resnet5_keras_baseline_COVID_Finetuning_fordemo.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.6.7"
    },
    "colab": {
      "name": "resnet5_keras_baseline_COVID_Finetuning_24April.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "machine_shape": "hm",
      "include_colab_link": true
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/VeereshShringari/COVID-testing/blob/master/resnet5_keras_baseline_COVID_Finetuning_fordemo.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "kApyA4pB-TqH",
        "colab_type": "text"
      },
      "source": [
        "### ResNet50 Keras baseline model\n",
        "\n",
        "This notebook takes you through some important steps in building a deep convnet in Keras for multilabel classification of brain CT scans. \n",
        "\n",
        "\n",
        "*Update (1):*\n",
        "* *training for 4 epochs instead of 3.*\n",
        "* *batch size lowered to 16 from 32.*\n",
        "* *training without learning rate decay.*\n",
        "* *Weighted BCE instead of \"plain\" BCE*\n",
        "* *training data lowered to 80% from 90%.*\n",
        "\n",
        "\n",
        "*Update (2):*\n",
        "* *adding competition metric for training*\n",
        "* *using custom Callback for validation and test sets instead of the `run()` function and 'global epochs'*\n",
        "* *training with \"plain\" BCE again*\n",
        "* *merging TestDataGenerator and TrainDataGenerator into one*\n",
        "* *adding undersampling (see inside `on_epoch_end`), will now run 6 epochs*\n",
        "\n",
        "*Update (3):*\n",
        "* *skipping/removing windowing (value clipping), but the transformation to Hounsfield Units is kept*\n",
        "* *removing initial layer (doing np.stack((img,)&ast;3, axis=-1)) instead*\n",
        "* *reducing learning rate to 5e-4 and add decay*\n",
        "* *increasing batch size to 32 from 16*\n",
        "* *Increasing training set to 90% of the data (10% for validation)*\n",
        "* *slight increase in undersampling*\n",
        "* *fixed some hardcoding for input dims/sizes*"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "60ahCsN4EKAE",
        "colab_type": "code",
        "outputId": "018f4c74-f8f1-4ca1-b88d-7185ae930d59",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "!unzip data.zip"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "unzip:  cannot find or open data.zip, data.zip.zip or data.zip.ZIP.\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "_cell_guid": "b1076dfc-b9ad-4769-8c92-a6c4dae69d19",
        "_uuid": "8f2839f25d086af736a60e9eeb907d3b93b6e0e5",
        "id": "D4RjvLfX-TqJ",
        "colab_type": "code",
        "outputId": "2a64f265-753c-4038-9087-60d55688750b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import collections\n",
        "from datetime import datetime\n",
        "from math import ceil, floor\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import os\n",
        "import pandas as pd\n",
        "from tqdm import tqdm_notebook as tqdm\n",
        "import sys\n",
        "\n",
        "\n",
        "\n",
        "import cv2\n",
        "import tensorflow as tf\n",
        "import keras\n",
        "\n",
        "\n",
        "from keras_applications.resnet import ResNet50\n",
        "from keras.preprocessing.image import load_img\n",
        "from keras.preprocessing.image import img_to_array\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from sklearn.model_selection import ShuffleSplit\n",
        "\n",
        "from tensorflow.keras.applications import imagenet_utils\n",
        "preprocess_input = imagenet_utils.preprocess_input\n",
        "\n",
        "WEIGHTS_PATH = 'drive/My Drive/resnet50_weights_tf_dim_ordering_tf_kernels.h5'\n",
        "WEIGHTS_PATH_NO_TOP = 'resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'\n",
        "\n",
        "\n",
        "TEST_IMAGES_DIR = 'drive/My Drive/data/'\n",
        "TRAIN_IMAGES_DIR = 'drive/My Drive/data/'\n",
        "\n",
        "\n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Vu4oeW23yNu3",
        "colab_type": "text"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yl7PogfSyN7a",
        "colab_type": "code",
        "outputId": "3f7478a0-9af9-4cb0-ac98-ee41f3592cfa",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive',force_remount=True)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jje61omNyOb1",
        "colab_type": "code",
        "outputId": "b9599b17-f411-403c-c8bd-7d500dfcad58",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "WEIGHTS_PATH"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'drive/My Drive/resnet50_weights_tf_dim_ordering_tf_kernels.h5'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XrqKG4g_FRkp",
        "colab_type": "code",
        "outputId": "5d8a1d8d-8aa5-4893-d7e5-c4682ed868be",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 376
        }
      },
      "source": [
        "import IPython.display as display\n",
        "from PIL import Image\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "import os\n",
        "import pathlib\n",
        "\n",
        "data_dir = pathlib.Path(\".\")\n",
