Anti_Spoofing_Latest (1).ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Anti Spoofing Latest.ipynb",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "1IQJ-sHAVfDQ",
        "colab_type": "code",
        "outputId": "ec286898-98c4-4b63-cade-1042d32e6774",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "import glob\n",
        "import numpy as np\n",
        "import os\n",
        "import shutil\n",
        "np.random.seed(42)\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array, array_to_img\n",
        "%matplotlib inline"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ofe5AlvvWFZm",
        "colab_type": "code",
        "outputId": "132af611-ab3c-47a0-b0ba-2e2e712e5608",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        }
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9y9ds4LHV-AR",
        "colab_type": "code",
        "outputId": "77ed99fb-0631-436c-bd77-76f4956c8137",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "files = glob.glob('/content/drive/My Drive/data/*')\n",
        "\n",
        "true_files = [fn for fn in files if 'true' in fn]\n",
        "false_files = [fn for fn in files if 'false' in fn]\n",
        "len(true_files), len(false_files)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(934, 953)"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 2
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2LTuxUKPXwWj",
        "colab_type": "code",
        "outputId": "e305c5e6-1364-4e93-c7f8-8c33f5d48fea",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "true_train = np.random.choice(true_files, size=500, replace=False)\n",
        "false_train = np.random.choice(false_files, size=500, replace=False)\n",
        "true_files = list(set(true_files) - set(true_train))\n",
        "false_files = list(set(false_files) - set(false_train))\n",
        "\n",
        "true_val = np.random.choice(true_files, size=200, replace=False)\n",
        "false_val = np.random.choice(false_files, size=200, replace=False)\n",
        "true_files = list(set(true_files) - set(true_val))\n",
        "false_files = list(set(false_files) - set(false_val))\n",
        "\n",
        "true_test = np.random.choice(true_files, size=200, replace=False)\n",
        "false_test = np.random.choice(false_files, size=200, replace=False)\n",
        "\n",
        "print('True datasets:', true_train.shape, true_val.shape, true_test.shape)\n",
        "print('False datasets:', false_train.shape, false_val.shape, false_test.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "True datasets: (500,) (200,) (200,)\n",
            "False datasets: (500,) (200,) (200,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "gv_iLDQhYZCK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_dir = 'training_data'\n",
        "val_dir = 'validation_data'\n",
        "test_dir = 'test_data'\n",
        "\n",
        "train_files = np.concatenate([true_train, false_train])\n",
        "validate_files = np.concatenate([true_val, false_val])\n",
        "test_files = np.concatenate([true_test, false_test])\n",
        "\n",
        "os.mkdir(train_dir) if not os.path.isdir(train_dir) else None\n",
        "os.mkdir(val_dir) if not os.path.isdir(val_dir) else None\n",
        "os.mkdir(test_dir) if not os.path.isdir(test_dir) else None\n",
        "\n",
        "for fn in train_files:\n",
        "    shutil.copy(fn, train_dir)\n",
        "\n",
        "for fn in validate_files:\n",
        "    shutil.copy(fn, val_dir)\n",
        "    \n",
        "for fn in test_files:\n",
        "    shutil.copy(fn, test_dir)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "hDMD47HNYY-z",
        "colab_type": "code",
        "outputId": "773ab958-3192-4181-b4b5-09ef871a4006",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "IMG_DIM = (100, 100)\n",
        "\n",
        "train_files = glob.glob('training_data/*')\n",
        "train_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in train_files]\n",
        "train_imgs = np.array(train_imgs)\n",
        "train_labels = [fn.split('/')[1].split(' ')[0].strip() for fn in train_files]\n",
        "\n",
        "validation_files = glob.glob('validation_data/*')\n",
        "validation_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in validation_files]\n",
        "validation_imgs = np.array(validation_imgs)\n",
        "validation_labels = [fn.split('/')[1].split(' ')[0].strip() for fn in validation_files]\n",
        "\n",
        "print('Train dataset shape:', train_imgs.shape, \n",
        "      '\\tValidation dataset shape:', validation_imgs.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train dataset shape: (1400, 100, 100, 3) \tValidation dataset shape: (512, 100, 100, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tiJA8sNmYY8F",
        "colab_type": "code",
        "outputId": "6ecb5fc6-9bfd-4d6b-cd1c-4b78bc191d45",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 134
        }
      },
      "source": [
        "train_imgs_scaled = train_imgs.astype('float32')\n",
        "validation_imgs_scaled  = validation_imgs.astype('float32')\n",
        "train_imgs_scaled /= 255\n",
        "validation_imgs_scaled /= 255\n",
        "\n",
        "print(train_imgs[0].shape)\n",
        "array_to_img(train_imgs[0])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(100, 100, 3)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGQAAABkCAIAAAD/gAIDAAA3uklEQVR4nM29aYxtV3Ye9u2119nn\n3HOnunVfsViv+Pj4OIum6FaLoqTutlpxO+60W7IjJ4FhxEDk/LCDOIAQRTHyJwMQGEGCTECCIEBs\nJD8iJICNBIbhIVFsQVJb3ZJpmqZJNrs5vInFepf1arjTGfZZe+38OPdW3ff4HsmWNXiBIIv3nnuG\ntdde61vjMTdvvvWln/6ZslRVBaCqoqH9Q1UzgqoS0XJRBCnwGWRhDABEWX9iANo4IDz4V/d/ZQAA\ntP7QwKbWMohCUyGcHxl624+89NJLo+HocDJ54403pPaXHt1lZgCz+byYzhDqh1z0go5uvDaks3/r\nX/mjX3750Q/P8Ovv01RTIgJARC1bWmLVGPWhJ1JVVYhIkPrTL/k56Pwy9t4HePjlPwd5L8uyOD0+\nlroy1u3t7g23hgBOT09Pz04X88WyKEO5QAwP41q/15ez424XzFakVmXVVkTuP5JV1TdelbBiTbz/\nEVXruvjM9VmJQ4wr4Tr/5IIX52fWjQPoQcyK9/3QGLLOBZG1uF38pC4LZqp9BRVKsu6g51J3enxa\nVDUAlzqXOt0aAvBVvZieIFT3XSwdjxEmW4O8qurEubyTTpftg69Ydi5lrCqfZNA5qaLxPobm0/l0\n75Oeny2u+fXQQx8uVmaDuZaYmSkIoO0GX+3QLOtw4lRVRAGEav76669nLhUJotI+ZLuPiEhE1j+/\nj3rlvEwSt1wuyir3jQHcA++JAUB1rTgeQBpl475/1+lhDKULWbbWnG+JVnLX6+EShmpTeV/XiBE2\nqc4+XkuOBXcss6rGIMZyDLKhTDcpqXzlOlwsK6nZe34os44mZ54SktWdERlVqBKRVQ21X8TgP8cz\ntxvKwBDi+YZtF0A3eN3Kyybrzxfpk9t8LVwhAMhc6r3AGMSwNhoGQNVUqlFakQkCMBBXW1V8EA8E\nIMRP0yJ3eolHsVjOUPWGDYM3jBJtqC4GwJaxZhYUDlYAKNRL8MtPZ9IPQueW8b4bf+BO3GCxte1d\nO8dRstBUQCuR1hKrxqqoqqpd0fPzG9iONRyiIHymARlprMFaVECH006u9QMFEKwanXOeGIAEUSXv\nBURREarlvVJgAAubm9TF4nTjme36gQnEG6ZdjU1jaNZHWgAwhLgpaw9b8YvrGuvyPCfLW6OR9CVl\n14ifzedZ1mFmVZRVFVVhEwRt5QgAQhlMiqj3YpcHUJjffPuNN7TyzQLSKY+OJc+zzQNa4RIRXi4X\nX/jCF370a988/y7v9CUqgi6LmghefVVUdVUbMtHE9k5m85PT01NdVlXTeO/n85lIEO816uTmbSkq\n65xhCl5QlggCKCBrJoR79JTNEARoeWoBA+NgHeSsZVU66HE3JWIwMzMpOnlvfGmsqlXV+KqqyjI0\n3hiC5QhdY6sHSvEDyObZ7OzEWRiAmZ27R2FdwE9RbsFnrUBkTiwRKedSVxoVhBMVMl0eDrtbyPKc\niKxlJkOpq+sagAFCq+HYEBEZRlnX3qed1LFrL0NERMR+3q5SVVftQlVVVdf1OEVVV9Oz6Xy+qKqy\nqvxyOS+K4oN33qiqqpPnl8bj2bL9rXHWofFMnHZyVWWqFoslAGvZOgcFM5dLimHxWYZ4k1s0n03Z\n2ijBkHWZU7nYhi3EJSJAWMHKuYaKwKqOjF0uj5GkVVUWdUWpYzBHUMKQyjrnLCGSNGVrhiOQMHNi\nCEREgPrMWpeC2UfAKCwABZRcCkBD8OzaO+DegIEKaLznoR9Cx8Tn9v6rllU1y7IsSQGIioQAgElY\n4RJOslREouhsOffeq6ozLCLz+Xy+nFtj66pelAtfeeenPizLohDvT0+nZ9NquSiLcpZbnNy+jmZR\n3DzY2xlOjk6OhKrZMbhLRMztKlMrT0TEIrIC9SRQViIRUUHV+NVxpKqGROEuNn8QOT8FRJhIrbaK\nS1VVokBajnwmEWDamwIxt/JJQWRTWzKzetUQYHTuvaqiQkdBBhqBLCeXQdWDG+/Jue5gRBY9ojGA\nSF1bcaoApInLabUotC68F3+F6//9f/lP0ZS3bnww7kmnb0hIAFptPeA+a+i9J8LaAxKvCqBoqhUT\nDZQUsGAioigqJMwsIudOExGJKhREEbjAfe0ZLphioCHoJxwuUQ0i7XlaTgEwRKH0SebYWmZeAUsL\nDWSUSGCYMiRZlmlQ770YUYqr0wYAoACEVgOgYYCjKhoYzZnAFJl8hnD64z/+U0jT2fxkZ4vYoZiX\nedavvLTMajnTEjOzSGjh7wVFVdUoosywG08LAhD8SqbWgoTWezzHjUFBRCoRAG3oSoXq2j+nT/pd\nn5tcQmIUUBhR9aKiCArVqHVkH0mIRPVC10TqNMzWKTQGkYY5oIlg+Nl0ce3qk5jNl2WlwTFb+QTO\nOL9VIuIYVeQe2EmWORFVJqJPQhSFQlSNIioANSsV3i7rucYhovsVbIgtp2K7ZL9jShgGChUDRFVA\nDCSqRrUeLpJ6uEDt0q72NwYAIAVAwMXDlovqqWd2Y7GsKog0TBZG7/eINhaW754cqKrWUAtmZSZV\nzYiyjlsfDKLA6pONBySQF22thjIjsjGrHWRVY1RjiJlRgYhWrCE+V5b23puJa0/1/JNWcltFBkBE\nALC1alQpYUsKaKSqEQJpgEYKQYVERIRVSZlblxCIYLMIntAaGgO4YKAwmBzd+ov/9s+b0bDnwBK9\n8rKCQjgbEACIAqoMEiLGyjds0YRZ708LioChVlkQEVtrmOqyJmOsYyLSqMyk5HQtLbTWiEEFbTDh\nXKkBAASK1RZ/KKSmtSQQEWfZuTFa8bHd4EyCVv2phijarKT181F7xtaYnEwOru7uhOMDSlA0UM4r\nZXaZl0IVDAigEGUCmIi4ros8z8+ZBaw4BVzoNlWl0EoHmZUTzwDIolVqxmwqcjr/39VzthwI0tpQ\nRK3LxXncw9iVCJs1X3T9Ww1AgnOWoQUQohpCy/PGe8vc/rblJoyeC2mrsABwkq1WyChIYZhsDBGT\ng9uPXdkPHx5aZzRi0O2jPB1tjY/BqsJQFQ9hVgV5UmURbWFAyywRoZQBpYj7VmvFSjHKUHPPd3aD\nWdbyWj7WwtU+GQmMwOgKLbZnMBTXwBgidgNtEBHZ9t+kG3rXNytRMp+q+FpOrUQycqthEVUqr5FE\ngIam0ykSq0HQxFDjZH5znLoXXtj/rTffVBgiKuq61qaiRAXcbjQVUXPBm9XyGlBUmPtvSGO0uJ+P\nGzbzYvdhQ98DD3XRLoT33ocnIjKGiGBp0xluWpzVgq971d89FM/39AOu2P57uSyAXI1qAxtw5zqe\nurr7x77yY8+kR0eTk9NlXWx3bs6aqddTbUiJmXKti3I5I6KmJhAlWUYG1Lo+7QUjsWEYUARAFFbB\nJjJmxZEW2tgWc91/f6tNEZQNwwJ2g6ERCgW1qlhJBX6l8jyxgrSpnHFkSaJKFNWg6wiKKEAugtEG\nNNiQBgQwMxm+j03Wcoyqqpw5LT2rbVRDKIAqVWWPGx/h5Bjv37jdvfVXXnzqaqdDu5f6qOkl575f\nnF433dcOlAGogWtNFpOxjCDBaBs7kNCqDDYETiwzO+tgCe6h6Nwl5JxjYuZ7wMOiwkqK7xWihwH9\nlVSGUPkKIeKfM1Z/fjliteIbvYBUnhuxxTJM5yBgfGnAg17ec2maElxofH+RP54PT+c32VfVfD4/\nODxMrMt7eW/Qt4kzsXWNAFWbWg0xKlTVe6+sFAiCtUJhinFlNBPn0mz/0mg8HvfywXA07HbTLIMI\nymW4eff45q2bpyenLcs2uXIfg1ZPdb5vw8rxoQ3m6tqqni9IC4MBIBIn9y5ARJuVYebKV5w5rcqs\nmwFojj88ev+92Swc38UR8KzFIB8i5X7azbrOiyphUGFezr/xU69wUS5n85k9mpAhN82y07Tb6VHG\nWdbJUpdkgyxxanXzLhEhUARaBV4iwbKGQI0AlVcvUFilhDiDFyyLpqgWxWKB81j45yH74LBBuvJ+\n7Er1yIUPaYlckj3wV+fExACyJJ1OpwASZ2aLMxV4jwIY9CG2RkxIGSW8apa6qpoc3pm++OWvcYte\nAGhU8b4CZrM5EnbOsaUsG3YG3TzNMpclaUJEzMzMa0RErRO8susiqvrercPbHx03Kik7YsZaBC4Q\n0+eD7/QQiyAbBmSTU2jVFqxKDA+HXa1j6zrO3/UAwHFZztVCGiyBTgZVrcum1EISTJv6g7sHvq7U\nusOjCZ9ODvcu7w62d0Sk8k3lK1WFqogAPLlz3d5lsi7LsuFw2OsM8zxLkkwJvV4XIBhix7qBFixI\ng0+IlKSpxAAxIUQ2zQqz3qd8KHis4ei9eOTBD0zrXBSpRRsCODfZAUqAae2+EbPxkwaIRhDFgEDF\nvBh1MwCg9OC961mDJeCBegm/mKVp/+OaG4L6qppX1B9pHPytv/Ut9uKZaXdnFwASaoMeC18XRVEW\ni4qZiJq6Co0vioLisUIz51ze6fZ6LkmzzGUuT/IkyzJyro13PfghHyJQK/jSQrx7jtnQX59TGB1D\n117Oeb4kAkaJoUJMAFSCuISXS9nftiiXVXGaJSAgB07meO8jP5ndcGx7nYzgFKObx/rayeFtOI6i\nWdZb3aUBAMs8cJxnqY5Ge3u7la/LhS/KpW8EgVXVeyHyc39GREiZIqWpy7rdzLlOnueDvkudcwk8\nGXCLP2CI46ZveQ8wIyLi86TciuoN1aafsJ6qK4+XbAs87kkRAiDDMCqVd8xIWL2EdTy+0+ksa9/v\ndi8Nh6iqt//ZG50uKIE2qIEiYD6NQyu+XGRZNk9xu8CSiMmyYZJCvHgCkWWyRESBwM4BSDjrqmJ7\ndX9NJYuiKJdlVVWtLKj3qtFnSekrjpx0nJue7Tyys729bVPC51DltI6l0L36KNmQrE2bcG4H2+Sh\nBsLanVCzinWQApDdSzvPPPnUcJDMFa/99j+ZnRXkODFA9C7Limr54jNPg/10fjLsoD9C52NUwOkC\nyGAArTGNcljKIbsFACk5inpfFUXJzGzUOQf1aomIzLnjYhiA9z5N8263R7uGGixD4+tqNpuXddWm\n7cQI1cjGo14nt2SjqPmszAo2Ef+9QUFKLsA54x4osJIjAw2IURW0EskNFeCSbPeRnct7SQh4/8ax\nr4UTV/mKEq7LWlSWs8UTT1wDyBegHIMcAARYlihqVBanBprLpCPH0XklUuWzyt9+59ZyXnf73dH2\nqNftcZYBUDLM7BtHRExgZiEiUmKCieRMHwmQjHf6CrB1VV3XvgKQ9kecutb2uRa7KhCgplpvdqLN\nLbl2pO8Dq5vqfZPl5w4CkaUEqgbrs1HkNhitQNXIG++898Y774kIsdZeqVGLGFQVSXF2Nrt985kv\nPA/jeyPkXTOKFjfkA8AoeopjAXqGs6EQEzlHwnBMRMuioMnHJwfxNkDO1lVjM7s13BoOh/3tcd7p\nZb3OKpaWsjA7dj5lIiIDZktEbHngHLmhZQ5IPvnkv3fEzOfMwn2eliHEVWS5k2YKr02tqt6HXq/3\n2sGHP/nv/Kvh5NTXWCCOdx5RfHQGjAB14B5rN68AYSjAgBL4+PjY17VUTbtY8EiIxMeTo9OTo1PO\nDgG4Ttbr9bv9wWh71Ov3AntfopPnzCxBXZrAbqhYggEx32/BLngX9R4Y9Mn6g1VIJwJraLoBUC/O\nuo4jIa6jMZtuf1zFcolIVSR4UuXE1E1cLhcAJncO9/evUH1799EM80qdbYAMIEAyswAzWDnT8zuC\n8GYYlRQtcjmHelJ5AOplOZ1V/kbrA/X7w8H+zu7Obrff29nZ+UzQ/ADaKAnTjb8/uQ0p3AM772Pc\n74A0Ik0zavzk46PuVjfeVjWNGFRVKecFUNxRx4JVXqaNHhO0jYWuLq8EAppNiKxI2BKRigJwZACU\n5eLwe6dpmnX7PZdYkc+b9do47UbIZaPwxKySYqSAsgMeDs9+p6SqbfmSPz6BZd8sfRmkRtnM/HqF\nHBGBlaACjaqigGgES9Vk7Gr41T6Se0MC4oLEIA0RMpe08TYC5cYOknSYdkhi1nFeRIxSJDWatgGC\nIBACPCWkCgTvhQBEaBu985UvZ7PFbD4vppZ5vigAsMlEfXc4fmS8099yEtDr9rv9HtqkXGsTwzmS\nAlZ4quW03rOjzSq+qKrw5KAezgs5U59WSgJUt5FkjsnBHp+GAx/mQAqMM7g0943I9Ex9QxyJbVt/\nwHXwlUjeGWCV0fduQ0xk7aG1QtdqUiVQmu0/fa036MOwF1ELNSADAkITTBuvgBIRGgCQKNFAivro\nzuTux0fT6dnR0dHR0REALf1wa4soIctVVZaLovVASb1a2tnZ+eKXvvLYk08HR3CklqyPikDr+A99\nIjy5ZpaeI7IlCddRgwCompl1g1h7AJifnpws3rsRPj7C29PQBYbAeJjr3Y9ZULRBOoBdyHNYa/nk\nZErcsUxtVo/5nrCZJRvQIhmACYBlJiI3GHTyvCjqXo98FIajoBSBAGMpqgaFqiZpomsMoBqlkYPr\nNycHB1J5X5VUeCR89YlntJGq8cNe1nt0ZzgYHH98PJlMFqczIjq8cePXl8vBzutfeOXHLz/zDEXa\n1GFEdA/GiHQuTWRIVeuq8U21+OjoxgfXFRknlHbdI089v9IDzEHdyRSHUyyBq8Bjj9qtYda/nGXc\nFevmgoPJ8eT4LLFmPBqyKMFLWzZkOSGNZjPjx0SqbTrJEMU2dmp4azhs7/X47jEBaSfP0oTZseW6\nqZjZMieJDc1FEYtA+p1MxRen073xON99FEB/e/jCT3758MODDz/8sJ+n89mpqGQ5D7c748Hesipn\ntT89vqs2+dZv/MZPcfbEk0+Xn6MwBoD3Pkj0TVWVvqz98cdH6dZIi/jM3uMAkrYUocDkcH7jCDPg\nm5fx+KOPONZnn3/hqF5k+ZYbjCXpSdr9/vff/863/2GjwlGVyF54VfcJNVFQBZtzcWuPuLy/1+t0\nffCTwwMJWleH3TTrdbsuYbiBzWyWZUyOeUP9ZbycF6//o3/80tPPPv3k01tbW51h32VZPt7e29t7\n6Ue+AA29zE6Pjz+eHN65fVBMjyZ3p0PQ+FF8cOPm9HCyc2lvZzim7cHnYZZzDgl18lT6sn1pdPX5\npySQqpdFMW8k+BoARBfzogP82JP5X/g3/0zX+uPb14+Oj567+jTyweiJpx9/+gvTRl955UsvPv/c\nr/5/f5c1VNzdOy9KajNqm1fl9Ve1CgBOWIH9Z5/XhKRc0vLUoWJVqqmqqVWwHoB1DJNlGWW9ZDCm\nvI+FvP2P/sn+pfFwp8899Siq08V8VtLN651OL887qeXvz4+LsprP5r5ZOM5Hl3rENBiNbt2ejEfu\n1gfvvtEfvvRHXgFgwEzpygu68AdWua9VQhVAJCaGCtqgQ7RZn30lInMAMP7Vb/3GS4+an/uXviSL\nu6HLve0BAYd3bzRZZ/TY3tvffW1374r3fv/R8UtfeJlVlZ1hlwNofTndEAdSOb9ualmgIGwNhm28\nUVXJwHGGRtQSaxtkEG4rcph8VRLUph2XunJRHVx/f5S504ODG2+/PZvNl2UFoJO48XhnMOjPZvPT\n6bRWX9WFigjceGfP5dkXX/nxx65cuTO52+8PDj66/aK+DIBUxQSWlSb9gWg6PctauylVeTR54eoT\nXPkqSp50Oil4J+elXv9oenr75lng5WKepllVVaM05cIXe91B3uuqqjcKwFknUIWKBPUbZV2ZY8Cr\nUMdRBAVAxLHNeA0g16lVR+TItgvN7DImAqx1GgXC0/LY+4Ydp8qhFtHqzuTmpUdeOr55tCwLWBYF\nsYNgMpnkg9473337x37yp37jN7+jqm11e7tOBlBS+Ivg36dkNNrEaPtHp5PnkS0Qyvk7b77x0nB0\n9+bt3atb1dxzJ0OCvnV7g+HhjZtud382mXCaeeDo+ICJKOtlttuzAEdFUE2I21Aca8uH1uWqZFWx\n9eSTT+XdHMUColmeW/V2zSdW1KpgQlsckFHinMsHSFJOSzLYGvcXc1UUBIhK4/2i8Tu7O9/4mZ+l\n9O+/8cYbxbxids66Ct7ZjK1dFOXWcPjyyy+/9vrrWd6ldbY5Rl2n3dYqfwNG3JfQ1xAEAgSpixB0\ncjx54tqj6tztu2Fncdrfw/ERQlVgONSEuCjJJVFxdDjh/khRLJvy3XffZQGYk7yNNCg8FCqtCWQo\niHSt9M8jx1cff7xDybSpKu8BqEjmVh4Pt7EncoGo9RA57TiXwmVEtHdl/+re/u2b153Liqoqqro/\nGOyNh1//+jeGO6PR9mh3vPuxnwBMoIUULiFVXS4Wx8fH+/v7b7z5JiK1WeuWFyEIzhcTIN4o+wr3\nG00NgqiV997rdDq9sn8ZafL+DMks7A6KvHYFAZgq0E/oaLEw6XhyOJm8d6OsvTDPZgXXU9/d2WWb\nwRIRuUZah8hIExvRHlOzgnbOdMSCmDuDERhVVRLBZQzKiUhavKrI3WAVfAOR7XCWJ5aU8O4bb1/Z\n3t0djx+/su+lCIJ+b0CK25MDqYpXv/1b3pdf+OKLd4+vTKdn07Mp1x0oZWnWGQ2LcqFs0l6nXNbv\nvPH2+JGRqsJmbGyWOpf3EmdjjGbdCBIJiFZVvQYR1LpATd77IgixqwrFcvHKKy9j2RwD3wYuXS8o\nQYcz5xw6Hl4rJgx0NMqyUa9oSAOm2YxhQEQSlVtUwOxSVlF48sZAFKQEQNSmLiEipizLqsYrtaUl\nrj2DW4MO1dYhRVupQrwKBGmsFkV5/dYyZzvcuURkjg+ntw8m33711TTLxftlWfhK9q/s74x3+r1h\n0tO6aFS1qf10Os2Pj3u9nIz7/jvffdG9uDPeKTQ22jRlM0ydVizBk9lw6Q2BjHiZ1xWtEufkrKu8\nsOMPbt34+a/++GJZCZAB3wno3yqGSTEY2IytklYh6rzQzqDkZOGDD3ZZl4wIdq4N22oAWYiXi+QV\ncwxqDKLRpq6STn7l8cdbnd9UlQUU2rpHSraVfnYpEAlWwWAmTpidKD31/A+9/+Zbla+zJJ3NZmVd\nskk5c7u7e2enp8vlQgTO8fRkCeXhcMiOlyLOJQpUpQcwHI6K5dETT127evWqZR4lnbpprOVIMMak\nSMoi+Lry3qtBr7cFi1qlmw9goQLyWQrJHHwjvli88OwzI6dfvYIbt3EbeGeJvQ7GVWALNcFHoFLJ\nThrQLISlD96D23I9SldyIapiVI2qUTGrWE0b7SbmRvzWaAuEWhuBWiYWqwgEImIiOq+LpJXjztY6\nEMGQc33X6XnFIsryaLmoClL40quAOMk6PRGReu36Ei8XS2YejbaVudPLZvO59/6Dd9//wpde7uQ5\nAEodZ85Z50NbBaRBhZ3N0WXms6J2sMPhsKibvNdHQ8i8qnqVNITjjw+euHoFd377h666LeOzW3gT\nOCzRL5FT6PQRA6oizrTFY1AgBzgGiPearMvS1hUmqho3Y0kGSZo1jR9dGrNzMj1dcYRZJbQ+NDNZ\n5rYYiIigBHbMGZC0ORwlJwpZhLuns+Pjo+WiSdkO8oyZ0yxNfFpqLUFOz+bzRdnppXln4JwTMk0l\n3ldR9cUf+eG93b087YgKJ5lL2DeSuVWugBMRLwoVQhYAgDNX3p1u7exSYuGdGk9BEXU6nQ539vDR\nIhWvEeMtvHuGqgUfCpkiBzLAA66tFwSWbeVfMTlEb3xukleKWhUR2ta3R6iBCkaDcb/Tl2nlD28O\nqbWUGWU5SIhIACVVJmcyJWrUpPlW3h8PR2MfKOds5/LV1199/WhyoPPCDbTXp2E/96UXFZ3PKy06\ng66I1L4p6sWiWeynPZ8al+Zpb5SlgyvXLkuIg0u7ABwRjDGRXFvWYIyJ1Oulqlo0itCg9ojMcPuX\n90xTZTZT0hrIHGN+Gu++3yxOktQ9spUfTYpZL1ucVWupQIIV4wiw7VKvbT0agUGEBoogVV03qJEx\nulGN7pJkPB4753xZtT2LICLnmEmVNzCGGqLIFrqy6V41RlQqO9uXn3++Oj46fGR3T2ScOg5FVcTT\n47tzVWXnxHsiSl3imNNeB0zFsnRpDmDn0vj5H/pDmhBRsrYkbWGIXS2zJTJcB1FfN7XAkDQiThOi\nEDyigjSJ1Kh2Oh0AiXM4K61SnoI22nQj0GxwpxWoBHAtswTgRgAEIrUXARqNcV2TBTKm9s3u7q6I\n+CBtyKm9ZTaJrBwyEq1gs8jWMCdwbFNyOayzjkndLJa7j1+9fO3J48nhOHedJGOCn56NRkeTyWQ+\nmxfrqjaXZb3REEpgIselr68+/WSedQrxnLSlSLiPVFWjtFHyJGWJq+x0CErqfQAbrkNlbVbN5wDm\n0+M+KjW+KlFX1TYwXaNbAiJQAhZwazatmKV4aFCY4qpiDUTMlHdybaRVBmhTEmSMc/AeADmCz5iN\ndY6IFQR7UTjUWBg4NXjlS1/59q/+fe+rsqmk8rODj+bzmRdxzvkgznKadzLnPKmSY+IqSr+XwtAi\n1OySh90qgLIsJaoq5BMdq23nAylKP+PGA3BAeTbNLAhIDIaAA+6um4rbDRXWheDZeld+Gp3X9hGR\nc66TdyRqFGVmYgNatey0IUNDxGzsuueFickSGQNDQUijREAgZVF1h8O2DOB0Pi+Loqiqpqq992xW\nhXAiWnofvC+rYnY6Lxt59bVXP/j+u83n6Ws/l/o1qRcRkcZ7aRteFQA7Wi4Lal0NMgA6wDbQRdtm\nCtuGSYEM6AA9CwZAqLRs5YUosjJHCxhnLKkJK9lIKN8eUjfTpqqqOZFCASJeNRFS6pyqgsi5jF0m\naphdlgzStEeRyKI2agCK6ezk+IO336+Wp/2U/N3DqvZRakBU1QcvpUcJEREito7AWU+Oqg/ffeuD\nZ354+e73b778pS+6JBsOVwZGVaOqj8GLl7jyZaFKCqhIJcxM7KIKmAmJlMt+DgCz4vTdt79zUNdT\nxbJYqeY+MFpLEwNkkOcwvTxYKoy2eifIqkwTFCV1Ti3BrIzgOe09uuu9Z0uq6uwqt2657WnZKIDh\nDtmMSWHdKsOZEEAJoF5B5ju/8Ztnd4+dkXnZICrp6tkoqiMitOsAVbAlUV3Ml+V8SUn32aefpU5e\nLhtNKc+EHEMJIA1aiYh/wHCBC+GiVVFNlqZSLgBQkIOPPuKGNIRF227fyiWw1eJrAx9RBOSGKMlI\ndFXEgMQAIDDbVRdKq9djEFjnnKtUhqNtZi7LOQCNq64kgNvuh7X0E4xTZTBTdGwSMWC0hpFAVM7K\no7vHoa7VKoeKRCpfAyACKbWqgkCOnXiRKAikrKr40z/3cy+89FIddXY2H/S6iXPL5ZLTRGOURpQu\nZi20VtJadg6+rlSVEkYjjfdQIoNeJwcw7A/LuUcDZy5sX6vIhwPMChzUmAL9ANxZeCxoZQ1ZnVmB\nrPvSdNY5bRRAlmVZmmJd3ifWETETW3arCo5IlBARkR2QBVMKS5Q5IpLYahCJBCENUCQ2yxISWkxn\n1EI6BUWwudA1rpuflQsQl0X15a997Utf+smT4DVgazD6zm/9VpamTzx1LbMJYFSpbaPavHe2FOOq\nMPH8Q4rwEopisZVzM1+e3T0TDylWiqllmQeun2IOTIERcCXFE49nVy7vdl22LvdZIdJ7S9oBtsY3\nKz8xSRjAqn7fZpQ4spZTx4lraypXZevI2+obJKSqahyAGBVGFRBAoqbOuYHrWiJUs+roYrNsZKer\nsmCFQK9cufL1P/71s2I5LavvvPH6jbc+OD05+fJXvnI0Odp5dNc5p22t4oOqKz65/NL4erkcbm35\nujo+lqVgUaBt2i+AEqgAAV5weOmL137ohae6Hc7ILE6Ojm5/yAAyKLQBWQYZKEh53RWilWQuLerq\nqSefUbQ1KpnrZ/3BLtmVrdQYgY4GApFGUgsKoAgEJcfnmXpLLAu8/c773fF+eedQtfvM8/vVbv9w\nu3N053h+OiWmyiuYpN1OXlx30Nse/sk/++d0NJxMjt98453px6eHH092dnYo4WldYHo6HAxdwtQo\nGYIonRejBk1Aicta6BbAqiJGfYem7x/82Z/5k0T06vtYLnAMAEiAXaCXotPFUqEBt793fX7nsDvs\nhaCAoJLPSLtLm/FJeefSGIZVlRLb39qCYi1Hn4o+NjoZSRnQTp7v7+8fxWq0vYUkH27vMmd5Z2uS\nn87mZ7JYgpIsc53+wA2615585ukXX+yPxnXTOHanH5++9+73d3Z2xuNxnnfUQERCkFYt0r3lXffI\nV1xNYlACAr3/4fUvf/FqNSveXuAa8PJlDLd6sCyBJEIiXDmbnkognC3qyvDVa8/AmPev32iZxUxW\n13kwVV37MgARUyagfDhUS5RwN90+j8ecS35bPEMtXjEPEH4Aja84y5bLZSXFi3/4i089tg9ZVItj\ntdnIdUa7V4qqampPlseX9/Yfu4phX1V9o9O5V9Hvv/3ewa3bXe7s7o7Hl8ZkkeSu100TR4BCoOJB\njHjPsJhVXABKpGqCqCj87es3XvrX/+XT27f+3T/63LPXnvZNOTk6dM4tvTx67YnGcne8e+vG7bfe\neue9W7caddnO5cefeGpBOQNgSh8mGcSpEmdZ3slzH6gtIFBi+kRty2eSBqj3IhWBXOaSQcYN97u9\nbDSls5P5ydS5zqX+aOfyTuXjzY9PTt+/LaVXgyzLnn/mhe+++b1B3tu98vhLr7zQSXO0eZGL+2xZ\n87Bri6dGg0KEyB99cP30gxvvvf2rX/vK14fj7vLkODdaQV95/oUzkd3HrjZu+OIPv/wzP0uvvfnm\n3/l7v/Lbv/nts+m018kYFiCSFqMQw3BoNLCIYY5GjGPQ7qUdAM62weJVNfXK326FywgiaVQNcOwQ\ndT33YlWKQEQgV82VY4dCuSir49OTyWTiEh6NRvn4ytbuNY3wjT9aLgGAXb+XZTuODItJ/ue/+jfr\noGkHX33pmX5/VZaxwmPSys6KK1irBcUFjAhQh85dnRA7Wmpx9/r1X/nfXnnxeY5HVPq9Ya//xJOT\ngw+P3nrjq3/qT33r1vWrez0K3qteu7L/M9/45v/5N/76a7/97b1Hd1bNerQOzsSo1rJacmSZmRIG\nsP/o3qdoBAB00Vb5ULKcMAciAhEb7vV6eZ63X7ElIvKNiPeVr6TyLs92Lo2GoyHB3bh5OH5kHDTu\nPtq/eu3xdpDD74BYQAQGOWBnMPSFRwKJzrOSwc6l8dF3j771rW9deeXlyZ1Dl3UEtKyV0+Tak1c/\neO/6siq4bWzkVrAJUHLs4AxbRwmDef/y5fEjO+WiSPOHFq0RfTazAKiqD40vvRe/KIpeZ9WrCib1\n4huvquPRWKFZkqXdThQCuddefTNIMxx3f/wnXq7q38nMMzE4r9/IiAKAsimPT713SGvN+pSAia49\ndvnXX38jDrYWkfNsaHsdjVQul60uT4gZGqdn096AYMm61XgKkUhGKSgY+5evSO3NQxK/GqOGsFki\n0bo+54ayHS8UoFWzZNvx3jfBz+fzLLmqERqUASlFggDY3t4GVklmR06J33zjHUQeDvr7+zuDYbeo\npqt+T9xjalc3g3VUer0251PnVNU5V5ReVQNwcHAgw/4jflAklWQFdVyepQnRtb3dt/7xa6/88Z89\nPLrrIgqRyut333knzbgWz4iQRrwIK8FADWLUpJOLCKfu2uNPpB2nUGa3sixtXvdewCBG2gbm9dyQ\n9d1vNFWKqJgaD+qZK5YLL6HX625+WPuymNMHNw8yl0ssnn7mmshSROjhAaWHkTMkoiISRAoogA8+\n/JgScCzZpXlnwD5ZFBhkPWaen51Oz84mH02W/tbk5Oy779764MadnUd62ggjs4OdsQjApMRkHQwJ\nJS7vPf3ii1ujnaBkDatfQ4QW0bjV2ApgVZfRcgoAGYd4EZwjohZnZIxFpVm3k+WuqouFnw06A3hB\nxw1GPQCIYpyEBtZkUYGm/2u/9rcT62qcffWrX8m6iYRAROo/TXW2/7lYLQWAWHkJQWpNmnpv0AHw\n5/6L/+Yv/8X/4GevmE5nmF/y6dJFLY/63ca6jJJf+Tt/62xWcuZK1UGfnnx2e7pYzIuG0YSqKF3W\nacXCWIqKLMt2dnYHveHKoDBiJGs4rkrFLqw0UVs/ZnVV8YYH2YFVVSOtippJQzuy8v6K0fNxeSry\n7vfe6eS5L/2Vx/ddkgCQps3RfZpmbI3g/We2RBQ5dRR8u99/7Cs/+s2vXb35rZucnfXDtDfopMxZ\ngCSqhii6XpfAVJblyclJ2ejCiy/ACFgsF5fyLswqLANKMpdtj8cAxAtZCNo2edJVsfT9HNHzNAeg\n93VznVv1CPHq60DRNF6Db9HvRvdX60UHKsvq+GTqg9/eGp7Z6aOXd13OIiLq217fT+fUA5jFhAjK\nHKlDogBOF8fpkA4jZAHnY35WdB1yN7PdjDg/MVTX/mxRqCHHeUhly8GnwmA76OTeS9ZZ6YKgMhqP\n+t2uqm7WajXBb5q8i2xQpDY7qzGSuafuut0X5zBaRI/unAHIMiZyGgAS1YtLsHE+iAQ6OytC44mx\nNeju7u4GX3kpEUVI6N7OgE8K8rlTsaHgpa25IceceAC9AQU9/frPPTU5Pp3eLX7sla8c3TmqF7OT\nop4uisnRtNZgrc3yjFOn0ZFXdUogOJcAsMyWmIzZfWT36pWrWZ6t6u5XU1EufHoTKa4H0rTZTRjV\nAA3QGHWTcPEPgNpLVdXSQEEK0lVX40XfQFV7pmzy0fGHtw4W5bJYLp999rko3odSoqgqRdJ76WFS\ntkkGYAFHOMX8aHppmyHlx3fOELx6HT+yq0pZ3p1F3JhMfvO7Z4cnwQuok2ikSlY9ys6AM8J3X/2H\n+1dfKCYT18l3d3ceG1ypPr7e63W7zs2VKRJZYuY22trenU27bZs0vFOgEW9txqmjSFXjqW3ZEdVG\niYgcE4i0hPpmUTQW4+28WJSdTl+MbCkMc2wqGK7qanZS/PY/fH3QH79z67tf/8bXs07He+89vGjT\nNNZaMh4rWBABWHvRD1VvRHiY+Vx/ARAbFSBVr9VX/8hPJ6mlBJJob4tnJ6ev/9Nf/8KPvnwlH+fD\nLHXXjzwhErGTgNp70GowD1dVAGbX3/pOO73xXeBbf/dhy9M6hBZIMMjBzrBN055j19/qbw3H49Gl\nvN/pbo8H/UG/13Uu63QyconzzMyS5M5wkvi8k2mYSXUmtal9rb6F7w3QbA2Hf/vv/UotenQ8ffqZ\na1f29ypfV75oN5e1NkYtV9twrQ02eqbuaWlsWoBDqqqWYJQQDeP27du7O+NyOdeIg48+IrU/+yf+\ntSzvnkzndfCjkYy2dt758Ojwo8nJ2TRaQzZDBMQD92CWT3eON0caBsyq9qMKqIAZcHBx5CdnJX8K\nWdghuh2bZZ1ub7g1Wp55ilnW67/wh752+/qbnJA0ikRdwsaQRK1EiZkNt6PTZKM2UezG4xjCWrKk\nZo3CpDH477/zxhd/8pnldB4Fy2VEEAWdnczn89NFWZWlwvD+7nDYzY7PZoeT46KslKBGdVO5/u7R\n52TT+uBwghnCDAtgAZwPH/g/3vybn5jC/FmUXcHW1mB7nOedZ599rt/vb29vdzqd2jjqUa/