Face_Expression_Regoniser.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Face Expression Regoniser.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU",
    "gpuClass": "standard"
  },
  "cells": [
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "o8XahJvJqdKM",
        "outputId": "a9ba18a0-3dd6-4a98-fdfa-95122d664f1a"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: kaggle in /usr/local/lib/python3.7/dist-packages (1.5.12)\n",
            "Requirement already satisfied: tqdm in /usr/local/lib/python3.7/dist-packages (from kaggle) (4.64.0)\n",
            "Requirement already satisfied: python-dateutil in /usr/local/lib/python3.7/dist-packages (from kaggle) (2.8.2)\n",
            "Requirement already satisfied: python-slugify in /usr/local/lib/python3.7/dist-packages (from kaggle) (6.1.2)\n",
            "Requirement already satisfied: requests in /usr/local/lib/python3.7/dist-packages (from kaggle) (2.23.0)\n",
            "Requirement already satisfied: six>=1.10 in /usr/local/lib/python3.7/dist-packages (from kaggle) (1.15.0)\n",
            "Requirement already satisfied: certifi in /usr/local/lib/python3.7/dist-packages (from kaggle) (2022.6.15)\n",
            "Requirement already satisfied: urllib3 in /usr/local/lib/python3.7/dist-packages (from kaggle) (1.24.3)\n",
            "Requirement already satisfied: text-unidecode>=1.3 in /usr/local/lib/python3.7/dist-packages (from python-slugify->kaggle) (1.3)\n",
            "Requirement already satisfied: chardet<4,>=3.0.2 in /usr/local/lib/python3.7/dist-packages (from requests->kaggle) (3.0.4)\n",
            "Requirement already satisfied: idna<3,>=2.5 in /usr/local/lib/python3.7/dist-packages (from requests->kaggle) (2.10)\n"
          ]
        }
      ],
      "source": [
        "! pip install kaggle"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7ieQ3mjjse0Z",
        "outputId": "dd36d20c-bb80-4e91-f8db-7cc470676f72"
      },
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!rm -r ~/.kaggle\n",
        "!mkdir ~/.kaggle\n",
        "!cp \"/content/drive/MyDrive/Colab Notebooks/kaggle_api_credentials/kaggle.json\" ~/.kaggle/\n",
        "!chmod 600 ~/.kaggle/kaggle.json\n",
        "!kaggle datasets list"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "epjg1EDEskx1",
        "outputId": "2012da4a-6a07-44ad-8302-62e0b04f0f56"
      },
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "rm: cannot remove '/root/.kaggle': No such file or directory\n",
            "ref                                                       title                                               size  lastUpdated          downloadCount  voteCount  usabilityRating  \n",
            "--------------------------------------------------------  -------------------------------------------------  -----  -------------------  -------------  ---------  ---------------  \n",
            "ruchi798/data-science-job-salaries                        Data Science Job Salaries                            7KB  2022-06-15 08:59:12           7909        267  1.0              \n",
            "surajjha101/bigbasket-entire-product-list-28k-datapoints  BigBasket Entire Product List (~28K datapoints)      6MB  2022-06-22 12:51:18           2800        113  1.0              \n",
            "victorsoeiro/netflix-tv-shows-and-movies                  Netflix TV Shows and Movies                          2MB  2022-05-15 00:01:23          17668        505  1.0              \n",
            "sameepvani/nasa-nearest-earth-objects                     NASA - Nearest Earth Objects                         7MB  2022-06-17 02:32:18           2682        121  1.0              \n",
            "zusmani/petrolgas-prices-worldwide                        Petrol/Gas Prices Worldwide                         10KB  2022-06-24 01:25:33           2521        121  1.0              \n",
            "mukuldeshantri/ecommerce-fashion-dataset                  E-commerce Dataset with 30K Products               546KB  2022-07-08 12:28:18            869         34  1.0              \n",
            "jimschacko/airlines-dataset-to-predict-a-delay            Airlines Dataset to predict a delay                  6MB  2022-06-21 05:45:44           2433         64  1.0              \n",
            "iamsouravbanerjee/nifty500-stocks-dataset                 Stock Market Dataset (NIFTY-500)                    35KB  2022-07-17 11:46:10            839         62  1.0              \n",
            "zusmani/pakistans-national-assembly-attendance-20182022   Pakistan's National Assembly Attendance 2018-2022    2MB  2022-07-15 03:33:32            259         32  0.88235295       \n",
            "dansbecker/melbourne-housing-snapshot                     Melbourne Housing Snapshot                         451KB  2018-06-05 12:52:24          91437       1112  0.7058824        \n",