        "images = list(data_dir.glob('drive/My Drive/data/*.jpg'))\n",
        "new_width = 3000\n",
        "new_height = 1000\n",
        "for image_path in images:\n",
        "    print(image_path)\n",
        "    im = cv2.imread(str(image_path))\n",
        "    #display.display(im)\n",
        "    width, height,_ = im.shape   # Get dimensions\n",
        "    if width < height:\n",
        "      M = cv2.getRotationMatrix2D((height/2,width/2),90,1)\n",
        "      im = cv2.warpAffine(im,M,(height,width))\n",
        "    plt.imshow(im.astype('uint8'))\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "drive/My Drive/data/image1.jpg\n",
            "drive/My Drive/data/image5.jpg\n",
            "drive/My Drive/data/image4.jpg\n",
            "drive/My Drive/data/image3.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "drive/My Drive/data/image8.jpg\n",
            "drive/My Drive/data/image9.jpg\n",
            "drive/My Drive/data/image6.jpg\n",
            "drive/My Drive/data/image0.jpg\n",
            "drive/My Drive/data/image7.jpg\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "fqw-N5SPKdIu",
        "colab_type": "code",
        "outputId": "5439c231-fc73-44c3-a7e3-4cb9726e9320",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "im.shape"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(2045, 4478, 3)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "GmgwadgA-TqQ",
        "colab_type": "text"
      },
      "source": [
        "### 1. Helper functions\n",
        "\n",
        "* normalizing the pixel values between -1 and 1 \n",
        "* read and transform dcms to 3-channel inputs for e.g. ResNet50.\n",
        "\n",
        "\\* Source for windowing (although now partly removed from this kernel): https://www.kaggle.com/omission/eda-view-dicom-images-with-correct-windowing"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "_cell_guid": "79c7e3d0-c299-4dcb-8224-4455121ee9b0",
        "_uuid": "d629ff2d2480ee46fbb7e2d37f6b5fab8052498a",
        "id": "f0URQlqi-TqR",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def _normalize(img):\n",
        "    if img.max() == img.min():\n",
        "        return np.zeros(img.shape)-1\n",
        "    return 2 * (img - img.min())/(img.max() - img.min()) - 1\n",
        "\n",
        "\n",
        "  \n",
        "def _read(path, desired_size, augment_data=0, plot=False):\n",
        "    \"\"\"Will be used in DataGenerator\n",
        "    Loads image, crops and resizes. With optional image data augmentation.\n",
        "    We assume that the image has been centered. \n",
        "    Input:\n",
        "    ----------------------------------\n",
        "    desired_size     :  desired size for the image (tuple)\n",
        "    augment_data     :  nb of data augmented samples (int)\n",
        "    \"\"\"\n",
        "    new_width, new_height,_ = desired_size\n",
        "    print (path)\n",
        "    img = cv2.imread(path)\n",
        "    rows, cols,_ = img.shape\n",
        "    if rows < cols:  \n",
        "      M = cv2.getRotationMatrix2D((cols/2,rows/2),90,1)\n",
        "      img = cv2.warpAffine(img,M,(cols,rows))\n",
        "    res = cv2.resize(img, dsize=desired_size[:2], interpolation=cv2.INTER_CUBIC)\n",
        "    samples = np.expand_dims(res, 0)\n",
        "    if augment_data>0:\n",
        "        # create image data augmentation generator\n",
        "        datagen = ImageDataGenerator(rotation_range=90,\n",
        "                                     width_shift_range=[-100,100])\n",
        "        # prepare iterator\n",
        "        it = datagen.flow(samples, batch_size=1)\n",
        "        for i in range(augment_data):\n",
        "          batch = it.next()    # generate batch of images\n",
        "          image = batch[0]\n",
        "          samples = np.vstack((samples, np.expand_dims(image, 0)))\n",
        "          if plot:\n",
        "            plt.subplot(330 + 1 + i)\n",
        "            plt.imshow(batch[0].astype('uint8'))\n",
        "          #img = np.stack((res,)*3, axis=-1)\n",
        "    if plot: plt.show()\n",
        "    print('samples size in read:', samples.shape)\n",
        "    return samples\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VO4vaCw7IfL3",
        "colab_type": "code",
        "outputId": "a2f9f01f-ccfd-4ab9-d5dd-c790a840c118",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 320
        }
      },
      "source": [
        "test = _read('drive/My Drive/data/image1.jpg',(512,512,3),9, plot=True)\n",
        "print(test.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "drive/My Drive/data/image1.jpg\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        },
        {
          "output_type": "stream",
          "text": [
            "samples size in read: (10, 512, 512, 3)\n",
            "(10, 512, 512, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5LXLCrMX7if2",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tGSfWNhna91-",
        "colab_type": "code",
        "outputId": "0ea17c11-c630-4efd-ff08-4fb06345fb71",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 269
