items(TEST_FILENAMES)\nSTEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE\n\nprint('Dataset: {} training images, {} validation images, {} unlabeled test images'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES, NUM_TEST_IMAGES))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"TRAINING_FILENAMES, VALID_FILENAMES = train_test_split(\n    tf.io.gfile.glob(GCS_PATH + '/train_tfrecords/ld_train*.tfrec'),\n    test_size=0.35, random_state=5\n)\n\nTEST_FILENAMES = tf.io.gfile.glob(GCS_PATH + '/test_tfrecords/ld_test*.tfrec')","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"#print(TRAINING_FILENAMES)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def data_augment(image, label):\n    # Thanks to the dataset.prefetch(AUTO) statement in the following function this happens essentially for free on TPU. \n    # Data pipeline code is executed on the \"CPU\" part of the TPU while the TPU itself is computing gradients.\n    image = tf.image.random_flip_left_right(image)\n    return image, label","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Define data loading methods"},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_training_dataset(training_fikenames=TRAINING_FILENAMES):\n    dataset = load_dataset(training_fikenames, labeled=True)  \n    dataset = dataset.map(data_augment, num_parallel_calls=AUTOTUNE)  \n    dataset = dataset.repeat()\n    dataset = dataset.shuffle(2048)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.prefetch(AUTOTUNE)\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_validation_dataset(valid_filenames=VALID_FILENAMES, ordered=False):\n    dataset = load_dataset(valid_filenames, labeled=True, ordered=ordered) \n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.cache()\n    dataset = dataset.prefetch(AUTOTUNE)\n    return dataset\n","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_test_dataset(ordered=False):\n    dataset = load_dataset(TEST_FILENAMES, labeled=False, ordered=ordered)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.prefetch(AUTOTUNE)\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def count_data_items(filenames):\n    n = [int(re.compile(r\"-([0-9]*)\\.\").search(filename).group(1)) for filename in filenames]\n    return np.sum(n)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Brief exploratory data analysis (EDA)"},{"metadata":{"trusted":true},"cell_type":"code","source":"print(\"Training data shapes:\")\nfor image, label in get_training_dataset().take(3):\n    print(image.numpy().shape, label.numpy().shape)\nprint(\"Training data label examples:\", label.numpy())\nprint(\"Validation data shapes:\")\nfor image, label in get_validation_dataset().take(3):\n    print(image.numpy().shape, label.numpy().shape)\nprint(\"Validation data label examples:\", label.numpy())\nprint(\"Test data shapes:\")\nfor image, idnum in get_test_dataset().take(3):\n    print(image.numpy().shape, idnum.numpy().shape)\nprint(\"Test data IDs:\", idnum.numpy().astype('U')) # U=unicode string","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# numpy and matplotlib defaults\nnp.set_printoptions(threshold=15, linewidth=80)\n\ndef batch_to_numpy_images_and_labels(data):\n    images, labels = data\n    numpy_images = images.numpy()\n    numpy_labels = labels.numpy()\n    if numpy_labels.dtype == object: # binary string in this case, these are image ID strings\n        numpy_labels = [None for _ in enumerate(numpy_images)]\n    # If no labels, only image IDs, return None for labels (this is the case for test data)\n    return numpy_images, numpy_labels\n\ndef title_from_label_and_target(label, correct_label):\n    if correct_label is None:\n        return CLASSES[label], True\n    correct = (label == correct_label)\n    return \"{} [{}{}{}]\".format(CLASSES[label], 'OK' if correct else 'NO', u\"\\u2192\" if not correct else '',\n                                CLASSES[correct_label] if not correct else ''), correct\n\ndef display_one_plant(image, title, subplot, red=False, titlesize=16):\n    plt.subplot(*subplot)\n    plt.axis('off')\n    plt.imshow(image)\n    if len(title) > 0:\n        plt.title(title, fontsize=int(titlesize) if not red else int(titlesize/1.2), color='red' if red else 'black', fontdict={'verticalalignment':'center'}, pad=int(titlesize/1.5))\n    return (subplot[0], subplot[1], subplot[2]+1)\n\ndef display_batch_of_images(databatch, predictions=None):\n    \"\"\"This will work with:\n    display_batch_of_images(images)\n    display_batch_of_images(images, predictions)\n    display_batch_of_images((images, labels))\n    display_batch_of_images((images, labels), predictions)\n    \"\"\"\n    # data\n    