            "ramjasmaurya/1-gb-internet-price                          Internet Prices around 200+ countries in 2022.      22KB  2022-07-17 02:56:12           1425         48  0.9705882        \n",
            "datasnaek/youtube-new                                     Trending YouTube Video Statistics                  201MB  2019-06-03 00:56:47         179561       4606  0.7941176        \n",
            "zynicide/wine-reviews                                     Wine Reviews                                        51MB  2017-11-27 17:08:04         163296       3335  0.7941176        \n",
            "deepcontractor/car-price-prediction-challenge             Car Price Prediction Challenge                     429KB  2022-07-06 11:38:32            682         47  0.9411765        \n",
            "datasnaek/chess                                           Chess Game Dataset (Lichess)                         3MB  2017-09-04 03:09:09          30534       1019  0.8235294        \n",
            "rtatman/188-million-us-wildfires                          1.88 Million US Wildfires                          168MB  2020-05-12 21:03:49          20748       1021  0.8235294        \n",
            "residentmario/ramen-ratings                               Ramen Ratings                                       40KB  2018-01-11 16:04:39          34864        793  0.7058824        \n",
            "dansbecker/powerlifting-database                          powerlifting-database                                9MB  2019-04-30 21:07:41           4956         60  0.5882353        \n",
            "jpmiller/publicassistance                                 US Public Food Assistance                          703KB  2020-08-21 16:51:18          16656        396  0.9117647        \n",
            "nasa/kepler-exoplanet-search-results                      Kepler Exoplanet Search Results                      1MB  2017-10-10 18:26:59          10653        664  0.8235294        \n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "! kaggle datasets download sionehoghen/facial-expression"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "bbvxkrWlsuki",
        "outputId": "f453feec-6581-4de4-a4fc-8b3f681d70b2"
      },
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading facial-expression.zip to /content\n",
            " 68% 41.0M/60.3M [00:01<00:00, 24.7MB/s]\n",
            "100% 60.3M/60.3M [00:01<00:00, 35.3MB/s]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "! unzip -q facial-expression.zip"
      ],
      "metadata": {
        "id": "rRBTRW2YsxMT"
      },
      "execution_count": 5,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "import tensorflow as tf\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "from pathlib import Path\n",
        "\n",
        "tf.random.set_seed(4)"
      ],
      "metadata": {
        "id": "FKef2PtZs5bu"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "train_path = Path(\"/content/train\")\n",
        "test_path = Path(\"/content/test\")"
      ],
      "metadata": {
        "id": "O5iHCioet8Zh"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# getting image paths\n",
        "train_image_paths = list(train_path.glob(\"*/*\"))\n",
        "train_image_paths = list(map(lambda x: str(x), train_image_paths))\n",
        "\n",
        "print(train_image_paths[:10])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "8VQVp6PRuGiE",
        "outputId": "ab40c75f-4b65-4cba-d538-4cb0b60b86ad"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['/content/train/angry/Training_2778588.jpg', '/content/train/angry/Training_8813906.jpg', '/content/train/angry/Training_38018954.jpg', '/content/train/angry/Training_67027311.jpg', '/content/train/angry/Training_34928751.jpg', '/content/train/angry/Training_41568333.jpg', '/content/train/angry/Training_95669204.jpg', '/content/train/angry/Training_41898754.jpg', '/content/train/angry/Training_68732558.jpg', '/content/train/angry/Training_39536203.jpg']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# getting respective labels of images\n",
        "def get_label(image_path):\n",
        "  return image_path.split(\"/\")[-2]\n",
        "\n",
        "train_image_labels = list(map(lambda x: get_label(x), train_image_paths))\n",
        "\n",
        "print(train_image_labels[:10])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xgd-90mmujwZ",
        "outputId": "6b99b602-8efa-4be4-ee6c-6c36edab47ab"
      },
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['angry', 'angry', 'angry', 'angry', 'angry', 'angry', 'angry', 'angry', 'angry', 'angry']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.preprocessing import LabelEncoder\n",
        "\n",
        "Le = LabelEncoder()\n",
        "train_image_labels = Le.fit_transform(train_image_labels)\n",