fH2aY\n3Pzglb/0c8Mh7Qzw1nfR6wKlltO5RFnO57PFom5oodAIdjQcDkVQ1F4j9FPSq79ndN/zb44F/uTI\n1x+I70B1G3duz+5gBtx59cHaBMDj49Hs7gcElznPBHYuTVML63u+8n46nR0WZRBRuKLWCgjEbJ1p\nhxX9/tJ9zx8f8jdgGeHzTYD4QYk0s4kIl6U/nkBF8k42nU5ddAPu1rm/2snmXk/mOj85XixKVXgq\nyOY/cF307x/9HnEKcIQqNKRkLMjC5Q5WQaqQJKXh1jBPB76S6XKeZLkYiNGgqJvq91+y/uApy7j8\naEHsYkCWYe/R3cXsjABRLYulryokXQ+aL/1ZPasjpCGv4sn+3jHrHMr/3k38/h3SbDZn6PUPT4oS\naDCbL27fPpgeTytUvlZVndW1gqrg52UkgrCZLqDuD0DB/8FT3ulOTpu338EMOAL+/L/3P+130B9C\nLVQQIpLLj5QBEpygFCTzohJnlS6mAZ2PI8e9nuoPaI/WZAyIKHyiGP1fBNLl9HvvXz8AbgIEPAN0\nFdtAJ7EYdJdNdVIWFN18VvJgfHR2AragwAj8c9/8iZsfHR0dTIJIVdVVHYuHPGFmV/PQiWhWlZ++\nv2KM4bNni//BkOP+/EQqoACuAY/3cfWS7Q27KTly6TT20tof1frs47u/8b2PkFqQacfM81/7H/6K\nBhPqkiynLnHOuU527lLZS1fhq3q+mC+WUK18NZvPl/PFzdt3pmdn8+XyYDI5m569+cabp7Pp8elJ\nsYBvEGLLyMgGMd4zPf9fBKKUJ6eHx4ABXgCeu7Ldz3n3yh4pScQQgqJ6585kb2/n+snkg5kKEoCd\nEo96NtYwg+HFyUJ17ps2R98DQMAwh4r0M+wMukD3i89faQ9oM5rVRlO4jwwgBkVQ41Z/e99AE40q\nElX19GRWhfr05MTXzbdff71YVEdHR0fHx6+99k9nn/m+mn9uujs7fvO978+AS8APXbadRAfdTp4m\niXGUOEqQK9iyXOr+G3/sq3/1//nWcegREalwQwx+eGB7c97Jhpuq541YFgDYOFl/OYCoKhKoxqpe\n9rodkTAadKqFV40aoUqDNAf18NhIgnz9Sy+gbXoKmlhXVwUiqepsPlVVjYiKKgoC+SAisv/YVY3k\ny2rpaxHJ8u5iOV8uFlXdVF6n07Pju8fFsrxx/frJ9PTo6Gi5WNy+NVusZfs/+vPf3Op133nrLoAv\nA3vDNDeqWkvRdLv5oNuRNLqmc/UxnkxP9px7qj/QhpeNWiJOIkoJnQe/4QLYxBbrlD2Cnpd70KqI\nQN25osvzYjbvDfrV2Tx43Pzo1tOvfBFptzuUu3cODz86PJtOjyYTESnL2vtqZysbdLv9fv/a/hWX\nZsLkmG/fvp3lq2Qqoiar6ivS6ObTCVZJVnUGWhZ51KwDTS0ssLutz2wD8P55bJTl8Ko7XJxlBm5X\nALCXw4fauTxpVBeFcCrST9Utm+lieuYiqqJ47Or+W2/d7PRH4pccqs+oav2BSZSj0aI6+uhgWcyS\nxL33j34bAigD6AAdzvb2r6jRVlQ1ynQ2feHJ56bTs//2v/vPf/Hf/4Wj42m2zr9+LrJElggQKc8/\nW4Uq9Z4yJgBk2VfNAgBQChSY1UVmlA3qlKqpeiI1Qo4OPz4+5cHh8QxMIh6qJGLwu4u2fOFMVF/f\n+uB9aEPBZwTHmlnJE+mlOsigzYykQFhAi+0sd4Ib3/ve0Z3JL/7CL/zSL/0SMkb2eW/Jl5XUXpv7\nX+xxnve9b5YBM1eNb1Xsux5HizCZ1tOFlEEX9WK+PJ3ePTo8PTou5hOPU5u+d+cIhkNdMoElENeC\n/qBZzpLE3TdB9cFxW0ufTHBu3iVYfSXPv/C8b0ooRMRLxRuAJHdOVwXjVNTVaDzyQXy5PDg4ePUf\nX6doAIiGNqJ9fhsaEY1quAhwv//ue9/+zd988cUXs07u6+qFH35OI+75STv+1TKCaNQ8z1OXDAb9\nHrAE3gb2J/DdgC5McpYL1ZUzDT4ajW7fOrp+prfKI59kIYgFEJR/+it//P/9e/93+shOko2b4+PE\n5pvts5sJ0c85BdIHUoXLuv2xyxKTZFk5W2ZpYjo56qouah98uSjatmIR8VHIgigbDfJf/Qd//5d/\n+X+cn5z2Bw9tsXeWNaJ9YRMRPfvcc9PplBLGw+bnARKEAbbsUsvMb/2ztwE8ASDB7T7PM3qv9nYa\nXa3OqgSZzJbzCgvNaviEKCVFAGzH/GGLWUBvbP6z//g/+dN/4efbvkqooJhH8aaN4XxiKkej/nyw\ngaps8lEKPm/yJqtiQKx1UXcUft2c185Ab6k5KgRSqEios25HjYgvtVZ1Dmijwut8jkKDNtrO4eJ2\n3HUplRoRBK/aaw9c50xk/e4IN9gqz07GncHe7s4jV5/+E9/4E7/y5uzKAOhvA+elaBvMhSOy5/Py\naP0+C/MTKUTwdoADPCCAB3pAz6I/tPvj/nA43N/f39nbfeb55/r9/u7u3vjSaLTVG/S7SdrDoA9m\nrLoiDQAsTyECCYh+PjsDViky6mXtAwQJiUuw1sEyr0MQX1eVeBWh0Gj7Ui+7enIRgqEWQwBIV6kg\nBViDlrUHiZj2NYPrJFjQzWxFnnfEl6PB8Nknn/8Pf+kv/9f/16uPWuxd3UZ/aNZLzht23wd2CTOv\n4OZFWusVhiNkl/nkrgSPJENdgAhVA42YrhvIgNV8dA944LyeRgBZdyIwQMAJwMCWQZpga8s9/tiV\na09eefa557/5R39itDUeDPpEZMdjUARbMIMcVKAKkaYsVapiXoaqQkbnekdrofbFBkF8KFVVA2lA\nO5DEQNtUedW0zT8RgEa0rRNENHS9rcHoH/zar/2ZX/zv2y6vr/30K/W6D2s1n3+jg8V7dc6tBuav\nh10q1PyIRULIVrl1qEI3XjlXtGUgq4I0hAgRaIQ8aFh2u5B5ZomtSlAfWqvjDAB8YRv9gXNphqiF\nLxnWpRkzn/X2/853/lkr1C16J0CBzGI8tFudzu7u8C/9xZ8fDfJRN88Y+SOXmbNOt0dZlzsZKxtj\nwAwiVFOEVWJ19cKSINPZ/Jf/+t/4r/7L//UjDws8vp3vP/lM1usKWKVYSc2925DaUflMq6lOhjRq\nEDXXgNyiA7DFRhJ+taVazHk+HVUUEdCI6l5njwlk0BoDE0GKLEHmoA06jIwts9kTIYCtJUaadAAo\npC6bsypoBBjUsVneJcMgglKREFvLlhbl8vjkZDza4i5Rg3mgSvTm5OSDGRbr1hG7EcxvX0nmgLaB\nMAK7W/z4ted63dF4Z1yUZa1iVYnR9mx9sqpsc9J8ELW8wh+mNX5PAalFbqG6eubWffEbtWPtvFcF\nYrh4aYZufLuqT0xA6/FQXUZGSCwcYxwRIzoOLrXwgQza6SjHtThn2JG2paE2oS7US1YClHmjM1+X\ngCIB2EPfCTkzd/JOb7jFrpN2MwVTOxFNtXVW76t42ARAm3Os2SXQ+4bqteJyj93XuAK2/z8j9eXP\nsQoQTAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=100x100 at 0x7F11648EEB00>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "lqV0TF6tYY4J",
        "colab_type": "code",
        "outputId": "7a2a3429-970d-4eff-cbf6-7a532aef174f",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "batch_size = 10\n",
        "num_classes = 2\n",
        "epochs = 100\n",
        "input_shape = (100, 100, 3)\n",
        "\n",
        "# encode text category labels\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "\n",
        "le = LabelEncoder()\n",
        "le.fit(train_labels)\n",
        "train_labels_enc = le.transform(train_labels)\n",
        "validation_labels_enc = le.transform(validation_labels)\n",
        "\n",
        "print(train_labels[695:700], train_labels_enc[695:700])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "['false', 'true', 'false', 'false', 'true'] [0 1 0 0 1]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UkxTltLwYY1J",
        "colab_type": "code",
        "outputId": "78612424-7ef8-464f-ad62-45364fbc133c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 861
        }
      },
      "source": [
        "from keras.applications import vgg16\n",
        "from keras.models import Model\n",
        "import keras\n",
        "\n",
        "vgg = vgg16.VGG16(include_top=False, weights='imagenet', \n",
        "                                     input_shape=input_shape)\n",
        "\n",
        "output = vgg.layers[-1].output\n",
        "output = keras.layers.Flatten()(output)\n",
        "vgg_model = Model(vgg.input, output)\n",
        "\n",
        "vgg_model.trainable = False\n",
        "for layer in vgg_model.layers:\n",
        "    layer.trainable = False\n",
        "    \n",
        "import pandas as pd\n",
        "pd.set_option('max_colwidth', -1)\n",
        "layers = [(layer, layer.name, layer.trainable) for layer in vgg_model.layers]\n",
        "pd.DataFrame(layers, columns=['Layer Type', 'Layer Name', 'Layer Trainable'])  "
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:66: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:541: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4432: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4267: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:190: The name tf.get_default_session is deprecated. Please use tf.compat.v1.get_default_session instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:197: The name tf.ConfigProto is deprecated. Please use tf.compat.v1.ConfigProto instead.\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "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>Layer Type</th>\n",
              "      <th>Layer Name</th>\n",
              "      <th>Layer Trainable</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>&lt;keras.engine.input_layer.InputLayer object at 0x7f115cd3e5f8&gt;</td>\n",
              "      <td>input_1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115cd3e748&gt;</td>\n",
              "      <td>block1_conv1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115cd3e978&gt;</td>\n",
              "      <td>block1_conv2</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>&lt;keras.layers.pooling.MaxPooling2D object at 0x7f115ccdcc88&gt;</td>\n",
              "      <td>block1_pool</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115ccdc470&gt;</td>\n",
              "      <td>block2_conv1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c48e400&gt;</td>\n",