images, labels = batch_to_numpy_images_and_labels(databatch)\n    if labels is None:\n        labels = [None for _ in enumerate(images)]\n        \n    # auto-squaring: this will drop data that does not fit into square or square-ish rectangle\n    rows = int(math.sqrt(len(images)))\n    cols = len(images)//rows\n        \n    # size and spacing\n    FIGSIZE = 13.0\n    SPACING = 0.1\n    subplot=(rows,cols,1)\n    if rows < cols:\n        plt.figure(figsize=(FIGSIZE,FIGSIZE/cols*rows))\n    else:\n        plt.figure(figsize=(FIGSIZE/rows*cols,FIGSIZE))\n    \n    # display\n    for i, (image, label) in enumerate(zip(images[:rows*cols], labels[:rows*cols])):\n        title = '' if label is None else CLASSES[label]\n        correct = True\n        if predictions is not None:\n            title, correct = title_from_label_and_target(predictions[i], label)\n        dynamic_titlesize = FIGSIZE*SPACING/max(rows,cols)*40+3 # magic formula tested to work from 1x1 to 10x10 images\n        subplot = display_one_plant(image, title, subplot, not correct, titlesize=dynamic_titlesize)\n    \n    #layout\n    plt.tight_layout()\n    if label is None and predictions is None:\n        plt.subplots_adjust(wspace=0, hspace=0)\n    else:\n        plt.subplots_adjust(wspace=SPACING, hspace=SPACING)\n    plt.show()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# load our training dataset for EDA\ntraining_dataset = get_training_dataset()\ntraining_dataset = training_dataset.unbatch().batch(20)\ntrain_batch = iter(training_dataset)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# run this cell again for another randomized set of training images\ndisplay_batch_of_images(next(train_batch))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# load our validation dataset for EDA\nvalidation_dataset = get_validation_dataset()\nvalidation_dataset = validation_dataset.unbatch().batch(20)\nvalid_batch = iter(validation_dataset)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# run this cell again for another randomized set of training images\ndisplay_batch_of_images(next(valid_batch))","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# load our test dataset for EDA\ntesting_dataset = get_test_dataset()\ntesting_dataset = testing_dataset.unbatch().batch(20)\ntest_batch = iter(testing_dataset)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# we only have one test image\ndisplay_batch_of_images(next(test_batch))","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# CutMix and MixUp Augmentation"},{"metadata":{"trusted":true},"cell_type":"code","source":"def onehot(image,label):\n    CLASSES = 5\n    return image,tf.one_hot(label,CLASSES)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def cutmix(image, label, PROBABILITY = 1.0):\n    # input image - is a batch of images of size [n,dim,dim,3] not a single image of [dim,dim,3]\n    # output - a batch of images with cutmix applied\n    DIM = IMAGE_SIZE[0]\n    CLASSES = 5\n    \n    imgs = []; labs = []\n    for j in range(AUG_BATCH):\n        # DO CUTMIX WITH PROBABILITY DEFINED ABOVE\n        P = tf.cast( tf.random.uniform([],0,1)<=PROBABILITY, tf.int32)\n        # CHOOSE RANDOM IMAGE TO CUTMIX WITH\n        k = tf.cast( tf.random.uniform([],0,AUG_BATCH),tf.int32)\n        # CHOOSE RANDOM LOCATION\n        x = tf.cast( tf.random.uniform([],0,DIM),tf.int32)\n        y = tf.cast( tf.random.uniform([],0,DIM),tf.int32)\n        b = tf.random.uniform([],0,1) # this is beta dist with alpha=1.0\n        WIDTH = tf.cast( DIM * tf.math.sqrt(1-b),tf.int32) * P\n        ya = tf.math.maximum(0,y-WIDTH//2)\n        yb = tf.math.minimum(DIM,y+WIDTH//2)\n        xa = tf.math.maximum(0,x-WIDTH//2)\n        xb = tf.math.minimum(DIM,x+WIDTH//2)\n        # MAKE CUTMIX IMAGE\n        one = image[j,ya:yb,0:xa,:]\n        two = image[k,ya:yb,xa:xb,:]\n        three = image[j,ya:yb,xb:DIM,:]\n        middle = tf.concat([one,two,three],axis=1)\n        img = tf.concat([image[j,0:ya,:,:],middle,image[j,yb:DIM,:,:]],axis=0)\n        imgs.append(img)\n        # MAKE CUTMIX LABEL\n        a = tf.cast(WIDTH*WIDTH/DIM/DIM,tf.float32)\n        if len(label.shape)==1:\n            lab1 = tf.one_hot(label[j],CLASSES)\n            lab2 = tf.one_hot(label[k],CLASSES)\n        else:\n            lab1 = label[j,]\n            lab2 = label[k,]\n        labs.append((1-a)*lab1 + a*lab2)\n            \n    # RESHAPE HACK SO TPU COMPILER KNOWS SHAPE OF OUTPUT TENSOR (maybe use Python typing instead?)