train-cat-dog-classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. Ref: https://www.kaggle.com/bavalpreet26/cnn-tutorial-keras-nb2\n",
    "2. dataset: https://www.kaggle.com/tongpython/cat-and-dog/notebooks?datasetId=23777&sortBy=voteCount"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras import layers\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = ['cats', 'dogs']\n",
    "train_path = 'dataset/training_set/'\n",
    "test_path = 'dataset/test_set/'\n",
    "lr = 0.0001\n",
    "epochs = 50\n",
    "algorithm = 'Vgg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 8005 images belonging to 2 classes.\n",
      "Found 2023 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "train_batches = ImageDataGenerator(\n",
    "    preprocessing_function=tf.keras.applications.vgg16.preprocess_input) \\\n",
    "    .flow_from_directory(\n",
    "    directory=train_path,\n",
    "    target_size=(224,224),\n",
    "    classes=['cats', 'dogs'],\n",
    "    batch_size=10, shuffle=True)\n",
    "test_batches = ImageDataGenerator(preprocessing_function=tf.keras.applications.vgg16.preprocess_input) \\\n",
    "    .flow_from_directory(\n",
    "    directory=test_path,\n",
    "    target_size=(224,224),\n",
    "    classes=['cats', 'dogs'],\n",
    "    batch_size=10, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAWAAAAB1CAYAAABwDTvQAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAABJlElEQVR4nO29d3gc1fX//7qzfVe76tWyJPfeG8Y2xsaYbnoPxRDTA6EGCJBAElJJqIEAwVTTezfVNu4FF9zlJllW76vtM+f3x8pVsrEkE/0+X+b1PPPYmrlz5+zMnfecOffcO0pEMDExMTH536N1tgEmJiYmP1dMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkrG0prJRqJWctAeXLwJpsw67pGA312Lt2xYga+LduRILBA8rbsLi6QIIgkQBGVCCoA01gUaADEt5vDxFRbftZB7P1f0KViKS3dSe7zSo2QInQpBsttlssFgzDYN+0QaU0RAxQgICmWRCR+LrDpD3nFsDmSJRYtAeIArYCDaCsOB12QqEg8QsJ4MJmy8HhKMfv9wMWIA2owOfzEgxkEI0lAk3ARkBITPJRX9ew3/GcTieaUjQFAm22Ny0tTQoKCqisrCTmL6aiVogJaAosVhvRaLSVvRRY80num4qrvpJkn4+62hpKdu0CICUllW7dCogZOoU7ishMdlERNOienYPSddbv2EWktqJdbUEpJZoGackuEhMz8SWnAlBTU822bdvbWt3Bj9P87+4W1d62cLB7TSmFLymJpsZGcrrk4PP5sFtt1NXUUlVXQ3JSEomJSQBEw2GqKyux2Z2kZaZjte6VJiMUpKkphN1tx+50o5Ri67at1FTX/OS6kEgm+Sm52AoAA6KNQn1VlORMC5pLQ1kA1N6T2YxEDPQ6weq1QCMsL17ealtokwCjbEAayt4Fu6WWsKbBGY8gIxJwrnifnnojO5r6UHPJebC2hNSln2NZ+h8qyuONFnsmXe78G6NOn0ZOZZArj8vihgDs3CAMKKqmsZeVbQuClD1wIUbZ3DaZ1hEUYAMiR6a6He3ZKT0tnbKyMlqXzkx03QLEADcQwOGwEItFsRpuolophh7FMAzilzQRSAUSgDqgEajer0arZiNmxNpjKgAWlU5M5gIu4ALgLRRRsrJy2L59+z4l++LxPEqPHneyfPl8oAvwb+BSBmSnsam0K9UNmcAPe/YY0TWBb+oa2PdOyUpPJBhq3xXKL8hlyfxPWDR/OR/dfxbPLwxTbWikWjROuOhiXnrpJXRdP2AvK6TN4ITPbuesmlpO6ZOMv7qa6Vddw5q1RVx44an89o/38/fiZZQ9vowJ4218VZrGY1edTqyhgeG33smO/z7VrrYAIAJOj4+PP3ufXr0GABaqqko5/vjjWblybXurPfAXYrD3UXmk0CwWunbPZ9eOXfgbG9FjMYqLisnMTuOkKZO4/MoZVDWFePKBP1NVXES+x8dOpbGjrpYIiu5du3PR5FPZFQ1Q3lhGdkIW2blJuB0KPdNFsieT++++/whb3To6YUbFLueBu64k+xQXESOK3hTGleZGaRqIEAsEqS0rJ617AUopiEQwCoNUPK8oLqwmdXtXehTbWm0LbRNghwf0vojTQ7hxCygdFmyF2kux9slnx/y70Kb2xzV3PcGNRVSvdENlXFLS+xxNQ8pNWN09GRq1ctUxGWRZFX/2gT5akTIqndWrqlh1ahq7oo/y2gMXQfW6jp/Bw0CA1nyg/yVl5eVYiD8M9r8hhgBvAj72FeBwOADUYhBD9DloWgVdMwfSq28GjU2ZFG3Np6nJS1TCRCLr0I1TgHqsVojFQO/gABxdjzZbqgE3AJ8QP5MHRrU2Y0RuJOYPNP9dBJwHBKiqCxKObmnxi/WqyhbHi4T9hMItVh8W0XAMZU0hqrnYknYSFt8X9EjNZoQvkc2bF3H22aexadNWIuEo9XV1NPj9hEIRolGdzZ8WMeyMTFw2K86sTB54ZiZFSxeRlm7DBgzw5GBL7cNpx4xgxacRrCgalItooFf7jG1GBPz+WhZ/9zk9ew5EKUhNzeKlV1/kkUefYPmipaxZs55YrH0PUaUgJvBTvSYOGzqSvC7VjBozjicf+wfRQAD7rkpWf/Yud3z4KQHDQlEwSH0kzGridghgV3DcSVM4c8YlrN2+mUf+8SxrtyxjQrA3mSP6MWLcUN55fBY7i4p+Isv3x5WsSD1rDYnHRMDpwqFs4LHt2e4vL+Pbp56loi7CJQ/cic3nAaWodYT4yljJwLO7ke5UcM5BDhB/ZT28hb3naZ/FJTBN0M4TLIlCUndh9KXC2X8STn9U6PEbQWmSMPpaUWPvEJU9Rk7/oU7WGobU6LosMwxZZ4iU6CKlhiEXv7BRevx1pvT6YLU4E5IlbuLh23hoW/8ny7L22JsB4mtRl1vgOgFD4rfkvktUHI6ATBgdk5uuL5FXXonKriJDolFDgkFDyopismmVyKqVIi+/0CBDh09t1d722CoiaCpNYKtAqcA3Ajmi1CApKOjX4hg2m0265ua0WJ+sIQ7V0qb+idYW67wej3g9nnbZO3DAADEMQwq3bJFPP/lInnnySTn//HMlJ8Ur//zNQGls2CmBQJ2UlxVL8fZVsmzRHJn56OPi6PKYjHitVOoDITEMkbp6Q3bW+eWuu26T1asXi4hIUXGDTL7071K2bZ1c+16DhAyRwopaybjkX+1uC/v+7qNG9ZU1a74Tw4iIiIghhui6Lo2NdfLf/z4mWVkprV7XJIVYWjm3gCiFJCQcubZwsHshwemUX136S8nskisKpKsT6ZWAJColHhDrAeU1kGMHdJfe3XvK7NlrpC4YlbqmiKwL1MsK3ZBgqFGaImHRdUMGDR70k+mCUkpyu3SRX5xzjsz74gvRYzExDEMMXZeGwkIxwmERwxARkZKN6+WOc86RV+95SOpKSmU3RswQo0EX/1cRWX7nloO2hbZ5wK0SBD5gz7tzXT0s2QpLAJyQNBhO/B3+EX1h3XYoG8Gnm2rpa9Ri+eF77j33dKqbAnyydgeFo/rw2WsPElj9NWNu+x1bw6GOm9dhmgOsPzFVCiwaBziDYaBni7KaBhMmaNx6q5PjpygczhzUPjEoqxWcXS1kdo3/PWhwAuU1J7FyxewjZq9FajGY1GxjAGgAyjEiWa2Wb+0Muh0OYtEY4dj+HnBpK16dWECsnvYZK0J9UxMqFmHS8SfjsMJJ007lqYeyiCgdpSmcTgd2mxvIoEuelfyC/mS+/DC9k+pZtKyQiUeNw+OGZfPX8cG3S+gzZiyDBo1GOVwU9OvG8srthPypaMpLRBQ24yzg5vbZuw+btmzju+/eZcCA0QAoFEpTJCQkMn36dei6nZtu+jXBA/paGoSDhLPiZGR4SU6OUVx8YB9NO0nJhtoKkL3X0h8K8e8Xn91jh9OwU9ykowNWdA68ygawsHAn0UiUr2a/StcR96G2WQgkNGBzN/Duty/x1mtfMjCvD0F/3ZGxu7Wf4vXyu7vu4hfTr8DhcsbDCkAkEMT//SaK122m38nHoywWsnr25oEXX8QSEzSXY08dyqKILKnm/RvmUGccPMhzBAS4JcrmQE44HwZehnbmEBicgqEUTosQKoRcA0qf+zPfz5nNHdNOoqKolL9fNQO55R7S09PYUFLMvDuuRW+1c+QI29q8HH631U+DIdDyOimgB/tG+B1WuO12uO0OjcRE9hPeg6GUor624ccLtgHN6cOl7yK4zzWyiEFZ6a4WZUUMbFYfSpXu14loT0vF3uiHfTrcFBqZyV7qAkFE9sZ8nRZFNNbULls3bi5kxOgJ9O8xgldffRy7x0l6WiaLVpdQUbGdlWX3kJbah65Zdk448Xjy+vdn045t9BjoJLF4De/PW8qoAWNwe6zU7lRUb3Lh3xWPqSuHxuShPbBjJdroQHQDh89HanITJe2ydi+ZGV6mX3EWp556FUrZWmxXSuPyy6dT31DPPff8lnBo77UwgD5uK/kj8rG53JTUh7B6g9RXBago9rN1a2MHrTuAYD200vlrALkuKAlC2Oskqz6Mz6VTGrERjkTx2qB+n9B+OBz/45XXXuSCi89lyPAh1EVS2bRpCxVRWLRsIe988jVau7oLDw+LHuOhv/6ZxtIwv7z9arw+D7qhY3O7yDzrBGTHjj03nqZpOFwuIB5NMHQdDANBmDXnCRYmFxHdfAgH7oi775pVcm7+rVgCYUk3RC4UkeTVIrwrYgsYknxvQC7ZEJVv6uvlt9/Ok8diujy8o1ruf+xZubKsTqa9/aVYrY4Ovxodjq0/0dLO104lYBFA3G6P2DRNkhLyBdbsCTukJIj87R6RUEDahK7HZPr0ywRFfNljq2rXuRUR7FabqFZ+v6Y0cTvd+7cJEJtFa1HWYVVi0dSev7OSe8tfbnxO+uR3b1HW47aKx6q1sy24BVIkr/s0qazySzQmUuUPyKSTzpVpF06XSZf/UYZP+a1oaRPl0ccekw++niMfzlko6zaVytu/aZJ/9KyRzRsNMQyRcNCQR46tkY8vLBa9VmRLpciC9Y1SXRuSSec0SVV1VAK6yIhLCjsUghgzvIu8/ugZsnjBLKmvrTrk9Q2Hg/L8i09Ljx4FLc673WaRFJ9DUjITJT3XI3ndEsTtaxniae99ttvegy0OkAKbU5I1q6RakKFuTV4bMVAGWTIFEKtCLJbW9z1m7BjZunGD6LouxeuWyeeznpKvPn1dbrxphrg9rna2hUPfv26HVf52xVmyZd5SWfHZVvnuv9ukamuTxCIRMXT9kNehqrxcrrvqKpk7e7bUfvm5XDCol4zv002OHtD/oG3hiIta9mm/kNENTZIoIr8XkXtFBL8IC0Scnxgy8DdBOfPZQvm3LjKoXhfbZ02StLha7ogZ8rAh8laVX/KGTvzfC7BC0DI6UYDjgmhRSrpkZsmvzz9f5r+9WcYPiIrNIjLlKJGVC0VikUO2gRYYhiFlW3ZK724FggVB2/cGtbbr3IoI9oPcNAmuBLn7ktukq7f12KTFoomm7S/Gds0u/bL6yHtPL5BYWJf7Z9za8kbWlLhVO9uC6iZcPks8ty6QOavDIobIluKQnHjTe3LGuX+RPkPOEUv+VULaCLHbnWKxJYvH3VMmZ38u3S2GpFpEHrpOZMv3Ip89KHKUU2SEJvLFDVH5bp5fnnnze1m0YZGMPW61VDYGpbTRkOzR/2p3W1AKmXFBH3no79dIVkaqzJk967Cu87p1a+X4qZMlOSVJbLa9IpucnCz9B/SUBJe91YfmTynAXk2TMampe45rV0q6uFySlZohE4+aKOMHDJVTRvVr1S6bFTlm7FEy9+uFUrV4jrzypyvl00+el4qyrTJs2ND2tYWD2GmzaDIuN10enn6FbJmzXEKVETEMkYhfl83flom/2t/ifAcCASnftUv0aExEROoqK2XZt99KzeoNUn3bqxIo2Sbh+ioJ1NcdtC0c4RCEhdrep+DHxSBgPHCnIbiWBQm+HSJ0YgprC4RJuo8tq3ax9vqnsO+IUi/fs3HmC1w3JZPaOjc9u11A0cpVxHND/0f5CTYXWBMhUEm+08HOUAgdyMjIoKqqqjnF66dDUxrjepzEVVdNZNJ555GZlYXFZuezyYrXX4XjjoX8vrTIN2xBvJHtjU2I8PasWSQGwqQ6bFQH9p5PaRGFO3zkIOv9QT8rl3zBcQNH8uby+TRF9g8beG0aUV1oMsCqaXTxJlEdijGs/yWs2dUV29ebeePLj1vUGzYER4u1h0lCEtSfTtObBi/X2jj6P+DWgmz+IIftRQOxZiejrD9AYyWRSCpoPpKj19AYPg6HUhCD4o/ggwYoeR0ao1AC/O0pxQXRNXwZe5birX2p9aSxsipM7dpUqta3P97u8ThJSc3jvRc/pqKqhqraH0//UErRt28/Zr32Oo319SxetIBnn/4vOdnZrFmznDXrtiCGcdDr9lMRMgyqAg3sTqI0lMLnTOS2e//EeZdfwDfvfc79f7y1VbuMGHy/YgVXTb+AUydPYshRg3nl6df43bZ72LnTf8RsdFktHDM0l2tPv4+JF5yNr5sPrTnGYfNo9BibgbLsf+MZ4TBvPPVfIhUVXHL3b3B6PfhSUxk2fjyhzUVw2lG4svPRdZ03X3jh4Ac/Uk+6+Ct0ltDvK+m33ZC7RGSOITJ1myEXfdwo9qfrhDoR63pDPq3W5eRf3CooTVB2AU3G3/d3aTQMue1uEU17WyBVwCtgO6JPuoMuVquguZo9w73rfT6fNCdv/6QecLY3UfxllWI0964e+LT9YfVqefKJJ2T+119LLBpttUxZ6S75/J13ZOXipRKLxfasj4bDUrNutbzz74dl4uAhoh0Br8d6wHnad0l0e+S0oyaLW9vfC3Y6E6Rrtz77revXrZfcfPF0KcgYII889pbcceFvJdHuat17bqe9pI0Q3IaAIW63yH13iqxdFZOBfcpE03QZe/RSOXXa1dJnyDRxu6fLUPsnMpGw3GQRGaKJTEbkrmSRv+WI3GMVGYBIMiL5iFztXClJzjTJzeshdvcgye9zteT2vEVQ2e1uCxYNsVstkuK0CyD/+MfvW20Xh8IwDAmHQmIYuuzcuU3Gjhryo223vW3hUHUq2K+9AWK1WOW4CZNl7KjJkpycfNB2lOy2i01DemVlywN33yQ5qUlSkOOS887tKrT3beiAYzitFrl48lGy8LOPpXKtX4ywIeF6EWO2SGy1SLRUJFojIuGW5zcWjYke3p0hYUhDaa2s+Owr+eKJf0jdlg1Su/EHefbJJ8WbkHDQtnAETrQSNIeQPEh6nvOh3L86KusNkYAhUtwg8l29SL0hMt0QUSKiDJHHdF0uumh6/BWluZ6xJ1wkZXW6PPKoCCwW+JWgLhfIOCInurWGUQCSZ7Ec8rWsHUu7brre2QUSixqy4Yc1Em6q3+9ih8NhmTRpkjidTjlp4kTZ/P33LW640hVLZMZ5F0iyJ1Hyc3KltLik1ZuyescOOe24E37ymy4/M1eSHfvHgjXNIR5P5n7r7DabnDhytFzQa5ycN2KKdEnPPmi9WnsFOH2E4Ii3dhBRSiQtNSpKxSQreadsXlsrgVBI5i9aJ6fl7JBznbocrUTuQyQPkQxEjkJkkhKZhMjRiPRTIumIDCAsk7xXytnTLxVf1/5y2+P/lG7jTxfo0+62kODYP0Z75pnjJBYLtbierRGLRcTQY/tdc8OIyVefvS8Ou/0nF+ApJ50iv7ntNunes6eofUJNqb5EyU5KlkSvV2yaJgkKcR3CFrvdLvffeYdMGTRcrrrkQnHb94ZP+vXMkrSU5Pa1hX2O4bFa5Mm7bpPVC76XXYW1YuiGxLaJVK0RiS0W0UtE9GoRIyoirTz/DMOQmq2VUramVObPnCcvXPyS/CHtGunlTJIBGSnSN8UnLovlkLrQsRCELQnPxbcy7MQTOH1wN87tmUK0pARfuRd7aiLZbkUXTWgIRRlqs2CxWuiqYCKKntdeyedffkpNRTkoG4vWjuFvLyquOBayumRSHr4Gye4LG0Z2yMTWUCgsKHZiEGsxAqpz0Fx2NAtkF+Sh2fd/2bbZbPzlT39izdKljBk7lh6DB7fYP6lXX04770wGDyzAl5VPclpKizIKSLTUcsXk8/j8m6+IdGAk3KEQoLiiBE3bf1CGUjr9BvdlxeJKxg4fhDcxgy++/YbPli3BplTzwAA5aL3tDgK52W8goAhUVcebvnJWMn/BUoZEhpHjtBLsksSmUkU3pbPA4qd/zEcQRQ0QEnAC2cB1VngvBl+JjePDdzHrzWtx5CrychMINJYDW9prLTarkOFUVNTHz8WqVZuoqSkhPb1b/BFCGIVjT3rU3t8lvPHqS1gaiznnqrvRLDaKNq9l/erlZGfn4LIo2jmW5bA4cdo0zrj0lxixAE+ccAJ/uO8uFixcBkCjvxG7pmFoGicN6kl+rwI2lpXx3bL1BEJRklxOBuYnEWwKs7y4lkgkwj8ffpQYisLaKsb1yyezb19eeeNDyioCWGztly6fTXHcMeM568xpjDn6BKyOVLK6J6JiCks+pB4szCcghhCNgL8uRmNlACJRdGsId30uJ1YN5dUqg528QDB0eOl9bfwVu5O2LJCUheUvT3H9qUfRt7YOe3oKzsYIIZuVoM1CfWOQgFh5f7HBcy8Xsvmkrvgu9jFdUwxUip5jxjLzufe5446NHHVKAYU1Q3n/Q8VvLoSrr1E8XC34PRH0uiM3X5AFCzo6ghA7xI3eGYjFiI+d9/habFNKMXrsWEaPHdtcWBAxEEPAMFBWK84EL6edfS5y1rl79mkNLas/vQaX4LVqVB+hsdetoTQNm82G0ixEo/ED6XqMU08+gQH5XagoLWbEmBNISe7BV3M/pryi+CezxZoI2cNh5wfsm6YKgM9bQO/+TtLTbVgdMbpPeokdm3LZZt/AsKP7cNRnJ/NS2M7uQd4KCAFpsRhd7cUMChewNNKdHlxNUfp/+XbTEhpDddCB+LrbJXRPhupG0A0oKqri5Sd/w2WX30XI6kXCu0jNGI7T491vPxHhsy++5NOPPiKj53COnXIqyz5/hyWLPua+Bx5m6ogevPHdTzO61JOUTHZ+Li4jxAv/eZpVK7/H79+b6hYxDCLN/Sgfr9lM+q4Kxo0YwLghffEbUcb27kYPrZKXlhXRr4+FVJ/O/GUhRCC0s4iaWjddm8KIQG1D+1MqrQruP/dkrv3H40hCDrXbGsnsnRy/X3Zf4EMQKolS3WTgS7eR3TsBm90X32cw+PsFePqbhwhGDj+3us0CrNltdDvnSsrOu55Ikp21sxfw5EejCZZ+TGb5GsJGBJGL8PRJIZTlpKrRjZE3AKp0MjcZXNHXggKcVsXJJ44iLWU0f3kYjp8Af18ECxdCfdBNl2ka/lFOdn7SNgsPituNveBMguteoyOj3y1KdXgYb2vo0WjcNdsjnLuPsX+LqKos4/03XmfJgqU0NTQikQCnnncGp537CxJ8iXt3N2LQtBW8PYhPgAMohdKsFDVVUv8Te/66ruNITeGqG+/g7XdeRQUjFG9YR0ZqGv1OOYl/vfQqdT98Q8H5dzK4bBtfVhT/ZI9ErxWy/hglNc3KqmcV+16+LVu87Cpx0rdvgJpImKEjxvCq4wm69da598Zf8tVXNlQ47kBnEvd+RwPJVoMVxqPkar8m3cjDap3M9k3P4t/URLimtEP2Wm0OLj8xheJXy9hepROLCQ8/+SWa5PCvmW/RvSCfc08/m2t+fct+D1qlFD16dCfZ66Rnzz4ATJo8lYIueYgrg5umT6HGksCXc5Z0yL7WOObE43nsrw9ht1np3j2Ps6dMJRxt/SGkG0JZZR2ffTmfmf+8mmlX/B4NJ2H/t0ze8iTJeV/T2Agnngpbtsb38TcFSHJ7GDBqNH0H9uK7T79ql50