        "\n",
        "print(train_image_labels[:10])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "we7F1R0IvBB2",
        "outputId": "de380319-8d22-4163-e90e-31d3b24c077f"
      },
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[0 0 0 0 0 0 0 0 0 0]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "train_image_labels = tf.keras.utils.to_categorical(train_image_labels)\n",
        "\n",
        "print(train_image_labels[:10])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "YQF5zRpfvQDe",
        "outputId": "d876eefb-2906-4b3d-fc64-c0d40134d85e"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]\n",
            " [1. 0. 0. 0. 0. 0. 0.]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "Train_paths, Val_paths, Train_labels, Val_labels = train_test_split(train_image_paths, train_image_labels, test_size=0.25)"
      ],
      "metadata": {
        "id": "U11s6JH5vadC"
      },
      "execution_count": 12,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "Train_paths[:5], Val_paths[:5], Train_labels[:5], Val_labels[:5]"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "jVyfc6aRvvA6",
        "outputId": "919727aa-2c9c-4b44-ebb2-1abb89f0efa5"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(['/content/train/fear/Training_79381162.jpg',\n",
              "  '/content/train/angry/Training_26591418.jpg',\n",
              "  '/content/train/surprise/Training_30949917.jpg',\n",
              "  '/content/train/neutral/Training_68423650.jpg',\n",
              "  '/content/train/sad/Training_15632432.jpg'],\n",
              " ['/content/train/angry/Training_52197003.jpg',\n",
              "  '/content/train/happy/Training_96485203.jpg',\n",
              "  '/content/train/angry/Training_66014049.jpg',\n",
              "  '/content/train/happy/Training_76276874.jpg',\n",
              "  '/content/train/sad/Training_79758901.jpg'],\n",
              " array([[0., 0., 1., 0., 0., 0., 0.],\n",
              "        [1., 0., 0., 0., 0., 0., 0.],\n",
              "        [0., 0., 0., 0., 0., 0., 1.],\n",
              "        [0., 0., 0., 0., 1., 0., 0.],\n",
              "        [0., 0., 0., 0., 0., 1., 0.]], dtype=float32),\n",
              " array([[1., 0., 0., 0., 0., 0., 0.],\n",
              "        [0., 0., 0., 1., 0., 0., 0.],\n",
              "        [1., 0., 0., 0., 0., 0., 0.],\n",
              "        [0., 0., 0., 1., 0., 0., 0.],\n",
              "        [0., 0., 0., 0., 0., 1., 0.]], dtype=float32))"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# compute class weights\n",
        "classTotals = Train_labels.sum(axis=0)\n",
        "classWeight = classTotals.max()/classTotals\n",
        "\n",
        "class_weight = {e: weight for e, weight in enumerate(classWeight)}\n",
        "print(class_weight)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "HN9h1wqav0xU",
        "outputId": "a7bb6148-4487-45af-bd61-738865dbf3be"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "{0: 1.8216325, 1: 16.139881, 2: 1.753881, 3: 1.0, 4: 1.4574039, 5: 1.4849398, 6: 2.327468}\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# load image data\n",
        "def load(image, label):\n",
        "  image = tf.io.read_file(image)\n",
        "  image = tf.io.decode_jpeg(image, channels=3)\n",
        "  return image, label"
      ],
      "metadata": {
        "id": "73kuuPOAwXeO"
      },
      "execution_count": 15,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# define image & batch size\n",
        "IMG_SIZE = 96\n",
        "BATCH_SIZE = 32\n",
        "\n",
        "# basic transformation\n",
        "resize = tf.keras.Sequential([\n",
        "    tf.keras.layers.experimental.preprocessing.Resizing(IMG_SIZE, IMG_SIZE)\n",
        "])\n",
        "\n",
        "# data augmentation\n",
        "data_augmentation = tf.keras.Sequential([\n",
        "    tf.keras.layers.experimental.preprocessing.RandomFlip(\"horizontal\"),\n",
        "    tf.keras.layers.experimental.preprocessing.RandomRotation(0.1),\n",
        "    tf.keras.layers.experimental.preprocessing.RandomZoom(height_factor = (-0.1, -0.05)),\n",
        "])\n"
      ],
      "metadata": {
        "id": "X1TJdZcGwoU3"
      },
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# function to create tf dataset object\n",
        "AUTOTUNE = tf.data.experimental.AUTOTUNE\n",
        "\n",
        "def get_dataset(paths, labels, train=True):\n",
        "    image_paths = tf.convert_to_tensor(paths)\n",
        "    labels = tf.convert_to_tensor(labels)\n",
        "\n",
        "    image_dataset = tf.data.Dataset.from_tensor_slices(image_paths)\n",
        "    label_dataset = tf.data.Dataset.from_tensor_slices(labels)\n",
        "\n",
        "    dataset = tf.data.Dataset.zip((image_dataset, label_dataset))\n",
        "\n",