\n    image2 = tf.reshape(tf.stack(imgs),(AUG_BATCH,DIM,DIM,3))\n    label2 = tf.reshape(tf.stack(labs),(AUG_BATCH,CLASSES))\n    return image2,label2","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def mixup(image, label, PROBABILITY = 1.0):\n    # input image - is a batch of images of size [n,dim,dim,3] not a single image of [dim,dim,3]\n    # output - a batch of images with mixup applied\n    DIM = IMAGE_SIZE[0]\n    CLASSES = 5\n    \n    imgs = []; labs = []\n    for j in range(AUG_BATCH):\n        # DO MIXUP WITH PROBABILITY DEFINED ABOVE\n        P = tf.cast( tf.random.uniform([],0,1)<=PROBABILITY, tf.float32)\n        # CHOOSE RANDOM\n        k = tf.cast( tf.random.uniform([],0,AUG_BATCH),tf.int32)\n        a = tf.random.uniform([],0,1)*P # this is beta dist with alpha=1.0\n        # MAKE MIXUP IMAGE\n        img1 = image[j,]\n        img2 = image[k,]\n        imgs.append((1-a)*img1 + a*img2)\n        # MAKE CUTMIX LABEL\n        if len(label.shape)==1:\n            lab1 = tf.one_hot(label[j],CLASSES)\n            lab2 = tf.one_hot(label[k],CLASSES)\n        else:\n            lab1 = label[j,]\n            lab2 = label[k,]\n        labs.append((1-a)*lab1 + a*lab2)\n            \n    # RESHAPE HACK SO TPU COMPILER KNOWS SHAPE OF OUTPUT TENSOR (maybe use Python typing instead?)\n    image2 = tf.reshape(tf.stack(imgs),(AUG_BATCH,DIM,DIM,3))\n    label2 = tf.reshape(tf.stack(labs),(AUG_BATCH,CLASSES))\n    return image2,label2","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def transform(image,label):\n    # THIS FUNCTION APPLIES BOTH CUTMIX AND MIXUP\n    DIM = IMAGE_SIZE[0]\n    CLASSES = 5\n    SWITCH = 0.5\n    CUTMIX_PROB = 0.666\n    MIXUP_PROB = 0.666\n    # FOR SWITCH PERCENT OF TIME WE DO CUTMIX AND (1-SWITCH) WE DO MIXUP\n    image1 = []\n    for j in range(AUG_BATCH):\n        img = transform_mat(image[j,])\n        img = tf.image.random_flip_left_right(img)\n        img = tf.image.random_saturation(img, 0.7, 1.3)\n        img = tf.image.random_contrast(img, 0.8, 1.2)\n        img = tf.image.random_brightness(img, 0.1)\n        image1.append(img)\n        \n    image1 = tf.reshape(tf.stack(image1),(AUG_BATCH,DIM,DIM,3))\n    image2, label2 = cutmix(image1, label, CUTMIX_PROB)\n    image3, label3 = mixup(image1, label, MIXUP_PROB)\n    imgs = []; labs = []\n    for j in range(AUG_BATCH):\n        P = tf.cast( tf.random.uniform([],0,1)<=SWITCH, tf.float32)\n        imgs.append(P*image2[j,]+(1-P)*image3[j,])\n        labs.append(P*label2[j,]+(1-P)*label3[j,])\n    # RESHAPE HACK SO TPU COMPILER KNOWS SHAPE OF OUTPUT TENSOR (maybe use Python typing instead?)\n    image4 = tf.reshape(tf.stack(imgs),(AUG_BATCH,DIM,DIM,3))\n    label4 = tf.reshape(tf.stack(labs),(AUG_BATCH,CLASSES))\n    return image4,label4","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_training_dataset(dataset=TRAINING_FILENAMES, do_aug=True):\n    #dataset = load_dataset(dataset, labeled=True)  \n    dataset = dataset.map(data_augment, num_parallel_calls=AUTOTUNE)\n    dataset = dataset.repeat() # the training dataset must repeat for several epochs\n    dataset = dataset.batch(AUG_BATCH)\n    if do_aug: dataset = dataset.map(transform, num_parallel_calls=AUTOTUNE) # note we put AFTER batching\n    dataset = dataset.unbatch()\n    dataset = dataset.shuffle(2048)\n    dataset = dataset.batch(BATCH_SIZE)\n    dataset = dataset.prefetch(AUTOTUNE) # prefetch next batch while training (autotune prefetch buffer size)\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_validation_dataset(dataset, do_onehot=True):\n    dataset = dataset.batch(BATCH_SIZE)\n    if do_onehot: dataset = dataset.map(onehot, num_parallel_calls=AUTOTUNE) # we must use one hot like augmented train data\n    dataset = dataset.cache()\n    dataset = dataset.prefetch(AUTOTUNE) # prefetch next batch while training (autotune prefetch buffer size)\n    return dataset","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if debug:\n    row = 6; col = 4;\n    row = min(row,AUG_BATCH//col)\n    all_elements = get_training_dataset(load_dataset(TRAINING_FILENAMES),do_aug=False).unbatch()\n    augmented_element = all_elements.repeat().batch(AUG_BATCH).map(transform)\n\n    for (img,label) in augmented_element:\n        plt.figure(figsize=(15,int(15*row/col)))\n        for j in range(row*col):\n            plt.subplot(row,col,j+1)\n            plt.axis('off')\n            plt.imshow(img[j,])\n        plt.show()\n        break","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Building the model\n## Learning rate schedule"},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_lr_callback(batch_size=8):\n    lr_start   = 0.000005\n    lr_max     = 0.00000125 * strategy.num_replicas_in_sync * batch_size\n    lr_min     = 0.000001\n    lr_ramp_ep = 5\n    lr_sus_ep  = 0\n    lr_decay   = 0.8\n   \n    def lrfn(epoch):\n        if epoch < lr_ramp_ep:\n            lr = (lr_max - lr_start) / lr_ramp_ep * epoch + lr_start\n            \n        elif epoch < lr_ramp_ep + lr_sus_ep:\n            lr = lr_max\n            \n        else:\n            lr = (lr_max - lr_min) * lr_decay**(epoch - lr_ramp_ep - lr_sus_ep) + lr_min\n            \n        return lr\n\n    lr_callback = tf.keras.callbacks.LearningRateScheduler(lrfn, verbose=False)\n    return lr_callback","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## Building our model"},{"metadata":{"_kg_hide-input":true,"trusted":true},"cell_type":"code","source":"# Copyright 2019 The TensorFlow Authors. All Rights Reserved.