        }
      },
      "source": [
        "\n",
        "\n",
        "from numpy import expand_dims\n",
        "from keras.preprocessing.image import load_img\n",
        "from keras.preprocessing.image import img_to_array\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "import matplotlib.pyplot as plt\n",
        "import scipy as sc\n",
        "\n",
        "img = cv2.imread('drive/My Drive/data/image0.jpg')\n",
        "# convert to numpy array\n",
        "res = cv2.resize(img, dsize=(512, 512), interpolation=cv2.INTER_CUBIC)\n",
        "data = res\n",
        "# expand dimension to one sample\n",
        "\n",
        "samples = expand_dims(data, 0)\n",
        "# create image data augmentation generator\n",
        "datagen = ImageDataGenerator(rotation_range = 90, width_shift_range=[-100,100])\n",
        "# prepare iterator\n",
        "it = datagen.flow(samples, batch_size=1)\n",
        "for i in range(9):\n",
        "\t# define subplot\n",
        "\tplt.subplot(330 + 1 + i)\n",
        "\t# generate batch of images\n",
        "\tbatch = it.next()\n",
        "\t# convert to unsigned integers for viewing\n",
        "\timage = batch[0].astype('uint8')\n",
        "\t# plot raw pixel data\n",
        "\tplt.imshow(image)\n",
        "# show the figure\n",
        "plt.show()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 9 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "RVD3MWoh-TqW",
        "colab_type": "text"
      },
      "source": [
        "### 2. Data generators\n",
        "\n",
        "Inherits from keras.utils.Sequence object and thus should be safe for multiprocessing.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GVupHBDB-TqX",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "\n",
        "class DataGenerator(keras.utils.Sequence):\n",
        "    def __init__(self, list_IDs, img_labels, batch_size=1, img_size=(512, 512,3), \n",
        "                 img_dir=TRAIN_IMAGES_DIR, n_classes = 4, train =True,\n",
        "                 n_augment = 9, shuffle=True,\n",
        "                 *args, **kwargs):\n",
        "\n",
        "        self.list_IDs = list_IDs\n",
        "        self.indices = np.arange(len(self.list_IDs))\n",
        "        self.img_labels = img_labels ### contains col1: names of images for loading + col2(!exits fr test) for labels\n",
        "        self.n_classes = n_classes   ### nb of classes\n",
        "        self.n_augment = n_augment   ### nb of additional data samples\n",
        "        self.batch_size = batch_size\n",
        "        self.img_size = img_size    ###  desired image size: (width, height, n_channels)\n",
        "        self.img_dir = img_dir\n",
        "        self.shuffle = shuffle\n",
        "        self.train = train\n",
        "        self.on_epoch_end()\n",
        "\n",
        "    def __len__(self):\n",
        "        return int(ceil(len(self.indices) / self.batch_size))\n",
        "\n",
        "    def __getitem__(self, index):\n",
        "        indices = self.indices[index*self.batch_size: (index+1)*self.batch_size]\n",
        "        list_IDs_temp = [self.list_IDs[k] for k in indices]\n",
        "        if self.train:\n",
        "            X, Y = self.__data_generation(list_IDs_temp)\n",
        "            return X, Y\n",
        "        else:\n",
        "            X = self.__data_generation(list_IDs_temp)\n",
        "            return X\n",
        "        \n",
        "    def on_epoch_end(self):\n",
        "        self.indices = np.arange(len(self.list_IDs))\n",
        "        if self.shuffle == True:\n",
        "            np.random.shuffle(self.indices)\n",
        "\n",
        "    def __data_generation(self, list_IDs_temp):\n",
        "        print(\"Self image size\",self.img_size )\n",
        "        X = np.empty((self.batch_size * (self.n_augment + 1), *self.img_size))\n",
        "        \n",
        "        if self.train: # training phase\n",
        "            Y = np.zeros((self.batch_size * (self.n_augment + 1), self.n_classes),\n",
        "                          dtype=np.float32)\n",
        "        \n",
        "            for i, ID in enumerate(list_IDs_temp):\n",
        "              test = _read(self.img_dir+self.img_labels['ID'].loc[ID] +\".jpg\",\n",
        "                              self.img_size,  augment_data= self.n_augment, plot=False)\n",
        "              print(\"Dim data gen\",  test.shape)\n",
        "              X[i:(i +self.n_augment + 1),] = test\n",
        "              ### Convert label  into one hot vector\n",
        "              Y[i:(i +self.n_augment + 1), int(self.img_labels['label'].loc[ID])] = 1\n",
        "\n",
        "            return X, Y\n",
        "        \n",
        "        else: # test phase\n",
        "            for i, ID in enumerate(list_IDs_temp):\n",
        "                X[i,] = _read(self.img_dir+self.img_labels['ID'].loc[ID] +\".jpg\",\n",
        "                              self.img_size, augment_data=0, plot=False)      \n",