        "    dataset = dataset.map(lambda image,label: load(image,label))\n",
        "    dataset = dataset.map(lambda image,label: (resize(image),label), num_parallel_calls=AUTOTUNE)\n",
        "    dataset = dataset.shuffle(1000)\n",
        "    dataset = dataset.batch(BATCH_SIZE)\n",
        "\n",
        "    if train:\n",
        "      dataset = dataset.map(lambda image,label: (data_augmentation(image), label), num_parallel_calls=AUTOTUNE)\n",
        "\n",
        "    dataset = dataset.repeat()\n",
        "    return dataset"
      ],
      "metadata": {
        "id": "l-fFds8UxoCB"
      },
      "execution_count": 17,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# creating train dataset object & verifying it\n",
        "%time train_dataset = get_dataset(Train_paths, Train_labels)\n",
        "\n",
        "image, label = next(iter(train_dataset))\n",
        "print(image.shape)\n",
        "print(label.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cw4apw-s16UI",
        "outputId": "c41cdc60-b798-4147-e97f-080db0e2a427"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CPU times: user 1.08 s, sys: 21 ms, total: 1.1 s\n",
            "Wall time: 1.83 s\n",
            "(32, 96, 96, 3)\n",
            "(32, 7)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# view a sample training image\n",
        "print(Le.inverse_transform(np.argmax(label, axis=1))[0])\n",
        "plt.imshow((image[0].numpy()/255).reshape(96,96,3))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "DqEgb_x_2NIw",
        "outputId": "45fc5a47-228c-44f0-cd21-583ea9d086b6"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "neutral\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f3df92948d0>"
            ]
          },
          "metadata": {},
          "execution_count": 19
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD7CAYAAACscuKmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO29XYxtW1Ye9s2q2lV1zrmXn+5ETacbhUZGtlqWCBYiWEQRAlshBJkXhDAWajtE/ZLY2HFkmuQBR3IkLFnG/RARXZlYJEJpfowCIpYdpwMPfmlDGxTbtNsmgJtuGu5FQNN97zn1t2ceqsaub3/1jTHnqnPururcGtLW2mutueYcc8wxxjfmz5qr9d7xQA/0QP//p727ZuCBHuiBdkMPxv5AD/QWoQdjf6AHeovQg7E/0AO9RejB2B/ogd4i9GDsD/RAbxF6LmNvrX1Ta+3jrbVfba194EUx9UAP9EAvntpt59lba/sA/jWAPw3gkwB+AcCf7b3/yotj74Ee6IFeFB08x7NfA+BXe++/BgCttQ8B+FYAqbG//PLL/e1vfzv29vbQWsPe3h729i6Di9ba1vGuiZ0g89R7x3q9Ru8dZ2dnuLi4wPn5Oc7Pzzf/e++4uLjYSst5ah1d3St5qIPOHLbLz+Vd8aP/uaxMRiPeVB5xDFnFj+8pH8xbpHO/jA+VR5bvbcjVT685XrL20Lo4imeudNJW4HmM/V0AfpPOPwngPzRMvB/A+wHg7W9/O77/+78fjx49wuHhIY6OjnB8fIz9/X2sVisr6KoBKsEtIVUKFS6Xf3Fxgddffx2np6f49Kc/jc9+9rN47bXX8Oqrr+L111/Hq6++itPTU3zmM5/B2dkZTk5OcHZ2tjF8ABsHt7+/vzmqA9zf39+ca50in3AojmfOj/ONvPf29nBwcGCdbmttk+bo6GjrWTZIJ/cZAwxZBP9nZ2dYr9d49uwZzs7OcHp6itPT00264Ke1hqOjIxwcHGx0JuSxXq9xcnKCi4sLnJ6ebhwxO12mkMv+/j4ODw83dd3f39/UPdMRPneGGTIK579erzd1vLi42OJF24PbPmSlIKL14Db+xCc+gYyex9inqPf+CoBXAOA973lPX61WODw83PxWq9VGwIocwE10ucpzc155PNcQA17L8kPYp6enODk5wdOnT/HGG2/g9ddf3xyfPn26uR9oH3mqAce5GjvfU0TtvW8pgvtxvcNYgve9vb3Ncb1ebykbG34o7fn5+eYel8vyDcNgx6R8KMUz6/UaBwcHWK/XG+NVh6BOkA1So4Ew/CxKYJ6c82PjqSINpx8uCoq6sePiPNUJc5tzXUKP1NijDaPNKh1/HmP/FIAvpfN3X11LKZSYf4wus8TKr/85zQwtGbNQT8sodHJysvl/dna2Ce2joVmJ2MjZ2NVLq6MJZQ1kGNXRKag6hcir9479/f0byh38K2KrI3IRmbaRSxNKCmDLmEMf+F6Ux07FRRIjGcRz/FP5B4V8NDJwsuC8g+egyFcd82y7szPTerj6OXoeY/8FAF/RWnsPLo38OwB8Z/VAaw2r1QoHBwebUMyFqlkIrXktNXKNBGb6dOplA9n1d3Z2tmXkqsAuTMuQPZ7LQmI2QBfG81HRKa5pOfyMypWP7CicoegYjObrjC/yODi4VEdGtChTjUIjnpn21/KiLUIXGSFVfuwUXX76U4p8w6FyG7rnRpEK6xc70jcF2Xvv5621/wrAPwKwD+B/7r3/y+qZ1hoODg42Bs9KPhP+uTD+zSbmLRBdkdwZfDROGPjBwcGmbxjRDCtYpjCZ4oUB8Dmn1751KEzUSaODGD/QujtFYvmPnBanz/JgQ2AD5i5QZUjM7+geOyiOLtkJu/A5G6vg/HksRgFInUiUUzkK5kPbntt1NjJ+rj577/0fAPgHS55x4YoKRr0+H/n/KHRxnn8JqadlBeAjN0Y8F4ofYxIxVuGM3TV2FcZxd8IhpqZTBVFFYaXj+854FPkdv1nUxXmPoio2ygwAMj703EWKWUSiMuTBNq0/58l1d3w4PqtBQG33LILT6K6iN32AjikUnPtlrhH1WuYMZkM2F94yOeei/6Pxub8eITtPvQHYjOy21nB8fLxBdTZ2RfSMsjDc3Qs+QzkjXI+uhzP+KD8Um9Et6qKk8psZexjlEbyo8bFD43ZwRhwDds7g47+G7/HTCI6nVHvvOD8/35KvUuVEtHxtK5aBi+ScMeu5k4nSTo096Lbht0ORF8VP5q2jHPX0mTcNhQsDWK1Wm58ie2YIXC6PsmbGzjzG9TDYcEDcV9RBHs2bjcYNTFVIrPJTp1IhehXRVQ5bAcBFSuxMXGTpHLtzjso/58+Dimx8jg8tz8lDZezuKw8V3YmxVyFHpvwzEUDk7dJxyDMybsdrePpATPc8z/3G+oEnT55s5oZXq9WNKRZXlp6PPLtDg5jXPTg4wMXFBQ4ODjZzzzyHrc+5gcJMxiHLmB7jvm/cq6YQs3zVCF3Uoo4lygUuHRs7KS4rHC0PEgdxyJ4huw7UaXfBhel8nQcwM+fLz4Sz17zcqPwIRHdu7C7kHDGpXp7zytK+KNR33r0qmxE9wnc2dkUVzssdszKrtIz8jNRBPMWm+QUi8XWHQnqPy9CQ1sk0kyHL0kVXLpJx/WIN6xlRs/EilWGG7BW687VIywauU4qZ09e6OOe3lHZq7C5cZFoa3leGrfccumc86rkL49gBrVYrHB8fb8L1vb09PH78GAcHB3j8+DEODw83fUTgJlpWTmTGMIJ05VYg+sXFBfb39zezBXEtZg0YrRg9nTwjnQuzVaYxXpAZluadRWqqN4Fs6phizIDRT43LzRxwnbP2zqKsyjGyY4kIQgcCRwas8nUOR8vKaOfIniHkLLpn95Z4u0zpluYV+UT4Hui9v7+/MfZHjx7dCN+VXLdAqXJonCaM4PT0dGMMEYLywBcbTYSqFYK5fqde5+hB0TRDML1WOWGdOgxj5uii974J1cPRcF3cwCHroo7LqOwzY+P7LJcsktCw3uXJ4zXcrVInx+VVdGdhvHrHpag+S05BHaq4sFj51Xx5VPfw8BDAtUGFsR8eHg5XCfIIrfLFlDlGrgMbQSh7HHlZciA3h/ncL1WFZ96d4jIPjMC83HW2jZ3sOWrhfNSJspzZuTnH5fh29x3fDAqaNusmuHq6NnVdi8iTu2jKz7019qAlCjBC90g3up8hjHpw5tcpQyjaarXavETByJ4t/ZyhapGECxdDEYLPCNE5jD89Pd3wx+vjT09Pb8wju0Uk2eIflSevDotjhm4ayrpQ1fHEU4XBG4fMMS6xWq1sXipP/Z/VUfUoM3S3WIrzz5Bcy+a2jtkVjmyYl9HCmp332d9scoas9x0fI96cUiuy8yIaXUswE4bztSXRjoaxvV+vRouogrsKgeK8KAjAxgm40WGVQ2YEQarUI0ethu7qxwbP5etqwEw+7tx1J2ccmhpiJRcHJM5YlTfOh1fn6VjEKAoJuhNkB+ZG1oPUK84agabl50d9Ln4+8+CHh4cbhOR+OS+cYQMbhZIjQx8pMytuIFsoRzik8/PzDd+r1WrzVtvZ2dkWn87wlOdA1Qy5VM6K8EE6Yq7lRX5sJIz+ka9GESybalaAFxCxIUX9srEMLo9/ugyay+DuiEaLbjCP89cQnqOxGXDY+Qq6KiTKaHR/Sflq8DPRRoYMuhorWzmWIVWm2EuMPZMLz88y+gHXxhWoH+/bRzTC/eysTzkTgbAxhhFlEY5zDlWe/D+cQPCsy38zebnyRsiuOutkkSGt64poXnGu/DLIADdX4t07YwewMQoVWmYQQWzwL8L4s1BeEYZ50/4rb7rB/S0+jhzKSKk4Hacf1UsXlYTBsUHHIF4gOysUr/9nZRr1C6M8t0ovDF4X7gTptBTXXZGV68mj7plTysYbNG2WjxpqhsiMxJxfPB/yjDEVbp/z8/ONEw5j5kVJETHyTIqLvjLaObIvGank5+LaizR4V1aUN9OtYGTX9NnzFaovNXZFjYxH129urW26HjEKv1qtNqvsQqmAmwttMlKnmCliHHU995I25To5B6flaYjvnJbyp+3C4TTnqWkyVGejd46L84iozA3acZdF86ho58YeA1gzCOHozTD4zJsD28sjud8dQg9D17AqK8cpUpVOr1UOIStfjUGRPo6BOtFnjKW13E+eJReqAtiSl/atsxDf1V8dhpOBe1b71xnKB68hJ5UjOxvWBUV2lkXII37ab9f6RlmRN6fRVZDOoSvt3Nh50wptpIzR0b3Ie0RLlFWVUJU37rHi6H3nueNY/TI+WLFcnbXBVeF54In55Xn5MPbe+9Y0T6y15/4xy1TPnbwqPmfaQ2UUz+m0oZM51z2MRTcRURRlYwbywUsdSGNE1jorurtR+XCGUSYPhLoBzll033mf3W3YANTG+qINebYMbRyeqtLGy4RdIfeoXiofNy6QNbrKVfPhUWceZIwXeSLE500b438mq6VUdX0UaStjz9pCHbbTNaeHWraGzSE/HghkQ2RUXyoLjQBm+GTnc++Q3Xk+lzajFxG+qwIwErDhhpHzTxtCedM6VGid3XcK5waZtGwN7fjIBg5co1wYNs+3h3ED2PTpXVg6kmucV6R88jNsSNxdUnSM+mkYr8bB6x44v9BNRU/lP855F524rtuszRigqz87VDcNx+fq7CvaObI7NFfFfh7SCs/ml6EVI3m1ZprzYRTIyl/CV/VbUh9nRIpMgfKsZG77MFf/zLDV4Bz/bBh8zOrkkDwbV8jkNHKOrp4jBz6iqh2166E8zOjSvUJ2XcfM95bQbNiYhYZ8zmnYu/beN7vS6I6xo8GVpdGHKrIzkCXIzv1Jh+w8L617v7V2/YIM/zh0ZUTlOmg+uvCGB5sc/9VLHioj7l4wPxmgVE6GZZKF/MwPRxiu/fjIz7D+z04bOmJ58rWK7mTzCqA29Awh3mzK+k36qxCdz13eS0gVlf/PoGzFY5anDjK58ZWl9WJHpSPwkUfleB0poo9k4Opa1UWjjEzW2u4KAI4nlbdzzMq/q5eTY0U7N/ZsJLnySlV4k6Vdkq8LB2PteOwNz18YUaPXRnN8zCpx5Bd8u5FeF/6F0vGPrwHXyOUcE5fBYXy8g+/2DdTQ2YWn2YKTTP5af+XRobKLsNS5qBxdpMRymtGjzOjddW5Pnb3JuonaJpUezESSd74H3YwBBy0xen5GPZ9TGB7kqQblMnTPnNcInWbyUMPJHKbW0SmAixa0DFbGzDC4rKqO/JzOP8/k4ZxAFdFoXZX3qj6Rn1t7z7LT//Fc5MXdiYr3uM6Gz0cdjKu6cerAHd3p2vgZchVY2h/W9Iri2v+LeeVAdt491vVTswU1GgrOhtoOifScn9c+ufKS9Uu5TFZWrhMP0DG6x9GtgsvqpjJxYbjyobJyBqxpq11utY6cN79g4paiOj64bHdPZcEIz0dn7C4acN2gWbqT0fg4jjxR5vVnwxbNS5VSUTzC9/X6eqcX/pyTzrG7ENDxrWH1jGwc0uqCDTb2MEy34IWVNuPblcnGrr/Ih6fsRm2SdTfU6NlQdIrWIbU+p2GvM3I1RuVDp1hDTmykmcNx52qkzkmpsUe5o2hE9SejOxuge7OoMihVLjbwMGb96qa+9+36ukt4yO7PIFemJEw8b5zlyw4hU1JFeGfs/G61yoVlzkcuW5VbjVadkdaDeeOwl8cWXJdnFHFlUUfWllxG1jaz+qB1z/ThNnRnyF7RCPFnysjCakZ0/r46h+2M7PwlVjcYpS/BKJq6iCLjmfPlga1sdNyVqUs6ncHwmm+WiSIjcL31Mn8embe7YmOolsZy3RRJdQqNjdnJJfI9ODi44dgyw1C9c4au0R5PsUbduF0YdV1bqgwyp60OiaMZlsfzGDpwz5CdBTLyvkqjNOqtNYQPo+Z59Gx5rJbpGjfuc704rT6nDa0KUDW0M2jN1yG7lu9QnsN5DukDUdnoI61rhwo92UllzlJ5DH547UBlQI6qbge3U+a0XXTGvDoQiP8sV+ckM31WHVJ+K7pzYx8h3YhUYTMBOUQ/OTnZLJg5OTm5gejaP+e8ojyHsHzuDNrxr8bm+mmzeTNyLkF2VuwwlljeHO+9B/GOtAC20NCFv85JKrJH+2i3gOWTTf9pOqUZw1Mnom3LclHeXB/eAYVrMz3qKDzzrDqegYejOzP22VA9Q/fbhDOsXKGc+q027rtXhp3x4SKSEXo4g9dfVobykJVZIXtVrg5IOWSPkF/lNUJ5NnCH7o5/NojMKXD5XJ6Wr7Jz5Syh6lnX968MP8h12Wb1S+nOPv+klFW2oiVptT+mn15+9uzZVp+dPbnz5hVPo2glM3K+pmWy4joFHfHE+fFAFofBDtn5ZRkuh1+WiWPsdsMviWRhbxbCj0BAQ3Z+3tFsGSwjN4jI6dgR6jgKy4j7/PxhSEf8XOTFL43pa7gcjcwa/J1tOMlUKa+r0GxUEGlVsXQUPlbJRRjPCsFfcVnqiBzvGaKO5MD1rtAsu5ahdRyDR03HSs0bJoRRR4gfYSvLOWsDvTYa8XZ1YmPPoogsqhhFDro9tXOCOv2nPGn9qqiFedP8gjiKCVm7KKaiO++zZ2jH910YvyR/hzLZSrkI3+NZRj6mGZ6cofM9Z+iZo9OyqzIBHxlUhqwKyuF+nMcxjJ4/kghsoyDPUeu9DO0zY+a0WditsuNntP/NPHAZHDVEPXl7Li3TjfpzFKGj+q7OSlwX5tt9HMMBR0V3auwVEimpUlThkDMSVRodiecvnEY+7OUdLzMhmbuuCDpb78pY1Ji5X65y5lFsRis2duWRDYANmgfv3IKjCsFHCKtp1fGOFJ2f0UiDjxq1aV3dqkNdtMP5hRxmZnRc/eOoTsoBl4sqMrrztfHuPHtmZGSO1FiycE8bhAed+FlXvmuQyuBn6lvVw/GTGb0rj1EtjorwbjrLGb+ifrYOwPGa6UHFczZar3kEsWNzU46cPxs185gZU8ajRknuV/E9w6OuWrzXyO5QXf8Dy0btI/3Iw3HeLpyPPEKpVMFGjaWLIrK0M2iu55VCcf0j/8zoWZkZrTUy4P4rywW4uf98LHLhL7cyqusUWRzVsTBaaf0YpVWObvBO5cH6of+dzJk/NXR1AOyQFDx4ilLzD8qmW/l5JW3rLB1wD1bQ3QbllpIzOA0v+af3nUEpZQasBjZr4O66lu+OqsQaqjtSw+PnWOlmEF6VtWpfh2YqxyyiqcBCy2U+2VHpsyOAYMOu0LeKuqpyMllr/vy/ioaU7nyA7nmpMh7309VzPCIf50H6dRTtv1eGPBP2Kb/ZtRklcejCR2dIyi/vosIbTnI+rLz6PbkYzIoPSaqzUJ7ZCAE/p6x11mW96qgUcbldWEZuhNw5VpVxZuDMrzpdRWmnF9z94TcLHe/Bv8qnigCACWNvrX0pgP8FwDsAdACv9N4/2Fp7G4AfA/BlAH4DwLf33n9/lJ+SCrBCvyV5umtOYO4XafRahd6Zl608c3ZNG3Bk7I40rFf+gJthb1zjcN0ZkRopK6tD9RmDd7LSejB/1So75ivaTmccgtwsgZY9QzPtxE4o0xU1fE6voOWizopmkP0cwF/tvf+z1trLAD7aWvvHAP48gA/33n+gtfYBAB8A8L2jzNSzOS83QrARmvN/5/l4X7l4hZXfcgvivpq+1hn8OFSNe6O6u5CP02oap9yaPitHr4chZHk41NURfpcmpqui/x6fjOaywlhZjk5WjFRaB+aR/2f1UZkx71k7ZEenoy6N0xM2WuXDza+rPNTgXdczo6Gx994/DeDTV/8/21r7GIB3AfhWAF9/lexHAPw8Jow9mNLwRMMcpsw5uIZVI+H/8YuBuDBy3nKKV39pSOhQKNK6KbqlIZ+rB4fjGZKNqFKAMDyXLxtDhsKs1Px1GV2dl81V8/lIRsrbyAFyHno/iygqB8xp1OAdYLH