\n#\n# Licensed under the Apache License, Version 2.0 (the \"License\");\n# you may not use this file except in compliance with the License.\n# You may obtain a copy of the License at\n#\n#     http://www.apache.org/licenses/LICENSE-2.0\n#\n# Unless required by applicable law or agreed to in writing, software\n# distributed under the License is distributed on an \"AS IS\" BASIS,\n# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.\n# See the License for the specific language governing permissions and\n# limitations under the License.\n# ==============================================================================\n# pylint: disable=invalid-name\n\"\"\"EfficientNet models for Keras.\n\nReference paper:\n  - [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks]\n    (https://arxiv.org/abs/1905.11946) (ICML 2019)\n\"\"\"\nfrom __future__ import absolute_import\nfrom __future__ import division\nfrom __future__ import print_function\n\nimport copy\nimport math\nimport os\n\nfrom tensorflow.python.keras import backend\nfrom tensorflow.python.keras import layers\nfrom tensorflow.python.keras.applications import imagenet_utils\nfrom tensorflow.python.keras.engine import training\nfrom tensorflow.python.keras.utils import data_utils\nfrom tensorflow.python.keras.utils import layer_utils\nfrom tensorflow.python.util.tf_export import keras_export\n\n\nBASE_WEIGHTS_PATH = 'https://storage.googleapis.com/keras-applications/'\n\nWEIGHTS_HASHES = {\n    'b0': ('902e53a9f72be733fc0bcb005b3ebbac',\n           '50bc09e76180e00e4465e1a485ddc09d'),\n    'b1': ('1d254153d4ab51201f1646940f018540',\n           '74c4e6b3e1f6a1eea24c589628592432'),\n    'b2': ('b15cce36ff4dcbd00b6dd88e7857a6ad',\n           '111f8e2ac8aa800a7a99e3239f7bfb39'),\n    'b3': ('ffd1fdc53d0ce67064dc6a9c7960ede0',\n           'af6d107764bb5b1abb91932881670226'),\n    'b4': ('18c95ad55216b8f92d7e70b3a046e2fc',\n           'ebc24e6d6c33eaebbd558eafbeedf1ba'),\n    'b5': ('ace28f2a6363774853a83a0b21b9421a',\n           '38879255a25d3c92d5e44e04ae6cec6f'),\n    'b6': ('165f6e37dce68623721b423839de8be5',\n           '9ecce42647a20130c1f39a5d4cb75743'),\n    'b7': ('8c03f828fec3ef71311cd463b6759d99',\n           'cbcfe4450ddf6f3ad90b1b398090fe4a'),\n}\n\nDEFAULT_BLOCKS_ARGS = [{\n    'kernel_size': 3,\n    'repeats': 1,\n    'filters_in': 32,\n    'filters_out': 16,\n    'expand_ratio': 1,\n    'id_skip': True,\n    'strides': 1,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 3,\n    'repeats': 2,\n    'filters_in': 16,\n    'filters_out': 24,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 2,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 5,\n    'repeats': 2,\n    'filters_in': 24,\n    'filters_out': 40,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 2,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 3,\n    'repeats': 3,\n    'filters_in': 40,\n    'filters_out': 80,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 2,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 5,\n    'repeats': 3,\n    'filters_in': 80,\n    'filters_out': 112,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 1,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 5,\n    'repeats': 4,\n    'filters_in': 112,\n    'filters_out': 192,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 2,\n    'se_ratio': 0.25\n}, {\n    'kernel_size': 3,\n    'repeats': 1,\n    'filters_in': 192,\n    'filters_out': 320,\n    'expand_ratio': 6,\n    'id_skip': True,\n    'strides': 1,\n    'se_ratio': 0.25\n}]\n\nCONV_KERNEL_INITIALIZER = {\n    'class_name': 'VarianceScaling',\n    'config': {\n        'scale': 2.0,\n        'mode': 'fan_out',\n        'distribution': 'truncated_normal'\n    }\n}\n\nDENSE_KERNEL_INITIALIZER = {\n    'class_name': 'VarianceScaling',\n    'config': {\n        'scale': 1. / 3.,\n        'mode': 'fan_out',\n        'distribution': 'uniform'\n    }\n}\n\n\ndef EfficientNet(\n    width_coefficient,\n    depth_coefficient,\n    default_size,\n    dropout_rate=0.2,\n    drop_connect_rate=0.2,\n    depth_divisor=8,\n    activation='swish',\n    blocks_args='default',\n    model_name='efficientnet',\n    include_top=True,\n    weights='imagenet',\n    input_tensor=None,\n    input_shape=None,\n    pooling=None,\n    classes=1000,\n    classifier_activation='softmax',\n):\n  \"\"\"Instantiates the EfficientNet architecture using given scaling coefficients.