        "            return X"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3dk9Plt--Tqe",
        "colab_type": "text"
      },
      "source": [
        "### 3a. loss function and metric"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Aj4BKXWH-Tqf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras import backend as K\n",
        "\n",
        "def weighted_log_loss(y_true, y_pred):\n",
        "    \"\"\"\n",
        "    Can be used as the loss function in model.compile()\n",
        "    ---------------------------------------------------\n",
        "    \"\"\"\n",
        "    \n",
        "    class_weights = np.array([1., 2., 2., 2.])\n",
        "    \n",
        "    eps = K.epsilon()\n",
        "    \n",
        "    y_pred = K.clip(y_pred, eps, 1.0-eps)\n",
        "\n",
        "    out = -(         y_true  * K.log(      y_pred) * class_weights\n",
        "            + (1.0 - y_true) * K.log(1.0 - y_pred) * class_weights)\n",
        "    \n",
        "    return K.mean(out, axis=-1)\n",
        "\n",
        "\n",
        "def _normalized_weighted_average(arr, weights=None):\n",
        "    \"\"\"\n",
        "    A simple Keras implementation that mimics that of \n",
        "    numpy.average(), specifically for the this competition\n",
        "    \"\"\"\n",
        "    \n",
        "    if weights is not None:\n",
        "        scl = K.sum(weights)\n",
        "        weights = K.expand_dims(weights, axis=1)\n",
        "        return K.sum(K.dot(arr, weights), axis=1) / scl\n",
        "    return K.mean(arr, axis=1)\n",
        "\n",
        "\n",
        "def weighted_loss(y_true, y_pred):\n",
        "    \"\"\"\n",
        "    Will be used as the metric in model.compile()\n",
        "    ---------------------------------------------\n",
        "    \n",
        "    Similar to the custom loss function 'weighted_log_loss()' above\n",
        "    but with normalized weights, which should be very similar \n",
        "    to the official competition metric:\n",
        "        https://www.kaggle.com/kambarakun/lb-probe-weights-n-of-positives-scoring\n",
        "    and hence:\n",
        "        sklearn.metrics.log_loss with sample weights\n",
        "    \"\"\"\n",
        "    \n",
        "    class_weights = K.variable([1., 2., 2., 2.])\n",
        "    \n",
        "    eps = K.epsilon()\n",
        "    \n",
        "    y_pred = K.clip(y_pred, eps, 1.0-eps)\n",
        "\n",
        "    loss = -(        y_true  * K.log(      y_pred)\n",
        "            + (1.0 - y_true) * K.log(1.0 - y_pred))\n",
        "    \n",
        "    loss_samples = _normalized_weighted_average(loss, class_weights)\n",
        "    \n",
        "    return K.mean(loss_samples)\n",
        "\n",
        "\n",
        "def weighted_log_loss_metric(trues, preds):\n",
        "    \"\"\"\n",
        "    Will be used to calculate the log loss \n",
        "    of the validation set in PredictionCheckpoint()\n",
        "    ------------------------------------------\n",
        "    \"\"\"\n",
        "    class_weights = [1., 2., 2., 2.]\n",
        "    \n",
        "    epsilon = 1e-7\n",
        "    \n",
        "    preds = np.clip(preds, epsilon, 1-epsilon)\n",
        "    loss = trues * np.log(preds) + (1 - trues) * np.log(1 - preds)\n",
        "    loss_samples = np.average(loss, axis=1, weights=class_weights)\n",
        "\n",
        "    return - loss_samples.mean()\n",
        "\n",
        "def sparse_categorical_crossentropy(y_true, y_pred, from_logits=False, axis=-1):\n",
        "    return K.sparse_categorical_crossentropy(\n",
        "        y_true, y_pred, from_logits=from_logits, axis=axis)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_WOiCkW0Hyhz",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "#model = MyDeepModel(engine=ResNet50, input_dims=(224,224, 3),\n",
        "  #               batch_size=3, learning_rate=5e-5, n_classes=4\n",
        "    #             num_epochs=1, decay_rate=0.8, decay_steps=1,\n",
        "    #             weights='imagenet', loss =weighted_log_loss, verbose=2)\n",
        "\n",
        "#obtain test + validation predictions (history.test_predictions, history.valid_predictions)\n",
        "#history = model.fit_and_predict(df.iloc[train_idx], df.iloc[valid_idx], test_df)\n",
        "\n",
        "# Define the Keras TensorBoard callback.\n",
        "%load_ext tensorboard\n",
        "logdir=\"logs/fit/\" + datetime.now().strftime(\"%Y%m%d-%H%M%S\")\n",
        "tensorboard_callback = keras.callbacks.TensorBoard(log_dir=logdir)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6Iyq1X8R-Tqm",
        "colab_type": "text"
      },
      "source": [
        "### 3b. Model\n",
        "\n",
        "Model is divided into three parts: <br> \n",
        "\n",
        "* (REMOVED) The initial layer, which will transform/map input image of shape (\\_, \\_, 1) to another \"image\" of shape (\\_, \\_, 3).\n",
        "\n",
        "* The new input image is then passed through ResNet50 (which I named \"engine\"). ResNet50 could be replaced by any of the available architectures in keras_application.\n",