eu43Tb7gD5cuidkMT6d2TidRpONP48Q5uBa48O7m7/5bmWEZQqPu0gf88/RrFsbZN/9FG99JgzNHHMfXOP6N/Y6CdeyHhQheNP+wktvguSooMqmovotqWS5E3iYqcBIyTNThbwRQL3btr7H4xVkphsShGj4R77hTmfCIEI/DMaxCxpXBGdgaPeiDnSDnAjnSCfe8Aa3a7q7BoinOPHYd2OKMf2kg4ENwTF6qqrKSycufejaKDxLc/9tgTXH3TLSz5+nu2zi3lna++47Krb+Krjz+IF40GqNqxiqo5M6GhBtDA8LNvkvycBUv+J6GXUF0tKY3b+PWMq9GT0kGEhQvnMXzgaNJsGVQ4u7Pjy5ksXbsUIf7scblS2fPAOEI4dKFfLMyAuyFv4N71Fg3+9geNYybacRcnkJeax7DefXn1tfu55/e/JrM4ibeCijrioYcKYAXCR9YQNxi1bE2ewKIub9CQW8uYv/gYcdEdDDllIo5YB89tTBg0+iyOG98LmzXe1oK6jT7Dp1BfH+CbOQv561//xrplC3bHNYH4PTVkwADyfF5SU1JBKZL6jmT4mZfhye5Ov0lXUFPU0eEhrfPt+x9x0smnc/u9tzHz2aeo8v94lkJTDJ57/C0CJSVYHS6ceiJdvFv48Bm47AqorIiX2302SyorUaEmZpx6MTlZme2yMyMjg5TuvSl/vYGUrl4sSRqO1DZUICANUPJpjK1PlLLyD9v58Ipl3HjV3Tz49S00GvVtsqfNgZS6xhrmfrSW0Ot3QuVG1iyZT6I7gXrtaLDPgMu9MMUBXg3ydPBq2IIKV6biVhs4m58auzXMYoO+fQyuuilI+oseavzQf7gi7NAYi9Ayktl2lEWhNAfG6q3QgZtDN4SY3UFXn48d9W070T9GbUMDTz7xBJvWr2fEoD6cO/0q9j6SFegxsMAxxxzD008+yZrKjei6jgWNfr16MGzIEESEb9/4L9vLirjolzeDLwsEwvPewj7qTJTLh4hQXX1kbT8YoUiMvz08ky7pKdSUlgLCN98sYPjIReQ5a0lWjbz15TzqamtI9Hg4ddo5GMrLq7OePKJ2WMJRcmgg3Ohg2MkWdqyJh9KUBj37KlJTFVq6HUEYPcSDwkOwKYW5b31JmjaeZbobUMQAl+an97k7qB/fFcafyQxp4Ou/vwvDRjFG8/PP639HuCHwIxYdGlFWbG4Lv7z+Vr5cchc7dlURDDZhc9kZMKAP8+cvZkd5BZdffS0vzHyOfoNH7PGEBw8bzvgJx2JvDmUFKurZtn0XPQf2wmrNJiElF7YdeREOBgPM+3Y28749/PQ7r9XKkEGjabLYSLLYsWRNwFZ3F+O7L6R+eBnWwGoWri8lGo1hACXlVVw+NotLrzmPpob2pac6Elx89e8mxtrysCdaW3q9u7vMDuX4lUP0C/CmJsBCF1uXz+WLmndpoh3Xve09iEo0W5JgGSJwtihrb8H5K2HkPGGqXyzzYmKtErFGRVTUEEvIEK9f5OaoyA7DkEXbG+WpkqhsNkQihkhxY0weXl0hwz8vkxlPGvKX/0bk+QpDnqk25DvdkL4jRnS4t9OWnSOkXyE4esvu0WYHLkm0TJfZveTmpEnfvBxxKyVnnHC83HryNHHYWx9NRAcnYBnpS5B1LzwmkaaG/btcdV2qf5gv1YWr5Yunn5QTu/UUDeT8gmzZ8N0H0rjhA/notUflksm9ZFvh23u7bQ1D9OoNYuhRqd48Vx7/7a1y7JDBHe/59iF4Dy8zZN8skwS7XUYPGyPnH3ucTBk6UHLSUuT+3/9eVqxYIXW1dXLxORcLh0j7a4+tvYYOl99uKZP7tgbk5m9qxZ7QJGCIwpABOWFZszCedaI3hqVk4afSULRRilatkG7J6TJeO0emaeUyFEMyMWScrUl6nPSeDFwakYe2h+XLkCFjfv2ipFxyuwyevUqyJk/d1/52tQWf2yq//+VI+c/9U2XEkLQ9v33woEGSlpa23/kYc9Qwqa4u39NMotGI1NdUN2c/GLJ92zo5+dhJcv+tv5Ut/yiTcwc+eMSzIDQrYrEr6dc7S4Yd3U8s1pbXr7VMoxSPRyb17y+DBw2Sp596QDYt/Eya6suksbpc9FhYqkq2yrUXnipJNk2yMryS0SNNUnIcYlWI1s40tOzEfvLhDVGJVkirmQ1iiIjevBgiEhORfbMgDBFjjiHFxzbI7d7nJUuNFQsH1YIf1YU2esBWcPXHkAxw5UGsC2I5EQaPh1N6gyuGq0qjqSx+RIdLke0HZzoMADIE7g44WVFg4VHisTRbSMPfJZURfRUyRJgS1RiUplivoDfQzulX9kOvDkN0JVAEVi/E6vbb7rK76ZpcwJry1jsoVFQIhAJ0ycli8vmnkea0krZsHiVRG2n2LKrK17A3ZtsxfmjwM+36O3nVm87IM87b+6qgFJt3lHLTdReiKS9VlckIA1i6K5kbr7qf6vL17AiE+df1F5BfMJY9j3al0FL6IGKgpJRvPniRb9e0nO6xrSR0i//iplWHKmUBNGxEiRJ3KhK9PsoqdzBu1AlsW1RBzz69uWbGlSiblcXz5rJ+81r2C9IeATQUqdkZVGyoIyHbxa05r+Gp2MQ2YkzIrcez8QSiA6Zg9bjJHDqC8Lr30D97k+tVNX3lbZRPp95xIzuVRnnkdZ5Z+jUJ70ZZWFNEyfVXEfBMwbNjJdu75+LXLwepo3k2qnZhc9jwJRVQG0th7fqv96xfvWbN/gUVLF22kvfefZPpV1yDUhasVhu+5Ph7Y1NjHdMv/QXffreCuUsW85T2PpX+dg9nOSi9+3UjpYeFgot74d6cQkVZBd3Tklm0tJCoxFtBqhUqmkPClualrqmJOevWIcDNN6wlOzGJsYN7cOLUKZx5zY0kZ+Uxetg4/vPqR/gyHPz11fuYv/kt8jdHefa59nXaWoNucr1WLPbmFQb7vWju+TcM2IFS4vONdQe9AXZ8Cd/c3ci/ttxMk2ygnIUduvPbnsshWeAZDokJYM1EDTwKmdoFfAKpGv5eClLA64JxAileWBmBv9hhvFIc3dfKABWffrsa+EOawkk8zubLVHQF3gZSgH4d+GGwO19DIZEalAqgVDJGrLxFOYs3k4rIwXtWr7vsCo45cQrfLV5Ar4F9eG3my+yqrANlpcEWT+xJAurpuAyHgGJ/E3WL5iPTzkJZmidhUYpRJ5zBRZeuZ/sWGwvnX0rQ68bmtDL6lM+Y9fTvOHdqD864618oS8uPMCilkdTzHM6b8Dnvrnmuw5MPaSEIt5x35wB0QN8z0b0OJGdnMWxYLhu2b6SkeCea1coJEydSHQ5SXlZF5BATMFmsLSekORzCBtjDkJnqIClJ6Ol7jHHRVURm/hs18hy0mA2jZhVG6Sa0npNx9YuS++F6bvngN8SefAq16zuKPPOZlJrIxS8X0qQLxryNnP/EL+nV241xiofw2Jt42ulD7MdB3kgIXQ0V37TL3tr6EF/M/Z4Z542gZ/c8ftiwtfWCzY3tjZf/wkmTJ5DdbTAiQlN9PaUVJWwr3MSSlRsQwB8I4N9n0vsjicXu4trf3o4vpysJvQPcfv6N3Hfv74kuKQTi190W2zuvoMH+rqFSENUNtlbX0CPcheMuvgy7Jxki1ZSWL8cAKkpquOWKO6F/MrnVWezY1fI+PhyiAmla3CVXBnFxpfmZvw4q5oInJz54UtJAHGCsheATULUSGlcIn/qbWCtzgcIOnLU4bRfgWBJk2sF9EdbRqaiJdmLjHPg2GoQMC7Z86KfBaUAPHVYIVCpY0wQLPZCoYBQwDlgDZBH3ct3Ev60QEvirAb8Kgbg7Jmj9cbKVUPwci45IBa1lk/qri/AfJDNCKcWGygp61NTTJb8P/7r3Ib7+6uv4a7hEiYTj8dTdjepIoIA3332XMRcPJGHQL1C44p6sxcKxJx7Pxb/4JcneBQzpnkRUi/DCzJV0tdk49RcXkZCc1lzLvk08HlyRpkbCMS/qCEyx2d0D5z4CT/4Ldv6wtzqn00ootH/vt8Vi2fPFiR9++IEffthfCIor9/fIFTBqeG+WrNi0TyX25kng2044EqHAYbC9dBeVRgobo5BerNPr4fvRHrNB7hjwpbP5mcvI8JXhq16MxPrDmJuwjRqKbH6UjJml2GhieK7GD9VOnNvf56G/BcjJT+eam37FrGAeuhc4Nh26pIPnFXgip132Gobw6ZItRJxejh2YROE2RSi8z/VSxF8njHhWTnFFNZ/Mfp1LL+/NV7NfJK+gN9fdfTObVm+lqTHw4737HcQf9lPvExrX7WRYfh6vvDiLd976Yr8yjc0m6+xtlbsRia83gNkL1vDo33/LH/72OOWl25j3TTzbIVRrMHBwV84/qy+33vghFpud9qAMiNZD+ePxiQ669QAtHNeaWBP4FGhJ0PQZ7LJA7Txw1gqNlcI3/npKCPMljwDbUErt1wnaHtomwFYvJNVBt6MhIQFjpB2jvx0tRQjZNHoPgQu0uOdqA5osYI/Ge5BxQ6IO2da4d98N6BaDeoF6W/zDOdVAJVCsgc8Z/3ZC+1FsJkKSxUJQ19n9wSGHpghLcnNC6O7OqIN3zIkIL7zwAi+//HK85EGyB47kZI8h4L+bd1I6/SEuOXsd0264B4c33qs9aOxoPnj5Ca684lI++WYHWR4vuZJO96MtjJ08qPlmC0BkJdicoLIxwsmUz/+WZ//6BM988wX6EXhUWBTceTn0PAmuvgLqZgM6LcQXDn7ODoYA1ZUHdBTq7U9aTnBY0DSNWFYS1VHhrQHnsTVqMOWr1Zz+4b9Jv2YDyphA935W9Mh8Qq8uw7+qN2n4wHYKJDso3Ho3mb+axGnjMklZHGTB3C9Z/OkLdL3vjyxL1Xi9GgoSYEe2QhIAT/uzbXYzb8FKBk/LxKFZCO3OK1bEJ3xO92C32YjsqmJjSYQ3v/ic8y66gfW73mbw6Ge5/pe3MP2aK+P7tHK5d2v4kciF0aJ2piYcTcIQN9dffRUfvP85urH/QXd/O6W142lAOlBD3Bvul+mietM8PvjgU1Zsis8JrhQM7DGAt39YzaCj+9G0tn2ZMjGB5Z9Cj4mQacD296C2EvRu4A1CQT40WSGKYCmrwV+xg7dKVzKf5WxkFbCVRipJ8jgZOXQI20rL2LL1IG8oh0HbBNieAoPPhzQvBOsxdnqwJOtoQcgdZGOUFcqjMFeBLuBV0Nsa/8zIqQLbLXHxzaA5tmtpfi1tNmQ48GVztL6qHLKyO/LwVjjd2QydeBqqcA12vYa6kp3M8znBdSbsnAeHlTKiAVb0DghAe9CBD1ds4vNVhVzy6Tecc8mVTDnvYqwJHgrGjueJd99gxew3EEpJyLQzZfIVeJKziD9UImDpidHk4quPvmLTom08/vzTFNZvx223cnleNsW1O/iqsf1CLMAblbC6BuQy4rGydRyxD+ttLW7fK2ZrhKM6DXWCNxrDk56Mdu7lfFEwjC9e/zPG3HVcddrbkLOcSNoOgoW7cBY0UbSwJ8liwWpUE4kJt6vb+JW/gcmTLqCk/B3WzllN8nmnkdj/Ij4QRZ+UuEhsrwHHWAh38G3fCjidDmYvj9AQie3xeBHi35QKhImlOyFN0fPsgazbUUxFWRmegq58+u1zaBYXnq4+AuU1rXppwpH7FlxKQiI+u4/bb/4177//OYbR8ngB4vfyxN7dWFFeTWP9XpfFAPzNY7yO65/LtEvPoWhbiH8//QGV/niMQAQe+f27OFOEB56azMtr2xcDVgZ8vxWWlsJVieBJgPklcGwjJLnAtgUCLogdKzTmNGAM+pp3fn83W2Lxty+r1crRvbrxw/ZiGgN+zpwwgbkuF0vXrm2XW9M2AQ43gbUPlDTCpC5YRtpI7GMlzymkxiDfB0UG9LLEY6JdgIzmeO8JMRiiga95mwKKm8XWS1yQLcBA4tsrox3zKhUGg6adwdiBUzhzxDjul82kzXwdbyxCo74cjM0/WkeiK4H6kIX4+/v/VoB3E9EN5pY28vGtd/OJaz3DTpqASh1Jv37J9Ov3IPGz1gRix9CbiIlBtCaJpvoAq16cw9/fnMfXm/qi8xBW9Qd8aiu3/Po48qtep+Ch9k1wDhAz4OZzoGwlcdGNxadjciZCVIdYByerOpLdcJrFis8WYVcA6psUabnJ9Ox7LJFjRtJU/W+MTTOxpC/HWBFD22XgHCWstATpHQvgdWajp51A0+A1fBeIcLLLyZjjetJ/WyHf5k0k/xgbG4HrNXhGBxLjKTW82DF5690thZtuPo61S4vZMGsRez6UrYgrfYoHw6Hjyc/FPrELssxKTVOEo3uexH3v3EaMbhgBC6LJkY2PtUKD38/5F13A3C/nHrL/1AYkOOz0GD2e4nlfUxMK7THLL/FvZZ87MRt3Um9m/eMsmvw1+wUMjYgQDMK9c7/DVtK+EEQVBvNFuE5T5Fkgugvyo0J9MIL1xAje2xJIG6jAopFj5PLy+4spaRZft9vJuIGD6NGnF4MmTqRnr37MfPppumdnM3n8eOYvWUIo0jadaGMMuB7KG+GkY3Gf5GRkD+jujAfYF+vCTqXoYo17svXE47u7gE3AXB0mWeOfltw9oHn3gD1pNkSA5cQ/p5KS1bF2I2iUJ2TRIFt5YsVXLF+7iyI9AFWlYGw7rDoaiMLAvrDh+079amdNXYjGpjCP/3cLdw64iZ6JPSAcgphG/fwa5hV5sGlL+fLj+SytC9Cw7Wyq/NVUNSQTMn4FpOLzzOYPV49g2/xNvLPhDe47KkRaK57K4RLVwN8PWAC2ROg9Ds4+C84YC2vK4JYroLYoPiCvswkGolQ6HFQlefE2hcgMG6R7HDi7alS5zmPBQy8zPrsRw+2ifocf34mwMmRlSGmUEd0El8vg4hP68Oc3NpEVeZMbT0slsL6RZYu+Y930KJrdDjps/g7YCpEGYOm3HbLZ5Uuhb68JbJj/GFYNdB2wWMAwsCS5kGgMQ0umy5DR9EgejWtcIRWB1Thr3NSVJWPLSeS4M0fxwedzCa7zQ9NPp8CFOzexeePGQ5YR4m+/W0p3kWREOLpXNz5du5GYYcSjKhYICjz2SRE9ztnM2OMv5sW5f2zeax8yFdp1+fBl+9wzgyjVQIYHVJ7QVB+lUH3Pjt4buP6ek6FPQnNvoVD48Zs8NGs2UYuF3OxsElxO7v3TnzCsVsZPmAACG9av56OPPuK8U05hc2EhRWVlbbKnbQJs86KGDcZ5rKKgL3RxQu9ojGUlYPNY6J4AG4l/RKeXCNao8LhN4wIF3hBU2sGt4tkdEeJebzXxYLgbqDfgkxBEGiHijQfE2+tHOF0Ors1MJH9RhAc/L6TI2IzmTkRcydBUhRUbiekpNNVFCEZraU3qJRiGNcsO+5guNII/wceNRJ2Fzb6amXMzWX3uJs4b8gJNhV0o12zMXZvDpthUhAZ0TgXyiZ/ZePBGEWFAjxX89YG+TB2/jT98qzP760ZuPzXuxbYXJzDjbrCdDaPy4LiekGgBTYE7FyzpoO2Kh9oP5hVZmkNVrZGengrEqDwwFtweWz02QroioCsitfX4XElkYqHK52FzVRba0N/Q49/X4BnZxEKjC46HFX1GjeWjkJWhhoZVSsn1lJM+bjAvRl24/zSDcT/UMse/ghPKdpLWvTtrGqD0wSAsccApCtTcDtm8fFUhp59/F1PGunG4rIStgN2Bsmu4+mZjH9abUJKNppxulMYSOG3AdF5a/R/8nxZi7d8La6oXLc+CWu7F4gph6DoS+mnmtJY2iPuGqkYc1UFiGOjNozMVkGDEHbIV28o5ZdpFnH38EJJ7+ihZvn8HrWwXnFt7EVAL22WrDQt9rYrGDOHZ6Dae8d/PVu19Tq8fgsjk/cqGNQc9Bg6mq1Vjw/Zi7rztNrLTM0hIT0fTLLzw/PO88cYbNPr9vPb++yTY2+6Vt0mAtYwMLFMz6Juj4wsYbA1BituClgS9EhRNOji0eDgwJaATWlHBLSOzCNXVU56WRFgpVhD3bH3En20+oJz4DV2m4HgHvLEFZgfgrII2/5492F0unpv9IiVlMeoSIlAvjB8xku9XriVktZLh9aEluLn4kWd55u7bCGzf0P6D0dyIsBI8UkHQZjSl8Zc/ZTBi8NNsWJvG3AVBPt6cy7qdOnUNOcT0jOZHx8nsjZgbWK319OxWyoUX6Fx1dX8yc3ZQP3s987fXs6wGZv2lI9PFxEOS/8gHld8yTr81BtXVoDd/U1XTNAzDQNM0lLKg61EsFisjBg+jrLqCoqKW4+edDo2MtOQjIsDBcAxHIMbULAc1yse3RY1EvU5SgkJttpv5084l5dZXmW6bg2viGVSN+x1pEcX6Jp3yQBNdElwMyV6Ce0mYtctreHBBMtdV2KkJ1vHmx2sIXdWVJaXAxlLo0RX6WSHavpS53SQkJOD1uhiZ14OqkrXMq7JjhMN4JgzCnT+AzCF9kH7ZDM0YSnJmDkkSISfpOkrds8nIzKBu8y5WJ20ltV8O1Hgp376LSMwf74XqZMYdPZrcbj148eWXgLgeNEg8NQygsbGJ599Z0PpotAgYby/CCLev9So0XCrCV+tm81/9HvysBl14afs8Gm+Yzt8f/DvdRg1BaRoDTjmL1487iUgoyLaibfTpN4Cl387DbxhUV1aycuVKbrj2Wt5+/30KCwtJy2r9g7SHok0CrCwWHNEyokvsbC2w4wtYWNXTg88BbodiWyM4gLVJsNyw0E2zMXjJZoqsMWqSknDYYQLxXn6NeNndYmwHshUkBiKcnWdlogUWFbb/i60NNTWsramJZwJodtwuJ0meBBobyklI64M71YPDaVDx3TdM6TaYj3ZsxOhASokAldZIx1StFRISfAxIn8CQoQUMG+PkwiuESCiXijKd7ZstVFTW8Pyz3/H5nD4k+NLJzq5jyDA/55+fxIRx+SSnusDwE3nlVxhffk9Vbfzt494FHeuE2X2m9ko+BATKgGURsOh7609OTqa6uprRg4fgdnn4euF3HDP1JN545RVWLl7OqaefSDiy/6vmzp2VlJdWdcDCvVgsGusEKmtiaBY7SUmK8poANix0z7GTl+Rhy1VXs+bpefSZ8B2VdhupykGopoHVhVvJGTqArslnc3LF3SyZ9SkZRhNjI2FWpWYyemg3PiirYeEsNyQkwy4D/rAYjPaloO0mGAzw6xmncfzUcTTyDPPf3YaRk0qsOoDzjJ4kDh2NEWxgSlo2u2xu8vyNDHbl8/7Q00lFsXPbB0Qa0wjqQkP9DiISw941jUhR1cFfO35CrAp8VqiJwrbCQi69+CLeevtNAsH4U1qHli+h+zjsu4daiwh1r9S0mNXucHETZVH0cbbwZwz2ti9DhPe++Zoxs17g1LrT0ZM85HfrhjctFbfLzYDkVCQapWjrVubNmsW9D/6Jh/75TzSlSElJ4fY77yR6QBs+HNokwDqKgNJYX2AnyW2n0eOnomsCiTvj00/38EGqDRI1SLJB5eg05pbbyMtM4DQbDCb+cmwlfm6jwAbimpUHIEJFWQPnaB4MzYI3x97xF/poCAdhsnvls62hEFCEYxkUVpWTF6vj483PckK/oSS6XCQlJFBaVU3oEJ+RPiQ2H8SO7NeHUxPz+d09jzDt+yZm3HUyTpeGwwVdu1np2g1EUjhuwomsWlJB1wE2MnK74nJb0TS1T6PV+fz5XVgWNLKjuZF3NMdgI3ADcD/xzpOZwBM7wFsL56aA3b43IaK6Oj4Z75JVK0lIzCA9K4e777qH1KQErC6FXVNE2P/+s1gs+BITqaqp6aClYNcUOhpGKEpTWEh16fRLSyA9zcLqOmg0hOCwY5kz5nSuGJbB0p1VDPcq5t/1a7SeeUx97G/o4QWMGZiH2xLFk55GQ3YVJaHefBRxkFJdSYbRheJRYcj3wjcZsNDfoYexrhsQszHgmONYsLkQ2yclWPr2wRXRcdnt1P6wmO7ZfVhRVY6WmQv2VNI0xYWp3dnQUM12V3fO7TqCEm9XPt32COVraiHLB5U1ENDB2Dsw4qdG0xS6IUQk/qa7rbyC72e/ypTuOXywtmUKl8tlJxKJIAK7vwTmtFiIGgYxEUb3TWD1uvY5Z0F2Ucg9CC33N0SY/9knDPU4ufnFl7BYrVx9/XVccMllJCanEti1i159uzF8wADSs7JRSlG4aRPPzJyJ02bD6nK32Z62xYDrDayBRDxZHrweIYqDFD/UiMLngUHNnWwGYEPhsUNjVhI+K/RSkEj8osfY6wE7iXu/AJpSHNUzjcK6KGkuK8l2dURyyCMIOYEgS5fsAIRo3Rw0ixOrvQs5Y6bx/uK30IIBpgw+Cnt+F2KrPmGD0cAPsWjbvMTgkRVfgG07V7GNVczbXESP3l5OvnjiXrdTj6BWvEPSe6uYeNx46HMiaAfmRwqq9FWKNxXzYjCebx1f2zGagBeBGcQF+AQgswskdgWtHtKOb/aKN7PHk9Hsdn5x7bWcP20SY0cO55mX32b2u7OwoKPU/rFit8vFc88+y7PPv8CHH37QoYT3WDRGT5vgTbaxsrgGX4JiZWUd43UXpYW1BCNlrPt6HuIbwmdfJfLB57czN8FB3dq5rI71ZMXSV3jsyVl88cVaJkycwK9uv5O7LruUNf3HMG6jn+JTs8gZopM/aC0LJh+N0bsbrFu93yTwbcVlg9EjR2N35bN+azmGHiFn8AB2ba+k3t9EliuZpKwMMhN8pDoVaWJQEhFyNQ8Wbwq9Rh/FDrcddxTC1eDOSqFhVw0keiHcAIaBlfi9eCRF2Gq1EovFnzx2h43Bp/QmGGgg15vBxrW70I0Y0Xo/JTWVnN4vl6/XbcW/jwEOt4NjLhvKms82UlPhJ6THcGhWjh0+jO+WraExFGLJcj+atX3TJIY59AP9402bWfrwPykPRjGAG++8m5f+M5Orr7+C4UOGk91nII1VleixGLuKirjyyivZsGEDqR4PpRUVbbZHtaVhK6UqgbZNeNlx8kWk5djaH6GTbIX/W/a2y1b4v2Wv2RYOC7Mt/LS0am+bBNjExMTE5Mhx5L73Y2JiYmLSJkwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SRMATYxMTHpJEwBNjExMekkTAE2MTEx6SSsbSmsVJpAPrAF8AP6/ttRCLLn/0pZsWhW7HYvOdmJeL0JKGvrmh8MwoYNBoYRAbYCwT3bRES1xU4Am1JiAKqFlWADNA26dcvD6U3GYrESiURYt24dun5g6TZTJSLpHa3ExMTkZ4CIHPYCyQIzBdIF2G+xoMSFRQDRQCwHbHdYbHJXv9Nk8SMfSywSk30xDJH7/iCilAg0Cgzbb9+22Lh7cYI4D7Bh30WBJHsdcslFp4sei8i8b78Wu8120PJtWJa1x15zMRdz+fktbQxB1AILgHoUce9ytwutIwSbfU0LYD9gz7Ae5e/rP+Lqf95CU2PjftsiBiz4ASQF0NygjqGj0ZFw85LSbM+BCFDbGKZh2w4a6qqY+eIzRKLRH633UK74xNET22WriYnJz5N2qNxcIIqzeWdHKyWi7BtA2EsMQXclULSzkQ0lQXYI/AA8EIV5NwFfAbMUPPIA5E5ru2n7sNsdNVrZpu0jo8u3FHLa2afy/POvH1a9eQkp2LSWkq4pjRk3zGifsSYmJj9L2iHAGwEhSDy22vQjpVOIx1wtgMuicd0pU7nxptu4/71VzECYDcxxwA1HwUlDIPN8xdHX+UgfNbntprVC1G1BDviVQ3ILmH7m2RxT0IfySj8L5qxAjJZSfWCA/PTjR3PrnbeRlJTSomxB1wKmTJxyRGw2MTH5eXDEsiAstPSG7UphaBoakAR0TfRy4YyzePjP9zElw8rUynq2inCPguuBsQYMDMOqCqG2vjtwHYd+6f9xHEq1+JGnnTad/775Bh++/T6XTZtGAhq2VvojYwf8Xba9kq8//Iyq2qoWZc+cdibpuWbfm4mJyeHTJgHWlIvspF7cdf2tnD3qKNITsnHZCnDbE8mx2Rjk8mLbJ+KaLEKOYRAmHgoINNazYcYMBuZEufSMQdS+NY9ta0u4RWBSCTz4K+Gb40poGnkfsW//DMwg7j+3n9qmGLF9nFsFVBdtoGhdPZ5BPema14epx5/KjdOm4XO0Fi1u3k8puhQMZv36XYjIftuSPSlcfOHFaJqZ1WdiYnL4tCkNTVPCNTMe5Ld/OgN/bS0VJWF2FFWzZu3nPPf0v6mqq0GCe9O4XIkp1NTXoCmIeqAsCJd9t4q3lq7FN8xC0bfv0bNPPhv657Ljn8BTETD+CjzTbFoJrUeZDx8F7CuXVqX4Ytl85F+fc/sdx3L00eO46qoZvPPeSxQs/IzVlYFW6xER3vni/RbrLZqDO37xOwYPHNwhO01MTH5+qAO9uUORnZYla1euJ7lLMmqfyMD27dsZMXw4ibW1BIEywK0UiZ4Eav2NRIh7wJpSjBnUizeffxTPxk0w5SI2OJw8l+BBSmHRZ8KOL0L4F76ElM2EcClQhIjR5jiEUuqAH7bbO41nB+d36c8tV0znL088SFpKCuu3bSPWjhzgicNO5eM/vIUnxQGDQHnVchEZ2eaKTExMfna0yQNOzUjHlb6/+AK43W7cHg+R+jpOcNkpDOoMSXXyXLWfAb19jD9mAG5bDoMG9mDSMb357s2n+ffjX/HiK3mMOWEaoxSoHAhMV2y/zMXimhks3XYhG+fXsuDPxx+ZX6q5AQGjCRAa/Dv5euFnVNXXUVpTc1hVWInL9+6IRpIniXtvvB33e3a4hdZTLkxMTEwORluShkeMGCEtiAQkEg7JhKPHykiXQ9aOHSxVf/qj3D9pggByxwXDpHDxzbJx/g3yzl9OlguGZ4nbokSB3DaolwS3ro+PxNgHo3mJGCJD4sdsc4IzIEod/gAKu4ZoP1Le7XDIgH6D9gwymXHeL+ODSioMkWh8cAnmQAxzMRdzOcylTR5wS/U2IFwN1mRcepDSYJhzlq1nwvbH0DwuFPDImyv5z7srEQF/VDD2CQw8umYz2s2/5sFZr2JxJ+9Zv9vBtqk2uugHmnf40RViAv3H5FE7tRb3hixqvm2guqJ8z/bUzAzGn3wSWzZsQgfsVgeXXHE5mhVIN9g/0mxiYmLy43Sw274ePHYsNidTBwzC4nFx1jETGJjbj3B9MQkKwrpQHxYaIvuLL0AEmPnpF2z+dFbb1PInwBDYurGWBPsgCkYci9dw4trn9NRUVvHJK6/yw8KFAPTq0ovBfQahUM2Dr62ws7OsNzEx+b9IBwRYQJWBSkBpGtf97e988sG7ZPXtyWa7EGtIpOkwNLUyYvDIX58gWreR+Bi6ziNQ28jGexbwxW+eYUfVDoL7BHXFMIhGInv+7t+jH960BECDsILZxIf1mZiYmBwmHRBgA0gEXCilcKVm0m3SVK75x7/45R/v5TNr+LD7pN5cvp7Nsx4Fifx44f8R+6WutbI94oogToFqkGsg/C3Uh/5HxpmYmPw/QQcEuBFIZt+RakoJVucCnClpGPYDp+M5ONUG/OM/Mwl9PhMxDhx/9r/jwIzjlOZ1QvxE7Zv84fBYUJqCIAQHR5hrhLGN/19ZamJi8v8C7RTg3ZLkPGB9BPiA7v2zGX/MMW2q8dU1If5zxV3Evn4LORLxYNvhPwB2EyU+2/Hu8XC1xGdU04n7+/tatWtXBbFYjMYEP18mfsGYG2O4Ujtos4mJyc+KdgqwQesj1GzAeCwWG0mJiW2qMQTcV+rn4VvuQHaspcNZBartAmwQH3u3ezjGoSzYsG4D2zdt44NXXmNYQV+81S5UqHNj2CYmJv+3aIcAC3G/0ErLiXI04DTCTRvZsn5Jm2tuAB5dW8z3T9yFRFqb0LINuBLatdvhBkCcVgeVX4cYmz+R3N/kI1MNou8UteuYJiYmP086EANufeIakXf47sMzWL5yQ7tq3WnAjf/5nLpvPuhYalpSMmT0bf/+h0BTFq7tfieDBg6i2+AeyD/D6K+VE+1R+JMcz8TE5P9N2inAh5oz4QyWfD+FYAfexhc1Rvnrn/6J4a9vfyWJbrjvv5CT1/46Dla118p5D/bDO1RDWRTa0XXYJlbgHlZwxI9lYmLy/y4dEOADvVMDaEQpG5de/XumHHMMfXr2pFduLm6t5Zy8h8IAXlu0nKIPD+8rFa1iAe3Mo7C/9i6ecy7E4jiww7DtaJoiP78r995+B/kDxqAKiZ8KPRPUUHD06fAxTExMfj60c6Sve5//C3EVqgC8gJ0u3Xvw/iefYMRiNJRt59IzTmLBxlKCbYgoFEUNXv3LY4il/YOR7ekaetowvM8/h+3xUdTd8xuItc0115SV3l3yOXHiaEZPmcjEqSeSnZmH2qGgN+BUYOvYiG4TE5OfJ+1QDsXeT24K6DtB84DKYHdcWCmF2xOffczj68/L7/2WDz+s5tGnZrJu6w6Mw4jtCjBr3Xq0rrltNxHQdOimC9k2RcjmZPuvriH69RyaZrec09fWayCWEaNJtsbw1zWRneKhSdOo7d+fUen9eH7S0eR3TUIpxZ6p4AqaT8XufkgBjP//DCQxMTH5/z8dd920VMBF658OWoZSXcnqcxy/7LaUYwf/ixMvu5qtZRWHVfVa3SCrrPzHC7aCxwYhO9RrkAUMcjnZ9bt7+XDBPPDvP/2k0diI89qrGDZhNEmG4NIUicTjMwNEsbYOMmrArYDdub7NMZX4vGtCLGKwzdKxzyeZmJj8vOjgZDwKlJsWEwTv3sZg4jPUVKKsQ1m8agO7KqvJTEjC40qka2om1115DUmJSa3WLkBZKNwuywRI0uJ2VQFfKkXxyMG4xx7VoqxetoOkx/7JJSL0t2jkKcU5SnGjUiRr4E2CN4GtleCXeM5yiHjgJUo8+OK3a3xm6djnk0xMTH5e/ETBy90hBgcwEmgCLcopvxjIqOe6cvrks8mdMI77f/cA9/7hAdJz83nggd8i0nL2iHYnoonQH+gPVBIfo+e1WfGccSrzv/kSYvuHC9IcNnKBrs3HXA58AXQDwho0pUJdKgSIB1o2NG/rTTzksp34h0dNTExMDpefsPfopeZ/+wN1wDEkZY3i7U9nkuAajpbo4KWX3+CH1eu57roref31l6gq30VtQyN6Oz4NdCAZSjGGuJfanfgPPVophl8+neSQzrKnHqVs82YAbE4ng04/nWoVDz1Em/fpzd6PXExQcU/aR3wWjEziYl3bvEQAf4etNjEx+TnxEwqwGxgI5AK/B/JQaiXpBeuAwYh4mTJlEp989AWTj5/A5ZddzvijRlNSGmTB4u948sm/E4m0v1PLAZwLbGu2xArkAd3cTnrefD2rzjqVhS+9Sk1dEz0H9KHfGWfQpBQ9iUe0FwD9iAvxGmAOMIW40PZs/tdFfExgKvE+uWXtttbExOTnSJs+yqmUqgR2/HTmtEq+iKS3dadOshXaaa+JicnPjzYJsImJiYnJkaODWRAmJiYmJu3FFGATExOTTsIUYBMTE5NOwhRgExMTk07CFGATExOTTsIUYBMTE5NOwhRgExMTk07CFGATExOTTsIUYBMTE5NO4v8DgMci6xPpLNoAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 360x360 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0. 1.]\n",
      " [0. 1.]\n",
      " [0. 1.]\n",
      " [1. 0.]\n",
      " [0. 1.]\n",
      " [0. 1.]\n",
      " [0. 1.]\n",
      " [1. 0.]\n",
      " [0. 1.]\n",
      " [1. 0.]]\n"
     ]
    }
   ],
   "source": [
    "def plotImages(images):\n",
    "    plt.figure(figsize=(5,5))\n",
    "    for i in range(len(images)):\n",
    "        plt.subplot(5,8,i+1)\n",