              "      <td>block2_conv2</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>&lt;keras.layers.pooling.MaxPooling2D object at 0x7f115c4a1c88&gt;</td>\n",
              "      <td>block2_pool</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c4a12e8&gt;</td>\n",
              "      <td>block3_conv1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>8</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c44e7b8&gt;</td>\n",
              "      <td>block3_conv2</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>9</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c467a90&gt;</td>\n",
              "      <td>block3_conv3</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>10</th>\n",
              "      <td>&lt;keras.layers.pooling.MaxPooling2D object at 0x7f115c47cd30&gt;</td>\n",
              "      <td>block3_pool</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>11</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c47c860&gt;</td>\n",
              "      <td>block4_conv1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>12</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c42bb70&gt;</td>\n",
              "      <td>block4_conv2</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>13</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c43edd8&gt;</td>\n",
              "      <td>block4_conv3</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>14</th>\n",
              "      <td>&lt;keras.layers.pooling.MaxPooling2D object at 0x7f115c3d6a90&gt;</td>\n",
              "      <td>block4_pool</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c3d6a58&gt;</td>\n",
              "      <td>block5_conv1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>16</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c386dd8&gt;</td>\n",
              "      <td>block5_conv2</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>17</th>\n",
              "      <td>&lt;keras.layers.convolutional.Conv2D object at 0x7f115c39b710&gt;</td>\n",
              "      <td>block5_conv3</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>18</th>\n",
              "      <td>&lt;keras.layers.pooling.MaxPooling2D object at 0x7f115c3b0cc0&gt;</td>\n",
              "      <td>block5_pool</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>19</th>\n",
              "      <td>&lt;keras.layers.core.Flatten object at 0x7f115cd3e630&gt;</td>\n",
              "      <td>flatten_1</td>\n",
              "      <td>False</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "                                                        Layer Type  ... Layer Trainable\n",
              "0   <keras.engine.input_layer.InputLayer object at 0x7f115cd3e5f8>  ...  False         \n",
              "1   <keras.layers.convolutional.Conv2D object at 0x7f115cd3e748>    ...  False         \n",
              "2   <keras.layers.convolutional.Conv2D object at 0x7f115cd3e978>    ...  False         \n",
              "3   <keras.layers.pooling.MaxPooling2D object at 0x7f115ccdcc88>    ...  False         \n",
              "4   <keras.layers.convolutional.Conv2D object at 0x7f115ccdc470>    ...  False         \n",
              "5   <keras.layers.convolutional.Conv2D object at 0x7f115c48e400>    ...  False         \n",
              "6   <keras.layers.pooling.MaxPooling2D object at 0x7f115c4a1c88>    ...  False         \n",
              "7   <keras.layers.convolutional.Conv2D object at 0x7f115c4a12e8>    ...  False         \n",
              "8   <keras.layers.convolutional.Conv2D object at 0x7f115c44e7b8>    ...  False         \n",
              "9   <keras.layers.convolutional.Conv2D object at 0x7f115c467a90>    ...  False         \n",
              "10  <keras.layers.pooling.MaxPooling2D object at 0x7f115c47cd30>    ...  False         \n",
              "11  <keras.layers.convolutional.Conv2D object at 0x7f115c47c860>    ...  False         \n",
              "12  <keras.layers.convolutional.Conv2D object at 0x7f115c42bb70>    ...  False         \n",
              "13  <keras.layers.convolutional.Conv2D object at 0x7f115c43edd8>    ...  False         \n",
              "14  <keras.layers.pooling.MaxPooling2D object at 0x7f115c3d6a90>    ...  False         \n",
              "15  <keras.layers.convolutional.Conv2D object at 0x7f115c3d6a58>    ...  False         \n",
              "16  <keras.layers.convolutional.Conv2D object at 0x7f115c386dd8>    ...  False         \n",
              "17  <keras.layers.convolutional.Conv2D object at 0x7f115c39b710>    ...  False         \n",
              "18  <keras.layers.pooling.MaxPooling2D object at 0x7f115c3b0cc0>    ...  False         \n",
              "19  <keras.layers.core.Flatten object at 0x7f115cd3e630>            ...  False         \n",
              "\n",
              "[20 rows x 3 columns]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 8
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "WdFdxkZoYYyL",
        "colab_type": "code",
        "outputId": "1639b5c9-54ca-4790-a671-636ec4e114fd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 302
        }
      },
      "source": [
        "bottleneck_feature_example = vgg.predict(train_imgs_scaled[0:1])\n",
        "print(bottleneck_feature_example.shape)\n",
        "plt.imshow(bottleneck_feature_example[0][:,:,0])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(1, 3, 3, 512)\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f90c040ef28>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 6
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQ8AAAD8CAYAAABpXiE9AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\neEhqYnhIavL/UgX9OpgWTYkAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XFp95KD8YYvs",
        "colab_type": "code",
        "outputId": "a861851a-2c0c-4821-d670-f24bd998d037",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        }
      },
      "source": [
        "def get_bottleneck_features(model, input_imgs):\n",
        "    features = model.predict(input_imgs, verbose=0)\n",
        "    return features\n",
        "    \n",
        "train_features_vgg = get_bottleneck_features(vgg_model, train_imgs_scaled)\n",
        "validation_features_vgg = get_bottleneck_features(vgg_model, validation_imgs_scaled)\n",
        "\n",
        "print('Train Bottleneck Features:', train_features_vgg.shape, \n",
        "      '\\tValidation Bottleneck Features:', validation_features_vgg.shape)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train Bottleneck Features: (1400, 4608) \tValidation Bottleneck Features: (200, 4608)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "pQd0MHg8YYs-",
        "colab_type": "code",
        "outputId": "fa99fa7f-6376-4a41-d09d-6051e1df396e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 474
        }
      },
      "source": [
        "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout, InputLayer\n",
        "from keras.models import Sequential\n",
        "from keras import optimizers\n",
        "\n",
        "input_shape = vgg_model.output_shape[1]\n",
        "\n",
        "model = Sequential()\n",