\n\n  Optionally loads weights pre-trained on ImageNet.\n  Note that the data format convention used by the model is\n  the one specified in your Keras config at `~/.keras/keras.json`.\n\n  Arguments:\n    width_coefficient: float, scaling coefficient for network width.\n    depth_coefficient: float, scaling coefficient for network depth.\n    default_size: integer, default input image size.\n    dropout_rate: float, dropout rate before final classifier layer.\n    drop_connect_rate: float, dropout rate at skip connections.\n    depth_divisor: integer, a unit of network width.\n    activation: activation function.\n    blocks_args: list of dicts, parameters to construct block modules.\n    model_name: string, model name.\n    include_top: whether to include the fully-connected\n        layer at the top of the network.\n    weights: one of `None` (random initialization),\n          'imagenet' (pre-training on ImageNet),\n          or the path to the weights file to be loaded.\n    input_tensor: optional Keras tensor\n        (i.e. output of `layers.Input()`)\n        to use as image input for the model.\n    input_shape: optional shape tuple, only to be specified\n        if `include_top` is False.\n        It should have exactly 3 inputs channels.\n    pooling: optional pooling mode for feature extraction\n        when `include_top` is `False`.\n        - `None` means that the output of the model will be\n            the 4D tensor output of the\n            last convolutional layer.\n        - `avg` means that global average pooling\n            will be applied to the output of the\n            last convolutional layer, and thus\n            the output of the model will be a 2D tensor.\n        - `max` means that global max pooling will\n            be applied.\n    classes: optional number of classes to classify images\n        into, only to be specified if `include_top` is True, and\n        if no `weights` argument is specified.\n    classifier_activation: A `str` or callable. The activation function to use\n        on the \"top\" layer. Ignored unless `include_top=True`. Set\n        `classifier_activation=None` to return the logits of the \"top\" layer.\n\n  Returns:\n    A `keras.Model` instance.\n\n  Raises:\n    ValueError: in case of invalid argument for `weights`,\n      or invalid input shape.\n    ValueError: if `classifier_activation` is not `softmax` or `None` when\n      using a pretrained top layer.\n  \"\"\"\n  if blocks_args == 'default':\n    blocks_args = DEFAULT_BLOCKS_ARGS\n\n  if not (weights in {'imagenet', None} or os.path.exists(weights)):\n    raise ValueError('The `weights` argument should be either '\n                     '`None` (random initialization), `imagenet` '\n                     '(pre-training on ImageNet), '\n                     'or the path to the weights file to be loaded.')\n\n  if weights == 'imagenet' and include_top and classes != 1000:\n    raise ValueError('If using `weights` as `\"imagenet\"` with `include_top`'\n                     ' as true, `classes` should be 1000')\n\n  # Determine proper input shape\n  input_shape = imagenet_utils.obtain_input_shape(\n      input_shape,\n      default_size=default_size,\n      min_size=32,\n      data_format=backend.image_data_format(),\n      require_flatten=include_top,\n      weights=weights)\n\n  if input_tensor is None:\n    img_input = layers.Input(shape=input_shape)\n  else:\n    if not backend.is_keras_tensor(input_tensor):\n      img_input = layers.Input(tensor=input_tensor, shape=input_shape)\n    else:\n      img_input = input_tensor\n\n  bn_axis = 3 if backend.image_data_format() == 'channels_last' else 1\n\n  def round_filters(filters, divisor=depth_divisor):\n    \"\"\"Round number of filters based on depth multiplier.\"\"\"\n    filters *= width_coefficient\n    new_filters = max(divisor, int(filters + divisor / 2) // divisor * divisor)\n    # Make sure that round down does not go down by more than 10%.\n    if new_filters < 0.9 * filters:\n      new_filters += divisor\n    return int(new_filters)\n\n  def round_repeats(repeats):\n    \"\"\"Round number of repeats based on depth multiplier.\"\"\"\n    return int(math.ceil(depth_coefficient * repeats))\n\n  # Build stem\n  x = img_input\n  #x = layers.Rescaling(1. / 255.)