        "\n",
        "* Finally, the output from ResNet50 goes through average pooling followed by a dense output layer."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-KCkHGPm-Tqn",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 233
        },
        "outputId": "07437223-4cce-4e85-8ee7-6d2e59819ee4"
      },
      "source": [
        "class PredictionCheckpoint(keras.callbacks.Callback):\n",
        "    \n",
        "    def __init__(self, test_df, valid_df, n_classes =4,\n",
        "                 test_images_dir=TEST_IMAGES_DIR, \n",
        "                 valid_images_dir=TRAIN_IMAGES_DIR, \n",
        "                 batch_size=32, input_size=(224, 224, 3)):\n",
        "        \n",
        "        self.test_df = test_df\n",
        "        self.valid_df = valid_df\n",
        "        self.test_images_dir = test_images_dir\n",
        "        self.valid_images_dir = valid_images_dir\n",
        "        self.batch_size = batch_size\n",
        "        self.input_size = input_size\n",
        "        self.n_classes = n_classes\n",
        "        \n",
        "        \n",
        "    def on_train_begin(self, logs={}):\n",
        "        self.test_predictions = []\n",
        "        self.valid_predictions = []\n",
        "        \n",
        "    def on_epoch_end(self,batch, logs={}):\n",
        "        self.test_predictions.append(\n",
        "            self.model.predict_generator(\n",
        "                DataGenerator(self.test_df.index, test_df,\n",
        "                              batch_size=self.batch_size, img_size=self.input_size, \n",
        "                              img_dir=self.test_images_dir, n_classes = self.n_classes,\n",
        "                              train =False, n_augment = 0, shuffle=True),\n",
        "                verbose=2)[:len(self.test_df)])\n",
        "        \n",
        "        self.valid_predictions.append(\n",
        "            self.model.predict_generator(\n",
        "                DataGenerator(self.valid_df.index, valid_df,\n",
        "                              batch_size=self.batch_size, img_size=self.input_size, \n",
        "                              img_dir=self.valid_images_dir, n_classes = self.n_classes,\n",
        "                              train =False, n_augment = 0, shuffle=True),\n",
        "                verbose=2)[:len(self.valid_df)])\n",
        "        valid_labels = np.zeros((self.valid_df.shape[0], self.n_classes))\n",
        "        valid_labels[np.arange(self.valid_df.shape[0]), self.valid_df['label']] = 1\n",
        "        print('valid_labels', valid_labels )\n",
        "        print('pred_labels', self.valid_predictions)\n",
        "        print(\"validation loss: %.4f\" %\n",
        "              weighted_log_loss_metric(valid_labels, \n",
        "                                   np.average(self.valid_predictions, axis=0)))\n",
        "        \n",
        "        # here you could save the predictions with np.save()\n",
        "\n",
        "\n",
        "class MyDeepModel:\n",
        "    \n",
        "    def __init__(self, engine, input_dims, batch_size=5, num_epochs=4,\n",
        "                 n_classes=4, learning_rate=1e-3, n_augment = 9,\n",
        "                 decay_rate=1.0, decay_steps=1, weights=WEIGHTS_PATH_NO_TOP, verbose=1):\n",
        "        \n",
        "        self.engine = engine\n",
        "        self.input_dims = input_dims\n",
        "        self.batch_size = batch_size\n",
        "        self.num_epochs = num_epochs\n",
        "        self.n_classes = n_classes\n",
        "        self.learning_rate = learning_rate\n",
        "        self.decay_rate = decay_rate\n",
        "        self.decay_steps = decay_steps\n",
        "        self.n_augment = n_augment\n",
        "        self.weights = weights\n",
        "        self.verbose = verbose\n",
        "        self._build()\n",
        "\n",
        "    def _build(self):\n",
        "        self.engine.trainable = True\n",
        "        \n",
        "      \n",
        "    \n",
        "        engine = self.engine(include_top=False,\n",
        "                             weights=self.weights, input_shape=(*self.input_dims[:2], 3),\n",
        "                             backend = keras.backend, layers = keras.layers, \n",
        "                             models = keras.models, utils = keras.utils,)\n",
        "        set_trainable = False\n",
        "        for layer in engine.layers:\n",
        "            if layer.name in ['res5c_branch2b', 'res5c_branch2c', 'activation_97']:\n",
        "              set_trainable = True\n",
        "            if set_trainable:\n",
        "              layer.trainable = True\n",
        "            else:\n",
        "              layer.trainable = False\n",
        "        x = keras.layers.GlobalAveragePooling2D(name='avg_pool')(engine.output)\n",
        "        out = keras.layers.Dense(self.n_classes, activation=\"sigmoid\", name='dense_output')(x)\n",
        "\n",
        "        self.model = keras.models.Model(inputs=engine.input, outputs=out)\n",