        "model.add(InputLayer(input_shape=(input_shape,)))\n",
        "model.add(Dense(512, activation='relu', input_dim=input_shape))\n",
        "model.add(Dropout(0.3))\n",
        "model.add(Dense(512, activation='relu'))\n",
        "model.add(Dropout(0.3))\n",
        "model.add(Dense(1, activation='sigmoid'))\n",
        "\n",
        "model.compile(loss='binary_crossentropy',\n",
        "              optimizer=optimizers.RMSprop(lr=1e-4),\n",
        "              metrics=['accuracy'])\n",
        "\n",
        "model.summary()"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3733: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/optimizers.py:793: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
            "\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/nn_impl.py:180: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.where in 2.0, which has the same broadcast rule as np.where\n",
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "dense_1 (Dense)              (None, 512)               2359808   \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dropout_2 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 1)                 513       \n",
            "=================================================================\n",
            "Total params: 2,622,977\n",
            "Trainable params: 2,622,977\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "q4-G-kcYYYqW",
        "colab_type": "code",
        "outputId": "b338f03f-e6a0-4d97-9070-20d7ce796fe2",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        }
      },
      "source": [
        "history = model.fit(x=train_features_vgg, y=train_labels_enc,\n",
        "                    validation_data=(validation_features_vgg, validation_labels_enc),\n",
        "                    batch_size=batch_size,\n",
        "                    epochs=epochs,\n",
        "                    verbose=1)"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Train on 1400 samples, validate on 200 samples\n",
            "Epoch 1/30\n",
            "1400/1400 [==============================] - 1s 604us/step - loss: 0.1343 - acc: 0.9550 - val_loss: 0.0240 - val_acc: 0.9950\n",
            "Epoch 2/30\n",
            "1400/1400 [==============================] - 0s 248us/step - loss: 0.0295 - acc: 0.9871 - val_loss: 0.0197 - val_acc: 0.9950\n",
            "Epoch 3/30\n",
            "1400/1400 [==============================] - 0s 256us/step - loss: 0.0208 - acc: 0.9921 - val_loss: 0.0464 - val_acc: 0.9750\n",
            "Epoch 4/30\n",
            "1400/1400 [==============================] - 0s 244us/step - loss: 0.0179 - acc: 0.9886 - val_loss: 0.0197 - val_acc: 0.9900\n",
            "Epoch 5/30\n",
            "1400/1400 [==============================] - 0s 258us/step - loss: 0.0139 - acc: 0.9950 - val_loss: 0.0422 - val_acc: 0.9750\n",
            "Epoch 6/30\n",
            "1400/1400 [==============================] - 0s 266us/step - loss: 0.0136 - acc: 0.9936 - val_loss: 0.0070 - val_acc: 0.9950\n",
            "Epoch 7/30\n",
            "1400/1400 [==============================] - 0s 273us/step - loss: 0.0150 - acc: 0.9936 - val_loss: 0.0056 - val_acc: 1.0000\n",
            "Epoch 8/30\n",
            "1400/1400 [==============================] - 0s 265us/step - loss: 0.0132 - acc: 0.9936 - val_loss: 0.0185 - val_acc: 0.9950\n",
            "Epoch 9/30\n",
            "1400/1400 [==============================] - 0s 262us/step - loss: 0.0126 - acc: 0.9943 - val_loss: 0.0330 - val_acc: 0.9800\n",
            "Epoch 10/30\n",
            "1400/1400 [==============================] - 0s 261us/step - loss: 0.0104 - acc: 0.9936 - val_loss: 0.0048 - val_acc: 1.0000\n",
            "Epoch 11/30\n",
            "1400/1400 [==============================] - 0s 266us/step - loss: 0.0184 - acc: 0.9957 - val_loss: 0.0172 - val_acc: 0.9950\n",
            "Epoch 12/30\n",
            "1400/1400 [==============================] - 0s 255us/step - loss: 0.0112 - acc: 0.9950 - val_loss: 0.0034 - val_acc: 1.0000\n",
            "Epoch 13/30\n",
            "1400/1400 [==============================] - 0s 262us/step - loss: 0.0101 - acc: 0.9957 - val_loss: 0.0045 - val_acc: 1.0000\n",
            "Epoch 14/30\n",
            "1400/1400 [==============================] - 0s 264us/step - loss: 0.0132 - acc: 0.9957 - val_loss: 0.0433 - val_acc: 0.9750\n",
            "Epoch 15/30\n",
            "1400/1400 [==============================] - 0s 259us/step - loss: 0.0085 - acc: 0.9964 - val_loss: 0.0126 - val_acc: 0.9950\n",
            "Epoch 16/30\n",
            "1400/1400 [==============================] - 0s 256us/step - loss: 0.0099 - acc: 0.9929 - val_loss: 0.0028 - val_acc: 1.0000\n",
            "Epoch 17/30\n",
            "1400/1400 [==============================] - 0s 249us/step - loss: 0.0142 - acc: 0.9950 - val_loss: 0.0037 - val_acc: 1.0000\n",
            "Epoch 18/30\n",
            "1400/1400 [==============================] - 0s 253us/step - loss: 0.0114 - acc: 0.9950 - val_loss: 0.0839 - val_acc: 0.9750\n",
            "Epoch 19/30\n",
            "1400/1400 [==============================] - 0s 262us/step - loss: 0.0127 - acc: 0.9950 - val_loss: 0.0073 - val_acc: 0.9950\n",
            "Epoch 20/30\n",
            "1400/1400 [==============================] - 0s 271us/step - loss: 0.0088 - acc: 0.9964 - val_loss: 0.0116 - val_acc: 0.9950\n",
            "Epoch 21/30\n",
            "1400/1400 [==============================] - 0s 263us/step - loss: 0.0127 - acc: 0.9957 - val_loss: 0.0204 - val_acc: 0.9950\n",
            "Epoch 22/30\n",
            "1400/1400 [==============================] - 0s 278us/step - loss: 0.0130 - acc: 0.9950 - val_loss: 0.0510 - val_acc: 0.9800\n",
            "Epoch 23/30\n",
            "1400/1400 [==============================] - 0s 247us/step - loss: 0.0126 - acc: 0.9943 - val_loss: 0.0012 - val_acc: 1.0000\n",
            "Epoch 24/30\n",
            "1400/1400 [==============================] - 0s 258us/step - loss: 0.0125 - acc: 0.9943 - val_loss: 0.0070 - val_acc: 0.9950\n",
            "Epoch 25/30\n",
            "1400/1400 [==============================] - 0s 287us/step - loss: 0.0136 - acc: 0.9957 - val_loss: 0.0322 - val_acc: 0.9900\n",
            "Epoch 26/30\n",
            "1400/1400 [==============================] - 0s 279us/step - loss: 0.0128 - acc: 0.9957 - val_loss: 0.0062 - val_acc: 0.9950\n",
            "Epoch 27/30\n",
            "1400/1400 [==============================] - 0s 268us/step - loss: 0.0126 - acc: 0.9957 - val_loss: 0.0254 - val_acc: 0.9950\n",
            "Epoch 28/30\n",
            "1400/1400 [==============================] - 0s 276us/step - loss: 0.0072 - acc: 0.9957 - val_loss: 0.0519 - val_acc: 0.9800\n",
            "Epoch 29/30\n",
            "1400/1400 [==============================] - 0s 276us/step - loss: 0.0121 - acc: 0.9950 - val_loss: 0.0123 - val_acc: 0.9950\n",
            "Epoch 30/30\n",
            "1400/1400 [==============================] - 0s 275us/step - loss: 0.0113 - acc: 0.9950 - val_loss: 0.0178 - val_acc: 0.9950\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "U0Ciu13S6R0G",