(x)\n  x = layers.Normalization(axis=bn_axis)(x)\n\n  x = layers.ZeroPadding2D(\n      padding=imagenet_utils.correct_pad(x, 3),\n      name='stem_conv_pad')(x)\n  x = layers.Conv2D(\n      round_filters(32),\n      3,\n      strides=2,\n      padding='valid',\n      use_bias=False,\n      kernel_initializer=CONV_KERNEL_INITIALIZER,\n      name='stem_conv')(x)\n  x = layers.BatchNormalization(axis=bn_axis, name='stem_bn')(x)\n  x = layers.Activation(activation, name='stem_activation')(x)\n\n  # Build blocks\n  blocks_args = copy.deepcopy(blocks_args)\n\n  b = 0\n  blocks = float(sum(args['repeats'] for args in blocks_args))\n  for (i, args) in enumerate(blocks_args):\n    assert args['repeats'] > 0\n    # Update block input and output filters based on depth multiplier.\n    args['filters_in'] = round_filters(args['filters_in'])\n    args['filters_out'] = round_filters(args['filters_out'])\n\n    for j in range(round_repeats(args.pop('repeats'))):\n      # The first block needs to take care of stride and filter size increase.\n      if j > 0:\n        args['strides'] = 1\n        args['filters_in'] = args['filters_out']\n      x = block(\n          x,\n          activation,\n          drop_connect_rate * b / blocks,\n          name='block{}{}_'.format(i + 1, chr(j + 97)),\n          **args)\n      b += 1\n\n  # Build top\n  x = layers.Conv2D(\n      round_filters(1280),\n      1,\n      padding='same',\n      use_bias=False,\n      kernel_initializer=CONV_KERNEL_INITIALIZER,\n      name='top_conv')(x)\n  x = layers.BatchNormalization(axis=bn_axis, name='top_bn')(x)\n  x = layers.Activation(activation, name='top_activation')(x)\n  if include_top:\n    x = layers.GlobalAveragePooling2D(name='avg_pool')(x)\n    if dropout_rate > 0:\n      x = layers.Dropout(dropout_rate, name='top_dropout')(x)\n    imagenet_utils.validate_activation(classifier_activation, weights)\n    x = layers.Dense(\n        classes,\n        activation=classifier_activation,\n        kernel_initializer=DENSE_KERNEL_INITIALIZER,\n        name='predictions')(x)\n  else:\n    if pooling == 'avg':\n      x = layers.GlobalAveragePooling2D(name='avg_pool')(x)\n    elif pooling == 'max':\n      x = layers.GlobalMaxPooling2D(name='max_pool')(x)\n\n  # Ensure that the model takes into account\n  # any potential predecessors of `input_tensor`.\n  if input_tensor is not None:\n    inputs = layer_utils.get_source_inputs(input_tensor)\n  else:\n    inputs = img_input\n\n  # Create model.\n  model = training.Model(inputs, x, name=model_name)\n\n  # Load weights.\n  if weights == 'imagenet':\n    if include_top:\n      file_suffix = '.h5'\n      file_hash = WEIGHTS_HASHES[model_name[-2:]][0]\n    else:\n      file_suffix = '_notop.h5'\n      file_hash = WEIGHTS_HASHES[model_name[-2:]][1]\n    file_name = model_name + file_suffix\n    weights_path = data_utils.get_file(\n        file_name,\n        BASE_WEIGHTS_PATH + file_name,\n        cache_subdir='models',\n        file_hash=file_hash)\n    model.load_weights(weights_path)\n  elif weights is not None:\n    model.load_weights(weights)\n  return model\n\n\ndef block(inputs,\n          activation='swish',\n          drop_rate=0.,\n          name='',\n          filters_in=32,\n          filters_out=16,\n          kernel_size=3,\n          strides=1,\n          expand_ratio=1,\n          se_ratio=0.,\n          id_skip=True):\n  \"\"\"An inverted residual block.\n\n  Arguments:\n      inputs: input tensor.\n      activation: activation function.\n      drop_rate: float between 0 and 1, fraction of the input units to drop.\n      name: string, block label.\n      filters_in: integer, the number of input filters.\n      filters_out: integer, the number of output filters.\n      kernel_size: integer, the dimension of the convolution window.\n      strides: integer, the stride of the convolution.\n      expand_ratio: integer, scaling coefficient for the input filters.\n      se_ratio: float between 0 and 1, fraction to squeeze the input filters.\n      id_skip: boolean.\n\n  Returns:\n      output tensor for the block.