        "        # loss function has been changed needs to be investigated.\n",
        "        self.model.compile(loss='categorical_crossentropy',\n",
        "                           optimizer=keras.optimizers.Adam(0.0),\n",
        "                           metrics=['accuracy'])\n",
        "    \n",
        "\n",
        "    def fit_and_predict(self, train_df, valid_df, test_df):\n",
        "        \n",
        "        # callbacks\n",
        "        pred_history = PredictionCheckpoint(test_df, valid_df,\n",
        "                                            n_classes=self.n_classes,\n",
        "                                            batch_size=self.batch_size,\n",
        "                                            input_size=self.input_dims)\n",
        "        checkpointer = keras.callbacks.ModelCheckpoint(filepath='%s-{epoch:02d}.hdf5' % self.engine.__name__, verbose=1, save_weights_only=True, save_best_only=False)\n",
        "        scheduler = keras.callbacks.LearningRateScheduler(lambda epoch: self.learning_rate * pow(self.decay_rate, floor(epoch / self.decay_steps)))\n",
        "        self.model.fit_generator(\n",
        "            DataGenerator(\n",
        "                list_IDs = train_df.index, \n",
        "                img_labels = train_df,\n",
        "                batch_size=self.batch_size,\n",
        "                img_size=self.input_dims,\n",
        "                img_dir=TRAIN_IMAGES_DIR,\n",
        "                n_classes = self.n_classes,\n",
        "                train=True,\n",
        "                n_augment = self.n_augment,\n",
        "                shuffle = True\n",
        "            ),\n",
        "            epochs=self.num_epochs,\n",
        "            verbose=self.verbose,\n",
        "            use_multiprocessing=True,\n",
        "            workers=4#,\n",
        "            #callbacks=[history]\n",
        "            #callbacks=[tensorboard_callback]\n",
        "        )\n",
        "        \n",
        "        return pred_history\n",
        "    def predict(self, image_name, path2image=TRAIN_IMAGES_DIR):\n",
        "        #### Predict one image at a time\n",
        "        X = _read( path2image + image_name, self.input_dims,0, plot=False)\n",
        "        res = self.model.predict(X, batch_size=1)\n",
        "        return res\n",
        "    \n",
        "    def save(self, path):\n",
        "        self.model.save_weights(path)\n",
        "    \n",
        "    def load(self, path):\n",
        "        self.model.load_weights(path)"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-1-ad11dec95ed9>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0;32mclass\u001b[0m \u001b[0mPredictionCheckpoint\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mkeras\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcallbacks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mCallback\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m     def __init__(self, test_df, valid_df, n_classes =4,\n\u001b[1;32m      4\u001b[0m                  \u001b[0mtest_images_dir\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mTEST_IMAGES_DIR\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m                  \u001b[0mvalid_images_dir\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mTRAIN_IMAGES_DIR\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'keras' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D-siq1_GxoP1",
        "colab_type": "code",
        "outputId": "4e45c8c9-50ff-4bd3-a629-9334137e5b23",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "WEIGHTS_PATH_NO_TOP"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'resnet50_weights_tf_dim_ordering_tf_kernels_notop.h5'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Kx7Fww_f-Tqt",
        "colab_type": "text"
      },
      "source": [
        "### 4. Read csv files\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SWt5X9Fa-Tqu",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def read_testset(filename=\"drive/My Drive/data/classification_labels.txt\"):\n",
        "    ''' \n",
        "    Data in data folder\n",
        "    '''\n",
        "    df = pd.read_csv(filename, sep=\" \", header=None)\n",
        "    df.columns = [\"ID\", \"label\"]\n",
        "    return df\n",
        "\n",
        "def read_trainset(filename=\"drive/My Drive/data/classification_labels.txt\"):\n",
        "    df = pd.read_csv(filename, sep=\" \", header=None)\n",
        "    df.columns = [\"ID\", \"label\"]\n",
        "    return df\n",
        "\n",
        "    \n",
        "test_df = read_testset()\n",
        "df = read_trainset()"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_fVDXUo3-Tq6",
        "colab_type": "code",
        "outputId": "b4f4644c-07d1-45d4-b48a-14c86b2fcf05",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 138
        }
      },
      "source": [