    "        plt.xticks([])\n",
    "        plt.yticks([])\n",
    "        plt.grid(False)\n",
    "        plt.imshow(images[i])  \n",
    "    plt.tight_layout()     \n",
    "    plt.show()   \n",
    "imgs, labels = next(train_batches)\n",
    "plotImages(imgs)\n",
    "print(labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_model(name):\n",
    "    if name == '1':\n",
    "        inputs = keras.Input(shape=(224,224,3), name=\"images-input\")\n",
    "        x = layers.Conv2D(32, 3, activation='relu', padding='same')(inputs)\n",
    "        x = layers.MaxPool2D(pool_size=2, strides=2)(x)\n",
    "        x = layers.Conv2D(64, 3, activation='relu', padding='same')(x)\n",
    "        x = layers.MaxPool2D(pool_size=2, strides=2)(x)\n",
    "        x = layers.Flatten()(x)\n",
    "        outputs = layers.Dense(units=2, activation='softmax')(x)\n",
    "        model = keras.Model(inputs, outputs, name='conv3_classifier')\n",
    "    elif name == 'Resnet':\n",
    "        inputs = keras.Input(shape=(224,224,3), name=\"images-input\")\n",
    "        x = layers.Conv2D(64, 3, activation=\"relu\", padding=\"same\", strides=1)(inputs)\n",
    "        x = layers.Conv2D(64, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(128, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(128, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding=\"valid\")(x)\n",
    "        \n",
    "        x = layers.Flatten()(x)\n",
    "        x = layers.Dense(4096, activation=\"relu\")(x)\n",
    "        x = layers.Dense(4096, activation=\"relu\")(x)\n",
    "        outputs = layers.Dense(2, activation=\"softmax\")(x)\n",
    "        model = keras.Model(inputs, outputs, name='Resnet_classifier')\n",
    "    elif name=='Vgg':\n",
    "        inputs = keras.Input(shape=(224,224,3), name=\"images-input\")\n",
    "        x = layers.Conv2D(64, 3, activation=\"relu\", padding=\"same\", strides=1)(inputs)\n",
    "        x = layers.Conv2D(64, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(128, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(128, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(256, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2),strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.Conv2D(512, 3, activation=\"relu\", padding=\"same\", strides=1)(x)\n",
    "        x = layers.MaxPooling2D(pool_size=(2, 2), strides=(2, 2), padding=\"valid\")(x)\n",
    "\n",
    "        x = layers.Flatten()(x)\n",
    "        x = layers.Dense(4096, activation=\"relu\")(x)\n",
    "        x = layers.Dense(4096, activation=\"relu\")(x)\n",
    "        outputs = layers.Dense(2, activation=\"softmax\")(x)\n",
    "        model = keras.Model(inputs, outputs, name='Vgg_classifier')\n",
    "    else:\n",
    "        raise RuntimeError('error')       \n",
    "    return model\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"Vgg_classifier\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "images-input (InputLayer)    [(None, 224, 224, 3)]     0         \n",
      "_________________________________________________________________\n",
      "conv2d (Conv2D)              (None, 224, 224, 64)      1792      \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 224, 224, 64)      36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 112, 112, 64)      0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 112, 112, 128)     73856     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 112, 112, 128)     147584    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 56, 56, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 56, 56, 256)       295168    \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "conv2d_6 (Conv2D)            (None, 56, 56, 256)       590080    \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 28, 28, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_7 (Conv2D)            (None, 28, 28, 512)       1180160   \n",
      "_________________________________________________________________\n",
      "conv2d_8 (Conv2D)            (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "conv2d_9 (Conv2D)            (None, 28, 28, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 14, 14, 512)       0         \n",
      "_________________________________________________________________\n",
      "conv2d_10 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "conv2d_11 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "conv2d_12 (Conv2D)           (None, 14, 14, 512)       2359808   \n",
      "_________________________________________________________________\n",
      "max_pooling2d_4 (MaxPooling2 (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 25088)             0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 4096)              102764544 \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 4096)              16781312  \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 2)                 8194      \n",
      "=================================================================\n",
      "Total params: 134,268,738\n",
      "Trainable params: 134,268,738\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = get_model(algorithm)\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:sample_weight modes were coerced from\n",
      "  ...\n",
      "    to  \n",
      "  ['...']\n",