8tF6cTxYtZcSGrhttPJexM7XWvgzAVwH4CIB3XDkCAPhtXIb57pn3A3g/AHzJl3zJFsOUpkT6jEbeNf6rkeuuqYH06/V68+liRh8OcTMkcUpVIW80mCLYaIDQlTEKhTPDqmTr6snXMtRklAqUZ2RXw1X5ZPy7ts7QjtM4Q8j0w7WL8qVyHEUPkYdGpXw9k3NVT2fYlaEDC4y9tfYSgL8P4C/33v9QGOqtNVtS7/0VAK8AwHvf+95eNbQpc1gBfj5DyDB0Xhp7cnJy4xfG3vv16jAuQz2vewFE66Z8OqXSt77U+J0icznuG2O6JLaKArK2YON1iujSs6ED12/NAdiqp8ojnnflZ0aqDpLfzHOGm/GtebptqdwSY5Ufy8ZNv1YGmxm6Puf66iNED5oy9tbaCpeG/qO995+6uvw7rbV39t4/3Vp7J4BXZ/IaEXu72fR8zNKwMbllsnwNuFYc/uQRD/ZUZWeKxLxo4+ncLI+CZ0rLChUGqWgRxxEy87NxXNIOnA+Hr2wcUQ8Xgmf86DUnw8jXrZQbGYE6FFcG88cy4vSOz+y+q2fmTJlmDLqimdH4BuCHAXys9/636dbPAHgfgB+4Ov70bRiYaWRNWymJyyeMJ1D92bNnGzQ/PT3d/BjZ+UuzrKS8yCSObpBHB790vXTvfbMOXx0Op3HviEf9wsADPVer1daLO4Bf1lnJM/IOylbHuV+kjfC9tYbDw8ON49rbu/48dBaxVF0OlW/wzgbLkYwzdpe3e8ZNy810l7QO7Dz4mEUcWd5O51XuI2cwg+xfB+C7APzz1tovX137b3Fp5D/eWvtuAP8WwLdP5PXCqELPIG48NjRnXIqsIXC9ns1vxpGfdYifoZKiOyM8nzPpa4/KW4bsM3Jdiuxalluc4xyOk9Msj+6oo/4jfpnnrJviunNVflkZ2b0qSsh4d05n9PzMaPw/AZDl8o1T3F3ndSMMCppRgqyR43nnRcNg4v117rPztBtvR8X8xLUI6RmJNCRV49WQUPvo2n2Iqb9A/ThXBxD14x1n+KOEKms2wEyeKktHasicN5+7Prs6LnVwTDoF5bofISMdvc7GF0aIzFOBwb+LOJyTYFlmjmbWQYzSxpH3wGPdqwz+zt5nX4Ia8VwWdsa1TCkcqmcfgVDEViMO49IwUcN4ra8LD/laZgSZgUSe4YRa294LjmkJsmfkDK8yKGdAzkFUMtM8KySPdomyKkPPUFYRPgvLK1lWRq71qkL46jlNf++NPWgm5Ir7twn3Ah15NF4/+pCFvjP93MwhsEIzL1XYXk35cFnsdALhgO3xA11yyejrZK/Kp3xkUdSojZwBOSPSfDQayfhbUpbWm8+1zCV5ZyBU1VNlW8mFz13/PHOaSjs39kCeTPGUKq9fPRvpAh11Xl1RXY3LrWjKDDwz9CzCyAzdDcRp/bRMNYj4f35+vlnbH/e4bpyfQ08tj/moDF2Vbtb4MgeXzXywrJW4LBflMM8qG/dshvxKHHG55/haZpjOkfA1HfiNNKx/Ge3U2GeZcs/x0d1X5HIhs44AZ4ocDe1GoKt6ZaGZcwx6nckZpkNZ5wQ4pAewNRKeGZtSZkDZvcpBuHqpMYWRqKPlbkDk515GirxmHIvjuZJ/5gjiXPWL618ZO5fl/vM1p9+aZsam7gTZZzw908jY1cgYObO5dO2j60sfMbDEH2lQ3jJkV76ZJ3dUp6NRASty1qhs6L1vr/7jbaT0e+mV8lXtkqFL5ti4ndWJqiNjmTvjjvl0fm8+5KTtmPHpeNZyVDczhx9t6T4m4vLKnF5GTu6une6lsVeVq9AxI0V0vl41KJfJA0mskJXizJJDZefAtFE5ZNU07jnniLh/r452pGSc/wiBuFx9Vh0ZGzJ36zRkzZyR4z8DiMw583lWX0VnV6fKyblugk5Jui5tJs8RTyMd3XkYzzvBZI16G6PSRnNTVepdGbGBy4bgd7PV0F0oXCkm81UpBCuuNiSH5xytZGF9PBdp4iWfSMfTgjx4x2WpjOI85BX5OXIy4Hx0t95AcuZNHS3XMZtaY3lzPdhRZsY4AwiOtG14kRTnrVENgK131YO0rvo/8uC0/Oy9Q3ZGG4csigT6rCP13u6YPaeGn60Um+3zjgx+VA81+iwcBPwqLDVcRnhGeR1xrhTFOeOZ9I53Vfj1er31+nDm/KJO6mhVRlx3JQ2xZ3RkRE7OzplzXbI2cHk7ui0w7tTY1+vLfdmjTxZe3KElsN24ce4oa2x3PSjK4e+MO2R3Bu8cw8wgnuOBG9whGstOF+Vo359lpnLjxSuhbGpYXFZQyGY2slF5cX15DCEWAnGkEgiZyc2Fx5y/1i/K5LwqR1AZv6Kt48uNxXCbaNTk8ol6aN2C4vpsF4xp58juvozBiqTXR4J35Bo1E5wqp75Bpukc6mehfOVsmHTlFstD68SIkA34ZTIJA9C+u0sb9XZy4nuZLDncDOVkp8bOi/utDhm5/i6SqZB0BvU5XWbUmt7x5H6ZbB3Ps/ri7ITbIKOdI/vTp09xeHi4tZMJKwKHcs8TXnGZ+m03/qabE74LWRXZRsbu8uXrmreWwaPm/BwbO78Z51bXATe/Bqt5sbz5/ozsZ1HdGScPyrGx64pBlaFz4lwfPjKPGf/Knysvrrm2VTlwvvwcOz+NEp3sK/3PyhvZzJ0YO6OQjkZWnmmJ8TMSxuCJWz2XLVIIcoiehfPO2Cv+NXJgZNcpMlawkJ0ObLlVeK5uXOfKGFjZHcqrjKIOmRyjzuoIdcuvGOhi1HbjE1xGlq4yDOYtiyTY4WTRENc9nLGuHagcu2tjlZ/aSPa/op0b+8nJydZrorwTKXCznx60FOW5sfTLL/w7Pz+3Xpcby/GTCdg12AgpnRKpd68iA72uBq38OpR3pHUY1cdFOM7ImWfONxvHcfy4fDL0VxlUVJUZvGX15YjGTSeGHBQ8sqjKRYSZ01WH4Winxn5xcYHPfOYzWK/XOD4+BnDzfWlHTtGZMi8diH5ycoJnz57hjTfewMnJCd54443Nm29h7IGqwVMlRG24rMEqJXTIqM9rndUhcl9fkU2n5zIHqnVzyFkZUabwnL/mo2UxGnMXJXMwmUN1YxfOCDPn7RwmT1tyWp2S5Hq2tj324NCcvzOvMhotutJj5uigMlUAACAASURBVPSVdj7PnoXSXKkQ/IwnHpWnyM57zgUv4Wh671sfKWQlcUquZen5DKrrs5VX13TqlPQYSsc8L4maKiOrkN21nbsWad0CE46uMhk4Hl15rFcZL8xPJoeqHnpUkHBOMYuyuE46BqP5VfVR2jmy/+Ef/uEmfA8PF3OtPCWnnsoZhAtTQ0gctgeyP336FM+ePcPnPve5zbWYCoztj4HrBQ/qlSsE5OvZTxuOnwn++cUVfZZH0ff29jbfTGejzwbGWE5OOTM5K9pk6Mx5sYPk+xVfcU1nGrKR+krmzI8rJ7vnIkSdwchQVuumIKHIrrM9XDceoMzknjmOe4XssWnEwcHBBt15LneEhCOU5HRh9IrsvGd88BL58rZJ1QKJqszsN8OvXtPBN0Y91++rDLmizJll9XD8Zt2dqszIyxmDQz/mLTvX/JfyoZQ5qUzmztnF/QzVI/9ZoBhFCY52juyf/exnN9sLR6MeHh5u0DX6zlUfvqIQjI6+838dqOOXY1arFdbrNQ4PD9FauzGAGGXMIn3cywxG72t/TaeiWD4REa1Wqy2ZqUIwOceQ8V9FLxmpk9FQluXHzzhU673fmJVwawm0O6A8anjteMjq4K7rtSyqyhyiRq2avppdYYo6z9LOR+PfeOMNHB8fo7XLVWqHh4fovePo6Ai9d6xWKwDbyhLkKuu8I/fTeW5dz8PouX8Y+8ZHxDFC6BHiZwjJ17hxdW883ZYqVvnpd9aCMrRR+WWKP4Pgrn5alpLrv7pnou76kswsOju9qXgfpVfDHDkCJ8dRvasoahQdqvyrttp5GP/s2TO8/vrrW332s7MzHBwcbPqgq9UKvV8vuNGFIRlyOSXNPCuHjTPLXV24lIVQqtiKcC5E40U/vffNRytOT0+3lseysfPYh+NzVHc1EOfInBzYAEdomRmUU/woj78oo5+Tztomu18pv6arUFjryFOzPPtQ8aflsdxHz6oO8b3ZUP5OwvjeOw4PDzch9aNHjwAAh4eHmyM3vIYqI/Sqpi4ifTgaDn9H3jrzzM5weKBJGzUUhdeExy+6FbG19bNnz7bQncN2AFuzB1EuK6AjdnQhMydHfYbHVjiPqA+Hsa5MNvhMOaNdeIVl1IunwTR8du0VaV204o6zEVLIIFvLoP+ZZ72mPKistK6ZYx+1OXBH77OfnZ2htbYZEQeAN954A+fn5zg6OtoodbbIYiRUpQrNeS06G74ifjZgNPqvZTt01z6aIj0bezwfqB/jDMF/dD8ULTNSnpbKUrsAnKe7Xhk7tw3Xxa1czOTM1zID03MnKxf9cPqIsgD/rfdRWU6ueq4OzaF6NZCpdCfz7AA2C1pOTk5wdHS0OYbBn52d4cmTJ5u+PE81qfFlfUxOGygeqKGj2xEex4cWDg8PsVqtNr94Vh2GelU1Ht5phZFAw3c27ED2i4uLjZz4yzRRn/Pz8w3yHR0dbW1BxbLKjGwWCYPUYfIUpZN/nGdopIis01W9X4/hRH2A7ZepXN1cO4TsOTpQRFfZ6HJklQvvlJPJ0507uagcwpmoI1KeVQcrg985sgPXIefZ2dnGGE5OTtB73yD96ekpDg8PN4rLITdQe8msX6OIzYNA6hD4XL3nTGMuue/674zkfM7yC6UPOYbDYMfkUDArP5Ohq09lYGxo/F+fd8bh2sm1VWaoeo0dwqiO7nnX3