        "df.head(3)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>ID</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>image0</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>image1</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>image2</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       ID  label\n",
              "0  image0      3\n",
              "1  image1      3\n",
              "2  image2      3"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 16
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wL4LR8Gw-TrA",
        "colab_type": "code",
        "outputId": "a907dc32-d9f9-47a5-ab11-a3f3fdb661eb",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 138
        }
      },
      "source": [
        "test_df.head(3)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>ID</th>\n",
              "      <th>label</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>image0</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>image1</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>image2</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       ID  label\n",
              "0  image0      3\n",
              "1  image1      3\n",
              "2  image2      3"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1WEYOH5x-TrF",
        "colab_type": "text"
      },
      "source": [
        "### 5. Train model and predict\n",
        "\n",
        "*Using train, validation and test set* <br>\n",
        "\n",
        "Training for 6 epochs with Adam optimizer, with a learning rate of 0.0005 and decay rate of 0.8. The validation predictions are \\[exponentially weighted\\] averaged over all 6 epochs (same goes for the test set submission later). `fit_and_predict` returns validation and test predictions for all epochs.\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zQpkQBx8-TrG",
        "colab_type": "code",
        "outputId": "5084b3bd-0692-44da-a6d8-e7ea5c7404bc",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "# train set (80%) and validation set (20%)\n",
        "ss = ShuffleSplit(n_splits=10, test_size=0.1, random_state=42).split(df.index)\n",
        "# lets go for the first fold only\n",
        "train_idx, valid_idx = next(ss)\n",
        "# obtain model\n",
        "model2 = MyDeepModel(engine=ResNet50, input_dims=(224, 224, 3), batch_size=3,\n",
        "                     learning_rate=5e-4, n_augment = 0,\n",
        "                    num_epochs=12, decay_rate=0.8, decay_steps=1, weights=\"imagenet\", verbose=2)\n",
        "# obtain test + validation predictions (history.test_predictions, history.valid_predictions)\n",
        "history = model2.fit_and_predict(df.iloc[train_idx], df.iloc[valid_idx], test_df)\n"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/6\n",
            "Epoch 1/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "drive/My Drive/data/image7.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image0.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image2.jpg\n",
            "drive/My Drive/data/image6.jpg\n",
            "drive/My Drive/data/image0.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 11s - loss: nan - accuracy: 0.7333\n",
            "Epoch 2/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "drive/My Drive/data/image9.jpg\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image0.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 8s - loss: nan - accuracy: 0.8333\n",
            "Epoch 3/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image0.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 9s - loss: nan - accuracy: 0.8444\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "Epoch 4/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image0.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 9s - loss: nan - accuracy: 0.8444\n",
            "Epoch 5/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "drive/My Drive/data/image3.jpg\n",
            "drive/My Drive/data/image2.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image0.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 9s - loss: nan - accuracy: 0.8333\n",
            "Epoch 6/6\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "Self image size (224, 224, 3)\n",
            "drive/My Drive/data/image9.jpg\n",
            "drive/My Drive/data/image0.jpg\n",
            "drive/My Drive/data/image4.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image5.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image1.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image7.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image2.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image3.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            "drive/My Drive/data/image6.jpg\n",
            "samples size in read: (10, 224, 224, 3)\n",
            "Dim data gen (10, 224, 224, 3)\n",
            " - 9s - loss: nan - accuracy: 0.7333\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4YfFbPCQWwZP",
        "colab_type": "code",
        "outputId": "e2bd010f-21d0-4a38-d7b3-d39cb03c8357",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 85
        }
      },
      "source": [
        "!pip install pyyaml h5py "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Requirement already satisfied: pyyaml in /usr/local/lib/python3.6/dist-packages (3.13)\n",