      "Train for 801 steps\n",
      "Epoch 1/50\n",
      "801/801 [==============================] - 68s 85ms/step - loss: 0.7005 - accuracy: 0.5088\n",
      "Epoch 2/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.6930 - accuracy: 0.5187\n",
      "Epoch 3/50\n",
      "801/801 [==============================] - 62s 78ms/step - loss: 0.6965 - accuracy: 0.5324\n",
      "Epoch 4/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.6879 - accuracy: 0.5467\n",
      "Epoch 5/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.6730 - accuracy: 0.5898\n",
      "Epoch 6/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.6155 - accuracy: 0.6611\n",
      "Epoch 7/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.5593 - accuracy: 0.7132\n",
      "Epoch 8/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.5082 - accuracy: 0.7544\n",
      "Epoch 9/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.4506 - accuracy: 0.7900\n",
      "Epoch 10/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.3993 - accuracy: 0.8240\n",
      "Epoch 11/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.3376 - accuracy: 0.8525\n",
      "Epoch 12/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.2691 - accuracy: 0.8879\n",
      "Epoch 13/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.1883 - accuracy: 0.9234\n",
      "Epoch 14/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.1170 - accuracy: 0.9532\n",
      "Epoch 15/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.1009 - accuracy: 0.9621\n",
      "Epoch 16/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0623 - accuracy: 0.9783\n",
      "Epoch 17/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0578 - accuracy: 0.9803\n",
      "Epoch 18/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0464 - accuracy: 0.9820\n",
      "Epoch 19/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0449 - accuracy: 0.9866\n",
      "Epoch 20/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0421 - accuracy: 0.9855\n",
      "Epoch 21/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0396 - accuracy: 0.9864\n",
      "Epoch 22/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0330 - accuracy: 0.9898\n",
      "Epoch 23/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0416 - accuracy: 0.9870\n",
      "Epoch 24/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0215 - accuracy: 0.9941\n",
      "Epoch 25/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0290 - accuracy: 0.9909\n",
      "Epoch 26/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0283 - accuracy: 0.9908\n",
      "Epoch 27/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0303 - accuracy: 0.9906\n",
      "Epoch 28/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0234 - accuracy: 0.9919\n",
      "Epoch 29/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0332 - accuracy: 0.9890\n",
      "Epoch 30/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0297 - accuracy: 0.9909\n",
      "Epoch 31/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0255 - accuracy: 0.9925\n",
      "Epoch 32/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0248 - accuracy: 0.9920\n",
      "Epoch 33/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0206 - accuracy: 0.9936\n",
      "Epoch 34/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0199 - accuracy: 0.9939\n",
      "Epoch 35/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0166 - accuracy: 0.9954\n",
      "Epoch 36/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0245 - accuracy: 0.9928\n",
      "Epoch 37/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0121 - accuracy: 0.9960\n",
      "Epoch 38/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0261 - accuracy: 0.9918\n",
      "Epoch 39/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0216 - accuracy: 0.9929\n",
      "Epoch 40/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0250 - accuracy: 0.9928\n",
      "Epoch 41/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0131 - accuracy: 0.9963\n",
      "Epoch 42/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0147 - accuracy: 0.9963\n",
      "Epoch 43/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0276 - accuracy: 0.9911\n",
      "Epoch 44/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0124 - accuracy: 0.9969\n",
      "Epoch 45/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0238 - accuracy: 0.9923\n",
      "Epoch 46/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0128 - accuracy: 0.9958\n",
      "Epoch 47/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0098 - accuracy: 0.9976\n",
      "Epoch 48/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0188 - accuracy: 0.9944\n",
      "Epoch 49/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0224 - accuracy: 0.9945\n",
      "Epoch 50/50\n",
      "801/801 [==============================] - 63s 78ms/step - loss: 0.0257 - accuracy: 0.9930\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer=keras.optimizers.Adam(learning_rate=lr),\n",
    "     loss='categorical_crossentropy', metrics=['accuracy'])\n",
    "\n",
    "history = model.fit(x=train_batches, epochs=epochs, verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import datetime\n",
    "datetime_dt = datetime.datetime.today()# 獲得當地時間\n",