\n  \"\"\"\n  bn_axis = 3 if backend.image_data_format() == 'channels_last' else 1\n\n  # Expansion phase\n  filters = filters_in * expand_ratio\n  if expand_ratio != 1:\n    x = layers.Conv2D(\n        filters,\n        1,\n        padding='same',\n        use_bias=False,\n        kernel_initializer=CONV_KERNEL_INITIALIZER,\n        name=name + 'expand_conv')(\n            inputs)\n    x = layers.BatchNormalization(axis=bn_axis, name=name + 'expand_bn')(x)\n    x = layers.Activation(activation, name=name + 'expand_activation')(x)\n  else:\n    x = inputs\n\n  # Depthwise Convolution\n  if strides == 2:\n    x = layers.ZeroPadding2D(\n        padding=imagenet_utils.correct_pad(x, kernel_size),\n        name=name + 'dwconv_pad')(x)\n    conv_pad = 'valid'\n  else:\n    conv_pad = 'same'\n  x = layers.DepthwiseConv2D(\n      kernel_size,\n      strides=strides,\n      padding=conv_pad,\n      use_bias=False,\n      depthwise_initializer=CONV_KERNEL_INITIALIZER,\n      name=name + 'dwconv')(x)\n  x = layers.BatchNormalization(axis=bn_axis, name=name + 'bn')(x)\n  x = layers.Activation(activation, name=name + 'activation')(x)\n\n  # Squeeze and Excitation phase\n  if 0 < se_ratio <= 1:\n    filters_se = max(1, int(filters_in * se_ratio))\n    se = layers.GlobalAveragePooling2D(name=name + 'se_squeeze')(x)\n    se = layers.Reshape((1, 1, filters), name=name + 'se_reshape')(se)\n    se = layers.Conv2D(\n        filters_se,\n        1,\n        padding='same',\n        activation=activation,\n        kernel_initializer=CONV_KERNEL_INITIALIZER,\n        name=name + 'se_reduce')(\n            se)\n    se = layers.Conv2D(\n        filters,\n        1,\n        padding='same',\n        activation='sigmoid',\n        kernel_initializer=CONV_KERNEL_INITIALIZER,\n        name=name + 'se_expand')(se)\n    x = layers.multiply([x, se], name=name + 'se_excite')\n\n  # Output phase\n  x = layers.Conv2D(\n      filters_out,\n      1,\n      padding='same',\n      use_bias=False,\n      kernel_initializer=CONV_KERNEL_INITIALIZER,\n      name=name + 'project_conv')(x)\n  x = layers.BatchNormalization(axis=bn_axis, name=name + 'project_bn')(x)\n  if id_skip and strides == 1 and filters_in == filters_out:\n    if drop_rate > 0:\n      x = layers.Dropout(\n          drop_rate, noise_shape=(None, 1, 1, 1), name=name + 'drop')(x)\n    x = layers.add([x, inputs], name=name + 'add')\n  return x\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB0',\n              'keras.applications.EfficientNetB0')\ndef EfficientNetB0(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.0,\n      1.0,\n      224,\n      0.2,\n      model_name='efficientnetb0',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB1',\n              'keras.applications.EfficientNetB1')\ndef EfficientNetB1(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.0,\n      1.1,\n      240,\n      0.2,\n      model_name='efficientnetb1',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB2',\n              'keras.applications.EfficientNetB2')\ndef EfficientNetB2(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.1,\n      1.2,\n      260,\n      0.3,\n      model_name='efficientnetb2',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB3',\n              'keras.applications.EfficientNetB3')\ndef EfficientNetB3(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.2,\n      1.4,\n      300,\n      0.3,\n      model_name='efficientnetb3',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB4',\n              'keras.applications.EfficientNetB4')\ndef EfficientNetB4(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.4,\n      1.8,\n      380,\n      0.4,\n      model_name='efficientnetb4',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB5',\n              'keras.applications.EfficientNetB5')\ndef EfficientNetB5(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.6,\n      2.2,\n      456,\n      0.4,\n      model_name='efficientnetb5',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB6',\n              'keras.applications.EfficientNetB6')\ndef EfficientNetB6(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      1.8,\n      2.6,\n      528,\n      0.5,\n      model_name='efficientnetb6',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.EfficientNetB7',\n              'keras.applications.EfficientNetB7')\ndef EfficientNetB7(include_top=True,\n                   weights='imagenet',\n                   input_tensor=None,\n                   input_shape=None,\n                   pooling=None,\n                   classes=1000,\n                   **kwargs):\n  return EfficientNet(\n      2.0,\n      3.1,\n      600,\n      0.5,\n      model_name='efficientnetb7',\n      include_top=include_top,\n      weights=weights,\n      input_tensor=input_tensor,\n      input_shape=input_shape,\n      pooling=pooling,\n      classes=classes,\n      **kwargs)\n\n\n@keras_export('keras.applications.efficientnet.preprocess_input')\ndef preprocess_input(x, data_format=None):  # pylint: disable=unused-argument\n  return x\n\n\n@keras_export('keras.applications.efficientnet.decode_predictions')\ndef decode_predictions(preds, top=5):\n  \"\"\"Decodes the prediction result from the model.