k426ng4zyziUVI5jNIoj/cK2cNbAtcoH/3U8OJ7e3ubUP709BSPHz/Ger3eINj+/v7GCWgFHcJwIwSqP3r0aIPiipZx79GjR1vortFE5klnkDH+OyPPxhr0eoS4Ydz8PPdzeZDTTQ3F/4yyhUBR/zBA7ss7gw/KukauDQFsZkuinmz4/NYf85gZI/9itsM5C+ZZDcmt9qye0ahFZxecgVYDfhnC3ztjB65DlBA4h54XFxd4+vQp1us1jo6ONo0boTw3rvPUQdn1UJQY1GqtbSlpDH5xOM8r67jhgmZQe0TOQbhwmB1DIHs4zeCfv0LLkVAWSmr+7r7ypYjplNvVW50d88Z5KqKzA+NxCza4zNBVT1TWbHhVWF+1GZelzzkHOIoMM/SuHPRID4E7CuOBm8gbg01Pnz7FxcXFpq8GYNOHDsNjh6FGEP1bXjDDo8X7+/s4OjraKA8rWEQN+/v7OD4+3jgGt8Mp1wO4iS4ZSmtjZQbG0QP/2DlFeh3Qa+26b+r2OtO6qOFyHdUYVKEZ3TPEcsbp5OkMPhytzjpoZMT3nDyjfSqZu0hE61mBjMtTz2dC+Wz2ievB9WE+K7oTY1flj77nxcXFZlSe0TTCbR6pD2egxh7z0zyoxdNSYexuCinGB/b2rlf18Tx/1oD8X9G3MvqR0ihv8V+VRRcR8fx7VVZmZBreqhGwcwBuDqzpQOSojsoD15unGMOJRZ66jVVWTxcxjSKZII4+ePpP043QWyMjV3+9pv85L+2SzdDOjV09ZQg9hBhh6LNnz24Y5/Hx8dby1ejLA9dr2fkDjWzoEeZqWKUGFZFDHJ3HdGFtdl+vz8glGhPA1hhH1JPv6RRh9qumaFThmA9FUU4XaThy0rqq0bv0Fcq5CGI03ZSF5xo1VUbFesLjBFWk5OTojFl5HdXfpXXdhxHtfIAujE5DrzDSp0+fbpTg9PR00y+NsDoW3BweHm4G0NhTR3cgvtDKW1dH2iA3bRFH16guFM8M2KG4QxRVZP4yKy++iRH2cIzMa3Q1eOaAj/pTZc+UqEJDPvJOM5qvkxUjczZdpM4vnB6vOcimO8OhuK6Eyt2V6eqhfEeUoc/zsbo+G9WpQ9IPY7hyMto5soeHjEZhBAlF7r1vRmHjzbTWLhfhrNfrzUo7Je67av/NhcAODTNjWEJZ+izUyxDdhYScRo2dB7WcgfMxyuV8NeLRNFyP7KiyGyngEqUf/ZwjyvJUHtSoqjpWo+Uj1HflOkdUXVtSLtNOjX1vbw+PHj3ammcPbxWGGoa9Xl9uYRVvq8WikcPDQ5ydneHo6AiPHj3C0dHRVgOEs4g+exgHN6COAge5hlWjdOGoaxh3HrzwGvdAoPgfyMXbLOs8u4bIYdwR7cSAJi+tZYeQRSw80Kb1YgfN97Q7w+MFbtxCyw2ZZuFs5BnIztOMuiknt8tMm2TX1Pm7uka0pXLS+i35r+RmK7Lwfca57jyMDyV2ygVcKx734U9OTgBcL7w5OTlBa9tb+4Tg3Ry1GnnWF1des+uzYZlGBhmi63UecVWjcRRy4HCe56Qdwju+FVFYyUZOkY8cQY3kzIY+a6Au+lJZZ+0x0+aOb64br25zDrCiJWmzrkhm8CPaqbHv7+/jC7/wCzcj5qF0/E65fjAgKhOLKmK0PPrzfJ0bSKfbmLSv5hqsmtrQMLhqcEUtRXlNy2sQWA4O/VT5GcFj4VEgvC73Zd44aqjqwDxkC1mizdixxlFRXA3Ulc91BLAZH+AxDAYMl+cMqitl6KsOOmt7vlZFQ65MrUvWrXDgUdGdhPGxrptf6YwG49c5W2ubeePeL0foLy4u8OzZs80zPBfrpsm4cYIyRHbCz0Jy4FrxZoyd/7twTJWHjZ4NSxue68dIHk5R0T6eU17Y8biVZRxlhKy1fi4MDr4CEZeShtOzA47aBhoSV+n5Ob3v2soZsUYqem2JE1CgWRq+B+3c2J88eYKDg4PNFFvMsYfB8tbOKlCeXguHEelDkZxxZlM8lRGqQcbzWUibNZRb8aWemxuQEZKdnyKiC2fZAGJemt+I4/TsUJRvRQ13ZAfEztkZJkdxTq6Vg2R+eACS33nXboN2DTUfF6YrHyMD5PbLdGAGCLJ0rv7cBgxsTi8dTRt7a20fwC8C+FTv/Vtaa+8B8CEAbwfwUQDf1Xs/rfLY29vDSy+9dMPYW2ubfnlcV2HEYFWgvRp7rIhzjakNWxk3G1tc58bUPNWDcxn6XxtRp8ciX07HkY4+x2G5Gj+P1CsKBjlj1/qoUbPTcXWKZ9iRsex0oE/z0PLjnMP4OHJIH2nY8TBlctLyK+PkvLL/DtU5XRx5XEZl69oknnNdUgWPjJZs8vY9AD5G538TwA/23v8IgN8H8N2jDKIPya+Oxhp0RZ5g3oWqVQiXUdaYKmBOy+eVl87C14wvh8T8c2/e6Vy6vrKr4bqbd9cpuIqPbIAvC6Ndnpr/SC5ZO414rfiaaQ/HQ6UjWdvH+YxOOD4qfXRljyIHpSlkb629G8B/BuB/APBft0tOvwHAd14l+REAfx3AD1X5RBi/t7e3tawVuNy8ArjuBwfzqvD8yxaSzAqzWrfuBJctI1UnxXnoslxOk62Ai+ejTOUv0vE35F2fNMqIo5bPSMTrvjmSyN5BUATjbpTyoHy7JcTaTkpRT14+G2XG4CyP72g7zhh7pQcOzV30wMcsP/d+Q2Xoqld6HBl50GwY/3cA/DUAL1+dvx3AH/Tez6/OPwngXe7B1tr7AbwfAN72trdtjLb3fsNweUrj6tkh8jlDy6gy7MxzSl2GoR03YMXXyONHWRGmOcWtnJxTbs1b03JZmdyijfioz2b14R/XJUMm5cXlpZEL8+XaZ5YcL1VbZs85x5D9z5zcEqoMf2jsrbVvAfBq7/2jrbWvX1TyZeGvAHgFAL78y7+8xwAd78++t7eHN954A/v7+zg9Pd3qh/CuMY8ePcJqtcLx8TGOjo5wdHR0YxeZaPwqbOdlqKOwiYlHpDNDZd4zxb+Sqw2xRxEJ56V1dooViK9OlPngYyYrRuP4hSzifuQT4yvaFnt7N18kqRBK5RhHdvq9X++DwIO78ZwuadWIROXr+KoMLkN3rUNcZ4fE7al1dM7YOXZtt4pmkP3rAPyZ1to3AzgG8AUAPgjgi1prB1fo/m4Anxpl1FrbWjWm/fZYFcXhneuHagg/E74DfiMGZ+z63Aw5JBul1efc89z4znlkz2t5M7yzgs0gL7+YovfVWXAUUPHq8lFeWQ5ubIBRvjKGJddUPkrqQPi65jtyuq6c2bauaGjsvffvA/B9VwV+PYD/pvf+51prPwHg23A5Iv8+AD89yqu1hsePH28+EhHKsL+/j2fPnm3WvvP2ybGRxOHhIR4/fozVaoWjo6MN2ut208K7/bk+Y7ZIhBtBR5a5XupwNOQeOSItI0vL+XBk4J4ZGZPeZxmqsavxRPq4p3VhZ52hJkcDMwivdeMIB7gen9Axhaz8ESqyg50N45XnDEicXnBUyPmyfukx7mflMD3PPPv3AvhQa+1vAPglAD88eiBG4yOc40U1sdgmvr0Wi244jI9NJ9jQ3cATkzPyyvDjmaCsITKU5fBMjXOGFPHdfeapSsv53aZMrodDyRljj2cjjYawzEPWBnyubcMy56k4NvhIq3moQ6uMXtPMRCbqZBy5aE6di0YvDjyyyIJpkbH33n8ewM9f/f81AF+z5PnWrhd5xEh8GPtLL720efFltVptXoCJpbDZTq/E21ZZFXq7/mf2oodDalWimfBZ81HU0OecEvA9/a/pnJJl6KQOS+t6G3LyUWfolFuReEm+cYwfRyQsgyx607ZhndG2yHh0+VeOyrVnpl+ZTlT8MO38RZh4e42Z29/f32yb3HvffDs9XnMNgz8+Pt7aH45DGWBbqByWu7ev1Nh18Yoae4SIjPBZY8X/DNkdUmh+2aBb1tgZZQg5Gw1k9VHHoPzzczpoGT/tx2d8ZjxqZBFtxBucRDreQjp710DL13cGtJ78vKsf5+WiRecAM4N1XTVXdkV3si0Vhyc86LZerzcbTPDrqTqfDtwUoBM2N5hen/kFr86ru/lv5oPTO8r4reSVHTODnlUCVVzOx+WV8ZvJUOsR/7NfZYQV4mUor2sIlFfHn6s3G6NbpVfpg0Yvzhkyn1XdtUwtP6M7Mfao3MHBAY6OjgBg8723+ITy/v7+5n13bjhWcN5nLcgpXIXskU+G8EGBFDGgyOsClDK0y9Ipv6q4nE9Fmp9eY9L8Wtv+yonmVXV9nENVYgcf93VakJ9zy5U1r6AY8I1FRtz318U+cS3aOpMHp1N5cptzG2m6im9F897rwcwgzrdyXI7u5PNPSmzMjOTcUDpSrF4wq3i2So7z4qNez8rMDMoZZIbEWaPOGKZLp9dG+er1Gcfp6j767+SRoXmG8M5wHJorsmfOduSUKmeZ3ZvVh6zM+D/bbktp5x92/NznPrc1Mst7x7d2/bFFNnY13CyE4qO77tKocumz7loMKlYDe5o3DxRpKBrPRX5utJqfczQyxkyxmRdXpqI3b+QZcuBoaORcXdSi00jcHtXz8Qt9CGRn5xX3gleVGf/X8RhGXk4X+Y7aRPnm/wocTgfdoi/WFwd2Fe38W2+xQi4YZQUJct5YBaP58nHEQ0ZZw0W5ulRUldrxq4rpyqkM8rY0Y+xBvOTV5cPtpAad/XcO0Dk597/qsqizCCMFtqcCdVRe83UGrHxquU62mpdLn+mF0xnmgwHClbFUX3Zq7Ofn5/jd3/3drUUw4eVitxreOw7YHgxTyhRaFUgRRI2O90wDbr7WGcbtwvFMsZmiTB4hZsqWys42ppMDR0N8T/ni