    "datetime_str = datetime_dt.strftime(\"%Y-%m-%d_%H-%M\")\n",
    "model.save(f'{model.name}_{datetime_str}.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "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>loss</th>\n",
       "      <th>accuracy</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>45</th>\n",
       "      <td>0.012761</td>\n",
       "      <td>0.995753</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>46</th>\n",
       "      <td>0.009781</td>\n",
       "      <td>0.997626</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>0.018828</td>\n",
       "      <td>0.994379</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>48</th>\n",
       "      <td>0.022460</td>\n",
       "      <td>0.994503</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>49</th>\n",
       "      <td>0.025677</td>\n",
       "      <td>0.993004</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "        loss  accuracy\n",
       "45  0.012761  0.995753\n",
       "46  0.009781  0.997626\n",
       "47  0.018828  0.994379\n",
       "48  0.022460  0.994503\n",
       "49  0.025677  0.993004"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "history_frame = pd.DataFrame(history.history)\n",
    "history_frame.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "history_frame.loc[:, ['loss']].plot()\n",
    "plt.savefig(f'{model.name}_loss_{datetime_str}.png')\n",
    "history_frame.loc[:, ['accuracy']].plot()\n",
    "plt.savefig(f'{model.name}_acc_{datetime_str}.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n",
      "Clipping input data to the valid range for imshow with RGB data ([0..1] for floats or [0..255] for integers).\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 360x360 with 10 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]\n",
      " [1. 0.]]\n"
     ]
    }
   ],
   "source": [
    "test_imgs, test_labels = next(test_batches)\n",
    "plotImages(test_imgs)\n",
    "print(test_labels)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10/10 [==============================] - 0s 19ms/sample - loss: 2.7377 - accuracy: 0.8000\n",
      "Test loss: 2.73773193359375\n",
      "Test accuracy: 0.8\n"
     ]
    }
   ],
   "source": [
    "x, y = next(test_batches)\n",
    "test_scores = model.evaluate(x, y, verbose=1)\n",
    "print(\"Test loss:\", test_scores[0])\n",
    "print(\"Test accuracy:\", test_scores[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "203/203 - 5s\n"
     ]
    }
   ],
   "source": [
    "predictions = model.predict(x=test_batches, verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2023, 2)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, ..., 1, 1, 1])"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_true = test_batches.classes\n",
    "y_true"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2023,)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_true.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0, 0, 0, ..., 0, 1, 1], dtype=int64)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred = np.argmax(predictions, axis=-1)\n",
    "y_pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(2023,)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_pred.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[843, 168],\n",
       "       [255, 757]], dtype=int64)"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "cm = confusion_matrix(y_true=y_true, y_pred=y_pred)\n",
    "cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "import itertools\n",
    "def plot_confusion_matrix(cm, classes,\n",
    "                          normalize=False,\n",
    "                          title='Confusion matrix',\n",
    "                          cmap=plt.cm.Blues,\n",
    "                          date=''):\n",
    "    \"\"\"\n",
    "    This function prints and plots the confusion matrix.\n",
    "    Normalization can be applied by setting `normalize=True`.\n",
    "    \"\"\"\n",
    "    test_accuracy = (cm[0][0] + cm[1][1]) /(cm[0][0] + cm[0][1] + cm[1][0] + cm[1][1])\n",
    "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
    "    plt.title(f'{title}, test_acc: {round(test_accuracy,3)}')\n",
    "    plt.colorbar()\n",
    "    tick_marks = np.arange(len(classes))\n",
    "    plt.xticks(tick_marks, classes, rotation=45)\n",
    "    plt.yticks(tick_marks, classes)\n",
    "\n",
    "    if normalize:\n",
    "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
    "        print(\"Normalized confusion matrix\")\n",
    "    else:\n",
    "        print('Confusion matrix, without normalization')\n",
    "\n",
    "    #print(cm)\n",
    "\n",
    "    thresh = cm.max() / 2.\n",
    "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
    "        plt.text(j, i, cm[i, j],\n",
    "            horizontalalignment=\"center\",\n",
    "            color=\"white\" if cm[i, j] > thresh else \"black\")\n",
    "    \n",
    "    plt.tight_layout()\n",
    "    plt.ylabel('True label')\n",
    "    plt.xlabel('Predicted label')\n",
    "    plt.savefig(f'confusion-maxtrix_{date}.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion matrix, without normalization\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_confusion_matrix(cm=cm, classes=classes, title='Confusion Matrix', date=datetime_str)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}