\n\n  Arguments\n    preds: Numpy tensor encoding a batch of predictions.\n    top: Integer, how many top-guesses to return.\n\n  Returns\n    A list of lists of top class prediction tuples\n    `(class_name, class_description, score)`.\n    One list of tuples per sample in batch input.\n\n  Raises\n    ValueError: In case of invalid shape of the `preds` array (must be 2D).\n  \"\"\"\n  return imagenet_utils.decode_predictions(preds, top=top)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def get_model(weights='imagenet'):\n    with strategy.scope():       \n        inp = tf.keras.layers.Input(shape=(DIM,DIM,3))\n        base = EfficientNetB3(input_shape=(DIM,DIM,3),weights='imagenet',include_top=False)\n        x = base(inp)\n        x = tf.keras.layers.GlobalAveragePooling2D()(x)\n        x = tf.keras.layers.Dense(5,activation='softmax')(x)\n        model = tf.keras.Model(inputs=inp, outputs=x)\n        opt = tf.keras.optimizers.Adam(learning_rate=0.001)\n        loss = tf.keras.losses.CategoricalCrossentropy(\n        from_logits=False, label_smoothing=0.0001,\n        name='categorical_crossentropy'\n        )\n        model.compile(optimizer=opt,loss=loss,metrics=['categorical_accuracy'])\n    return model","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"skf = KFold(n_splits=FOLDS,shuffle=True,random_state=12)\nfor fold,(idxT,idxV) in enumerate(skf.split(np.arange(5))):\n    if fold==(FOLDS-1):\n        idxTT = idxT; idxVV = idxV\n        print('### Using fold',fold,'for experiments')\n    print('Fold',fold,'has TRAIN:',idxT,'VALID:',idxV)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"probabilities = np.zeros((count_data_items(TEST_FILENAMES),1))\nfor fold,(idxT,idxV) in enumerate(skf.split(np.arange(5))):\n    print(); print('#'*25)\n    print('### FOLD',fold+1)\n    print('#'*25)\n    if PHASE == 'train':\n        files_train = tf.io.gfile.glob([GCS_PATH_STRATIFICATED + '/train%.2i*.tfrec'%x for x in idxT])\n        files_valid = tf.io.gfile.glob([GCS_PATH_STRATIFICATED + '/train%.2i*.tfrec'%x for x in idxV])\n        \n        NUM_TRAINING_IMAGES = int( count_data_items(files_train))\n        NUM_VALIDATION_IMAGES = int( count_data_items(files_valid) )\n        STEPS_PER_EPOCH = NUM_TRAINING_IMAGES // BATCH_SIZE\n        print('Dataset: {} training images, {} validation images,'.format(NUM_TRAINING_IMAGES, NUM_VALIDATION_IMAGES))\n        \n        train_dataset = load_dataset(list(pd.DataFrame({'TRAINING_FILENAMES': files_train}).loc[:]['TRAINING_FILENAMES']), labeled = True)\n        val_dataset = load_dataset(list(pd.DataFrame({'VALIDATION_FILENAMES': files_valid}).loc[:]['VALIDATION_FILENAMES']), labeled = True, ordered = True)\n        sv = tf.keras.callbacks.ModelCheckpoint(\n            'fold-%i.h5'%fold, monitor='val_loss', verbose=0, save_best_only=True,\n            save_weights_only=True, mode='min', save_freq='epoch')\n        model = get_model()\n        history = model.fit(\n            get_training_dataset(train_dataset), \n            steps_per_epoch = STEPS_PER_EPOCH,\n            epochs = EPOCHS,\n            callbacks = [get_lr_callback(BATCH_SIZE), sv],\n            validation_data = get_validation_dataset(val_dataset),\n            verbose=2\n        )\n        model.save_weights('fold-%if.h5'%fold)\n    \n    elif PHASE == 'inference':\n        model = get_model(weights=None)\n        model.load_weights('../input/leafdiseaseefnet0/fold-%i.h5'%fold)\n        test_ds = get_test_dataset(ordered=True) \n        test_ds = test_ds.map(to_float32)\n\n        print('Computing predictions...')\n        test_images_ds = testing_dataset\n        test_images_ds = test_ds.map(lambda image, idnum: image)\n        print(model.predict(test_images_ds))\n        probabilities += model.predict(test_images_ds) / FOLDS\n    \n        \n        \n    else:\n        print('PHASE must be train or inference.')\n    \n    \n    del model; z = gc.collect()\n    \npredictions = np.argmax(probabilities, axis=-1)\nprint(predictions)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"if PHASE == 'inference':\n    print('Generating submission.csv file...')\n    test_ids_ds = test_ds.map(lambda image, idnum: idnum).unbatch()\n    test_ids = next(iter(test_ids_ds.batch(NUM_TEST_IMAGES))).numpy().astype('U') # all in one batch\n    np.savetxt('submission.csv', np.rec.fromarrays([test_ids, predictions]), fmt=['%s', '%d'], delimiter=',', header='id,label', comments='')\n    !head submission.csv","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"","execution_count":null,"outputs":[]}],"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":4,"nbformat_minor":4}