        "colab_type": "code",
        "outputId": "a9946f73-ddcc-45c8-a364-2f62c307fbda",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 308
        }
      },
      "source": [
        "f, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 4))\n",
        "t = f.suptitle('Performance', fontsize=12)\n",
        "f.subplots_adjust(top=0.85, wspace=0.3)\n",
        "\n",
        "epoch_list = list(range(1,31))\n",
        "ax1.plot(epoch_list, history.history['acc'], label='Train Accuracy')\n",
        "ax1.plot(epoch_list, history.history['val_acc'], label='Validation Accuracy')\n",
        "ax1.set_xticks(np.arange(0, 31, 5))\n",
        "ax1.set_ylabel('Accuracy Value')\n",
        "ax1.set_xlabel('Epoch')\n",
        "ax1.set_title('Accuracy')\n",
        "l1 = ax1.legend(loc=\"best\")\n",
        "\n",
        "ax2.plot(epoch_list, history.history['loss'], label='Train Loss')\n",
        "ax2.plot(epoch_list, history.history['val_loss'], label='Validation Loss')\n",
        "ax2.set_xticks(np.arange(0, 31, 5))\n",
        "ax2.set_ylabel('Loss Value')\n",
        "ax2.set_xlabel('Epoch')\n",
        "ax2.set_title('Loss')\n",
        "l2 = ax2.legend(loc=\"best\")"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtoAAAEjCAYAAAAMm6Z/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 864x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "voT7N-YHZyVf",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "model.save('first_model.h5')"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "uPmvzJL7Z8CM",
        "colab_type": "text"
      },
      "source": [
        "Evaluating"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wvHDGGP4ZySF",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# load dependencies\n",
        "import glob\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.preprocessing.image import load_img, img_to_array, array_to_img\n",
        "from keras.models import load_model\n",
        "import model_evaluation_utils as meu\n",
        "%matplotlib inline\n",
        "\n",
        "# load saved models\n",
        "tl_cnn = load_model('first_model.h5')\n",
        "\n",
        "# load other configurations\n",
        "IMG_DIM = (100, 100)\n",
        "input_shape = (100, 100, 3)\n",
        "num2class_label_transformer = lambda l: ['false' if x == 0 else 'true' for x in l]\n",
        "class2num_label_transformer = lambda l: [0 if x == 'false' else 1 for x in l]\n",
        "\n",
        "# load VGG model for bottleneck features\n",
        "from keras.applications import vgg16\n",
        "from keras.models import Model\n",
        "import keras\n",
        "\n",
        "vgg = vgg16.VGG16(include_top=False, weights='imagenet', \n",
        "                  input_shape=input_shape)\n",
        "output = vgg.layers[-1].output\n",
        "output = keras.layers.Flatten()(output)\n",
        "vgg_model = Model(vgg.input, output)\n",
        "vgg_model.trainable = False\n",
        "\n",
        "def get_bottleneck_features(model, input_imgs):\n",
        "    features = model.predict(input_imgs, verbose=0)\n",
        "    return features"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "T2HmBZ9tYYnp",
        "colab_type": "code",
        "outputId": "f4761e0e-9bf2-4dc8-9d57-caf7e79c61e0",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 50
        }
      },
      "source": [
        "IMG_DIM = (100, 100)\n",
        "\n",
        "test_files = glob.glob('test_data/*')\n",
        "test_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in test_files]\n",
        "test_imgs = np.array(test_imgs)\n",
        "test_labels = [fn.split('/')[1].split(' ')[0].strip() for fn in test_files]\n",
        "\n",
        "test_imgs_scaled = test_imgs.astype('float32')\n",
        "test_imgs_scaled /= 255\n",
        "\n",
        "class2num_label_transformer = lambda l: [0 if x == 'false' else 1 for x in l]\n",
        "test_labels_enc = class2num_label_transformer(test_labels)\n",
        "#print(test_labels[0:5], test_labels_enc[0:5])\n",
        "\n",
        "#le = LabelEncoder()\n",
        "#le.fit(train_labels)\n",
        "#test_labels_enc = le.transform(test_labels)\n",
        "\n",
        "\n",
        "print('Test dataset shape:', test_imgs.shape)\n",
        "print(test_labels[0:5], test_labels_enc[0:5])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Test dataset shape: (200, 100, 100, 3)\n",
            "['true', 'false', 'false', 'true', 'false'] [1, 0, 0, 1, 0]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZAXhlEQ3aC7C",
        "colab_type": "code",
        "outputId": "3fe93d13-3e2f-4aaf-c695-e38f19fbafe1",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 504
        }
      },
      "source": [
        "import model_evaluation_utils as meu\n",
        "\n",
        "test_bottleneck_features = get_bottleneck_features(vgg_model, test_imgs_scaled)\n",
        "\n",
        "from keras.models import load_model\n",
        "tl_cnn = load_model('first_model.h5')\n",
        "\n",
        "num2class_label_transformer = lambda l: ['false' if x == 0 else 'true' for x in l]\n",
        "\n",
        "predictions = tl_cnn.predict_classes(test_bottleneck_features, verbose=0)\n",
        "predictions = num2class_label_transformer(predictions)\n",
        "meu.display_model_performance_metrics(true_labels=test_labels, predicted_labels=predictions, \n",
        "                                      classes=list(set(test_labels)))"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model Performance metrics:\n",
            "------------------------------\n",
            "Accuracy: 0.995\n",
            "Precision: 0.995\n",
            "Recall: 0.995\n",
            "F1 Score: 0.995\n",
            "\n",
            "Model Classification report:\n",
            "------------------------------\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "        true       0.99      1.00      1.00       100\n",
            "       false       1.00      0.99      0.99       100\n",
            "\n",
            "    accuracy                           0.99       200\n",
            "   macro avg       1.00      0.99      0.99       200\n",
            "weighted avg       1.00      0.99      0.99       200\n",
            "\n",
            "\n",
            "Prediction Confusion Matrix:\n",
            "------------------------------\n",
            "              Predicted:      \n",
            "                    true false\n",
            "Actual: true   100        0   \n",
            "        false  1          99  \n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "stream",
          "text": [
            "/content/model_evaluation_utils.py:62: FutureWarning: the 'labels' keyword is deprecated, use 'codes' instead\n",
            "  labels=level_labels),\n",
            "/content/model_evaluation_utils.py:64: FutureWarning: the 'labels' keyword is deprecated, use 'codes' instead\n",