            "Requirement already satisfied: h5py in /usr/local/lib/python3.6/dist-packages (2.10.0)\n",
            "Requirement already satisfied: six in /usr/local/lib/python3.6/dist-packages (from h5py) (1.12.0)\n",
            "Requirement already satisfied: numpy>=1.7 in /usr/local/lib/python3.6/dist-packages (from h5py) (1.18.3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TN62nd8AWY9X",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.save('data\\final_weight.h5')\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uCRk9n8S0E7V",
        "colab_type": "text"
      },
      "source": [
        "Using the Tensor graph analysis"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "XUscXfsW-TrO",
        "colab_type": "text"
      },
      "source": [
        "### 6. Submit test predictions"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ghTcolzR-TrQ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "test_df.iloc[:, :] = np.average(history.test_predictions, axis=0, weights=[2**i for i in range(len(history.test_predictions))])\n",
        "test_df = test_df.stack().reset_index()\n",
        "\n",
        "test_df.insert(loc=0, column='ID', value=test_df['Image'].astype(str) + \"_\" + test_df['Diagnosis'])\n",
        "\n",
        "test_df = test_df.drop([\"Image\", \"Diagnosis\"], axis=1)\n",
        "\n",
        "test_df.to_csv('submission.csv', index=False)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UX2ywVS5YkWN",
        "colab_type": "text"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "pur_H8QUiBsp",
        "colab_type": "text"
      },
      "source": [
        "### 7. Integrating questionnaire data\n",
        "\n",
        "Add:\n",
        " + Questionnaire information\n",
        " + County information"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "1huwRCSMlyMn",
        "colab": {}
      },
      "source": [
        "df['When did the individual first have COVID-19 symptoms?'].apply(lambda x: datetime.strptime(x, \"%m/%d/%y\"))"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "1YktBNQzrITO",
        "colab_type": "text"
      },
      "source": [
        ""
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U1wPszZonG1b",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df['Has the individual had close contact with anyone who has:']"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ambEo5XpmlPJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "df.columns"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XeRjNxMUiO7T",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def preprocessing_questionnaire(df):\n",
        "  #### parse the columns:\n",
        "  df['day'] = 0\n",
        "  df['month'] = 0\n",
        "  df['year'] = 0\n",
        "  ### age:\n",
        "  df['age'] = df['How old is the individual?'].apply(lambda x: 0 * (x == '0-18') + 1 * (x == '19-35') \\\n",
        " + 2 * (x == '36-50') + 3 * (x == '51-65') + 4 * (x == '66-85') + 5 * (x == '85+')  )\n",
        "  df['PCR']  = df['Has the individual had a positive RT-PCR result?'].apply(lambda x: 0 * (x == 'No, they have not taken a test.') \\\n",
        "                                                                            + 1 * ( x == 'Yes, they took a PCR test and got a negative r...')\n",
        "                                                                            + 2 * (x == 'Yes, they took a PCR test and got a positive r...') )\n",
        "  #### Need to check where are all the entries\n",
        "  df['Chills'] = 0\n",
        "  df['Cough'] = 0\n",
        "  df['Fatigue'] = 0\n",
        "  df['Fever'] = 0\n",
        "  df['Loss of smell and taste'] = 0\n",
        "  df['Muscle soreness'] = 0\n",
        "  df['Sore throat'] = 0\n",
        "  for i in range(df.shape[0]):\n",
        "    month, day, year = df['Timestamp'].iloc[0].split(' ')[0].split('/')\n",
        "    df['day'].iloc[i] = int(day)\n",
        "    df['month'].iloc[i] = int(month)\n",
        "    df['year'].iloc[i] = int(year)\n",
        "    symptoms = df['Did the individual have any COVID-19 symptoms? (Tick all that apply)'].iloc[i].split(', ')\n",
        "    for symp in symptoms:\n",
        "      df[symp].iloc[i] = 1\n",
        "  df['sympt_onset'] =  0  ### TO DO: current problem with the dataset (M/D D/M)\n",
        "  return df\n",
        "    \n",
        "\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "tduR_eIY-Trc",
        "colab_type": "text"
      },
      "source": [
        "### 8. Improvements\n",
        "\n",
        "\n",
        "Some improvements that could possibly be made:<br>\n",
        "* Image augmentation (which can be put in `_read()`)\n",
        "* Different learning rate and learning rate schedule\n",
        "* Increased input size\n",
        "* Train longer\n",
        "* Add dense layer and regularization (e.g. `keras.layers.Dropout()` before the output layer)\n",
        "* Adding some optimal windowing\n",
        "<br>\n",
        "<br>\n",
        "*Feel free to comment!*\n"
      ]
    }
  ]
}