            "  labels=level_labels))\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jvn_H5vd50x5",
        "colab_type": "code",
        "outputId": "eb788d9b-991b-4b36-c745-8804f37e5c99",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 340
        }
      },
      "source": [
        "meu.plot_model_roc_curve(tl_cnn, test_imgs_scaled, \n",
        "                         true_labels=[test_labels_enc], class_names=[0, 1])"
      ],
      "execution_count": 0,
      "outputs": [
        {
          "output_type": "error",
          "ename": "ValueError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-25-b9d11dd9b204>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m meu.plot_model_roc_curve(tl_cnn, test_imgs_scaled, \n\u001b[0;32m----> 2\u001b[0;31m                          true_labels=[0,1], class_names=[0, 1])\n\u001b[0m",
            "\u001b[0;32m/content/model_evaluation_utils.py\u001b[0m in \u001b[0;36mplot_model_roc_curve\u001b[0;34m(clf, features, true_labels, label_encoder, class_names)\u001b[0m\n\u001b[1;32m    143\u001b[0m     \u001b[0;32mif\u001b[0m \u001b[0mn_classes\u001b[0m \u001b[0;34m==\u001b[0m \u001b[0;36m2\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    144\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mclf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'predict_proba'\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[0;32m--> 145\u001b[0;31m             \u001b[0mprob\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mclf\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict_proba\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mfeatures\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    146\u001b[0m             \u001b[0my_score\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mprob\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mprob\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m-\u001b[0m\u001b[0;36m1\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    147\u001b[0m         \u001b[0;32melif\u001b[0m \u001b[0mhasattr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mclf\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'decision_function'\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[0;32m/usr/local/lib/python3.6/dist-packages/keras/engine/sequential.py\u001b[0m in \u001b[0;36mpredict_proba\u001b[0;34m(self, x, batch_size, verbose)\u001b[0m\n\u001b[1;32m    244\u001b[0m             \u001b[0mA\u001b[0m \u001b[0mNumpy\u001b[0m \u001b[0marray\u001b[0m \u001b[0mof\u001b[0m \u001b[0mprobability\u001b[0m \u001b[0mpredictions\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    245\u001b[0m         \"\"\"\n\u001b[0;32m--> 246\u001b[0;31m         \u001b[0mpreds\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mpredict\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mbatch_size\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mverbose\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    247\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mpreds\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmin\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m<\u001b[0m \u001b[0;36m0.\u001b[0m \u001b[0;32mor\u001b[0m \u001b[0mpreds\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mmax\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m1.\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    248\u001b[0m             warnings.warn('Network returning invalid probability values. '\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36mpredict\u001b[0;34m(self, x, batch_size, verbose, steps, callbacks, max_queue_size, workers, use_multiprocessing)\u001b[0m\n\u001b[1;32m   1378\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1379\u001b[0m         \u001b[0;31m# Case 2: Symbolic tensors or Numpy array-like.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m-> 1380\u001b[0;31m         \u001b[0mx\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0m_\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_standardize_user_data\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mx\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   1381\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mstateful\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m   1382\u001b[0m             \u001b[0;32mif\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0mbatch_size\u001b[0m \u001b[0;32mand\u001b[0m \u001b[0mx\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m0\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m%\u001b[0m \u001b[0mbatch_size\u001b[0m \u001b[0;34m!=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/keras/engine/training.py\u001b[0m in \u001b[0;36m_standardize_user_data\u001b[0;34m(self, x, y, sample_weight, class_weight, check_array_lengths, batch_size)\u001b[0m\n\u001b[1;32m    755\u001b[0m             \u001b[0mfeed_input_shapes\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    756\u001b[0m             \u001b[0mcheck_batch_axis\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;32mFalse\u001b[0m\u001b[0;34m,\u001b[0m  \u001b[0;31m# Don't enforce the batch size.\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 757\u001b[0;31m             exception_prefix='input')\n\u001b[0m\u001b[1;32m    758\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    759\u001b[0m         \u001b[0;32mif\u001b[0m \u001b[0my\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m/usr/local/lib/python3.6/dist-packages/keras/engine/training_utils.py\u001b[0m in \u001b[0;36mstandardize_input_data\u001b[0;34m(data, names, shapes, check_batch_axis, exception_prefix)\u001b[0m\n\u001b[1;32m    129\u001b[0m                         \u001b[0;34m': expected '\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0mnames\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0mi\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m' to have '\u001b[0m \u001b[0;34m+\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    130\u001b[0m                         \u001b[0mstr\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mlen\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mshape\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m)\u001b[0m \u001b[0;34m+\u001b[0m \u001b[0;34m' dimensions, but got array '\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 131\u001b[0;31m                         'with shape ' + str(data_shape))\n\u001b[0m\u001b[1;32m    132\u001b[0m                 \u001b[0;32mif\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mcheck_batch_axis\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    133\u001b[0m                     \u001b[0mdata_shape\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mdata_shape\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;36m1\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[0;31mValueError\u001b[0m: Error when checking input: expected dense_1_input to have 2 dimensions, but got array with shape (200, 100, 100, 3)"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DJdCVzenAMI-",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        ""
      ],
      "execution_count": 0,
      "outputs": []
    }
  ]
}