8pk3dsbO2Hl8g5G9ci7OqFTxq7DbGVB0+xjZWebhyHg2ZRRpOcegfX+9P4vwFVWRR+aAnM5ntHNj/73f+71N4/A7yaEoseEiK6sLZ1SxVVBu3pWNPihzEHGPuwwq0ArZg1gp4xkN12YQbUSZ49PrWmcmncoEtjfR0DBejVy7NI4/JWdoLCu3OIZ55IFb173jL+NEPhydOV7clK4uqnL1WxLSu3aZfYZ51TQVDzvvs3/mM5/ZMnb9jlo0ju6qygjCP6dg0aDqUNQBAN7Tq/JUQtZw0P13z+p/VtrnNXzloXIAHNEoX0C+AMkZeIXqylN2rvJ219RZ8Qq6qE+M+egGF/zZLy0/k706fZXVqJ3YEbj6Vw4yixhH+uVop8Z+dnaG3/qt39raGIE/t6ShGIBNaO8GhHQzRuC6wrzjKi+r5Z0+Ig0LNJTEOZtwGJFWHU5cD3KKlF1Xg3f33fP6fyYCcsYEXPfdNbKJiCtk7tphZOxZ12fEk1IYTvxU7uywOX3oGX9gI+rhnIcau37XPXPcGb/aTm7aktMvpZlndmrs6/XlF1/4E0vhfRmN3eIZFhArma7GA66RQMN19vKRLo7xTJRfeWPOM2ukCp0UERyKZ8g+E/Jl/Dl0cEjK5zoAV82tV4iZOaWM3+xZ5/x0UYo6zwAWduY8eJflz/Xg81lDVzlzfpXD0zKXlFXRzvvsr7322mZ7Y/6EU3zMsfe+9SYbK1ogfHziib/SqgMojOyxnj6WUYaD4W2WwzFEtyKUgze0jHfw2fGol84QB7i55NH1V52Rjww8Q1P3fOUAnLPSsRTd7jvScH6KllX0UUUDLBvOT+UUPOrinNCl1i6n3rivHm1ezQapM4wyZqKQzJGxXF3kqHm6rsJtjX7nyH56erknpaJrhPGVx2fj4rfmnLEzsse526AwKBNu1odTXkeRgEPxCtm1rOzeiyR1UOrUKgOt0FrPswhg5LQAb+x8DHnpj/v1US9dLejK0nN1ALcxvNs4vHiOeVhKO59njz3fw+OGh42dTQCPdFHJQNjz8/PNxxsVdYHrPdcY2WOO/+joaAvZI83e3t6Nj024F2RGv4ycklYePGtcNRD+XxldZWh8ro41kD3QXEfhXRmj+eHIi/PJxj5UPm6qkp135M8j8KEDupiF9+Xnkf8sMnF1GZFznjPIPgrjlxr8zo09XlsFrtc0u343kBuAIjsPGsUvugGRLroIjO5xfbVabY0dsEJVc/yVsVcNkSF6Fg7OKFx2/zbXnQFmylk5m4qvKmJg/lQ+I2TXRU/hAHiAltuf5TsTWYxkWDnl5wWK2yJ60M7Xxo8q51DFKaAaPIfzLBRefhsbZwSyx6uy/J58vHyjswLq9bm/xx/5Gxlw1EVDs/jP6YCbG0RoPvx/NGA2Ui4na535mDV2RStXlnMoDuFYDkx6j8djogxn7MD2a7COfy2jSpfxpfUdETs4J6/M4GcdwL0xdhaqDnSpEqmRn5+fb4Xz3Ag83x5TcLEbDht7OIII9SOsd1NhoSQxjRMoAdz8jpzSyMiVQhZunjdzhCNj1xDbOY0ZY1fe9fng39WR+eGBPr6edXn03K1pZ/6ijcKJA9fTrpyPysaF0pnzZMoiUn7O5ZmdV2MEmUwc3ckrrkFOKEGsNEsWc1TIBlwbj+5XBlx7f12kwQbMSBvoHyihHtkN6DAvmZFrw1aK54yd6+WcZSYbl7YK47P6jNqA07i6VAbB91wYnyk833Nvw6lhVmg9Ms5I71A449sBX1YX1SnWkwo4dm7szjPxT70rf901UNyhjE6F6ThAGHgspggUD2SPXXH066e6si2MXFfZ8aaH6iC4cUYhV6Y4TnaZcQL5dJhzDBn69349MMfdpAqt4+jKyBxE5YhUPxTF2SCir+6MRJ/nOXg1Fu6ycfSQ8a11Gzkd1gNXhra7k7GL9EZ0ZzvVVJ4VuKk4WXjKeaoSqAfkxgS25zp1zliV1OUZz0UeuhhH68MhpuuXcb2zZ/V6ZSyKqhWyz6St5K9tluU5Ssu/LAwOcgat6TKkUx1U49RyKsRcShp6uzKqqNfxOEN30mcHPKID2zuLANhCdP6Oe6TXkDvK0L6blhnpwmDZUHUNvXph7u/FwF70BWPUn+sQ/1kBqxCNj5UMM9R2+Tin4NCf5dJ7/gKM41ERsnIuWh9Oy9cqlOQ2yRwJ58XpuX257Vlv+Jp7SWgJ6bMxwKvRBqe9jX5UenOne9BVHpWNPxQhCyE1RHPK5cp0ysmKx2G68h6GENN72egzp1fK0Mulr1Bx1A/XOmYhvrtWjZhX6Ffl6+rBaZxsZul5kd0BBuddOZ1RWZpnlZf+z/RhCbrfSZ+dPaUzOvbwgejcV498+C0nXh+dKSP3jdSz89SMG4VnTx/nbOz8n9+qq5Qla+zM6NkB6tHVWZ9TYx+F9GrkPH7C5asj5bxGTsIZeiaTCumCMuPlY7Q3fz4q2h7Alp5l5VftWKVz+bh6aXivkepoLMTRnY3GZ+GRKgYvnNF5dA572NgjH/ag1U8H4LLQUY1CR6h1NZY6Mj7OkEPKyDuOLiqp8nKGN2vsOiiqCJUZe8ar443z0/8jWalz12PoDXfH1uv11uAqp8lkWBm3K1vvuaiB72e8a11n2p3pTpA9GzFVhVZjZ3SNZ2d25FRD5nfZ+b5zAMybosOoK5DV7TYG74zRGVKVz8hZZPVwU3CVsbt8dLbE1S3IhdNVffg51g8db+H+OUd0PN4SkWKUwd24yhDZgCsHNWp7lmnlFFgGle4z7dzY3eAXCyeEy5saxoIZXbiijelIPTngv1WePetQyilTNZ2lz2dTV6OwfgbZMxo5C77H/3n9ukZXM8ieRQjO+UU+3A3KeHfyYzmyUceRB8R49Vx0A/nFqrjuBiNV52aNbUSK3pWTc2WOyt75F2EyRNdzhyqaTxZy6392MGz0bglqxY+rh6L/CNkrhZ1BhBEKZ8/x/2x+faacSk6j+irCVzyrsivKuTQzpOjMusR9+axPnOWp/Ls6jdLwvQq83PmMLO4M2Z2AOGTiwSCeagOulzryYNrojbkw8sxJKNplyhoU+UW53G8PJOH8NRTmMpSWInzl3TMjdgiv/LCz1YE2V5Yrj7cZq5DdhciZwwk94YigkiWnC+LFObGJyuHh4Vb7uaiS89U8R868khn/gmd9BZtDex2jmulC3OlyWQC2koD/GgdwM2TUlVDZj9NUxs6KMzJ257AyxHL3smlEp5wOoTnfLM2MsWf5VKi2BL1GZXOaCqkyQ8pknfHp0FFDf6dPnG+FvKrLGR963dXL8ersYYbu7EUYHd1Vw+aBuSAOw3lwhafOtJHcSLv29dUAs3M2Ts6H0SaW4jpFcXPWkZ/SCN3dc5niVAgZddRn4pihvobUjs+szplDrPrAjOaZM9VxAV3/oPyyXgDY6BJHmL33relGLkvRXXUqaxOtm/533ZesrApglO4E2Z2HV7Sr+uoO1V1/vDL2yI/5Yf70OPLSrGTBP5eVOY1M6auGGzVqZtx6XimfOyp/GZ8u3xHa6rNZ/lV+6shmjUB1I4sWszrMoOuIB03LebookrshkUafU9r55hVnZ2cbg4xlsPFfR7fZ2F1ozuQcgTP2SJuR5u36g27BhY6Q6ziCU8KZKbMslOdzV5/MuEeGP3J0KpcsXM0cCj/r6uj6wK7uWld2tm5dxsyahDCgmIrrvd+YAdIxC5aD63ZU56P6u+c+b5C9937DwOOdcN41BNj+zlWQEyjTTP9c++kuDz2qojiFDGOPFXTsEEbjADMG74xGeXZpRwOOSw2eUU4pM/TKYWQ04sXxywaefaUmkzn32fn1Z7fQRuWha9udkc8a+kgPRsZ+L5E9PGgYPo+CMmUGy3mywbk+l/Lg0FJRktNEnhw6ZQ2o/MTzaqyzBl/1jRlZsrQzRuJ4q+4voRklr67pfeewWd7aV1djHyF81U1UHuM5nR2YIdWvGb3i55a2A3BH+8YD2NoJJjzo3t7eje2gXGge19mgQuBs8GqkM30rh+z8rGsMbahQuuCbUd4p2kw4r2VVvFe8KQ+cpprKm0FqV7+4tgTBKnKGHvJz211rnXQK0PHHIftqtQJwvV+i9t8V9R1YuPpyPnrUbmLwyEfXjR3RzgfooiFaa5swnsNeFnxm9Fm/PZ6vkF7TMjnh63Xmy3lbRdvMMSivWXpHSwylMtrsvruXRR+zfGeyVsqU1hkU562GX4XvPAOUlc3gwsts1dmMupVVusxgoywn68rJjmjK2FtrXwTg7wL44wA6gP8cwMcB/BiALwPwGwC+vff++1U+6/UaJycnALC1ZXPs8hqrl4DtJa2K7NnUVpQRRu6iAScgdR5OsXiajZ/VEDCuuakZlkOmiEuNxqFmlj4zfOcIdDoum8LK6ZlnsAAAIABJREFU8hwt+LmNoWchPIfv/N69G6Bzi5tUv+IX+xQGMMW1iBCYB9ZDDsszI3ftxOM+zLO2w4zMHNW7I17TBwH8w977HwPwlQA+BuADAD7ce/8KAB++Oh+S2wyhWkCjfacM4RVtVRHUo7trM0Z3mwjDoXn2f/TTtKNnM34yhHcydLKoDL1yBvp/ljJZz8pN9SFrcwWAairOyWSmHq5OGYi5Mm4DEsAEsrfWvhDAfwzgz18VdgrgtLX2rQC+/irZjwD4eQDfW+XVe98sQez9coAu9n7jfm703fXdcn0BJgvtZpU1iPMPLw1cvyChg4ZuQIWf4/xdF0JRr1r4MZInE0cezvm55zKn4RbcZI7H5ZM9x8+o4qsM43qm/JmDZqPWj0/y9wVY3zh/XorNa+Vjq/EoWyME5yz0vzvn+uuPr2vbubwqvZkJ498D4DUAf6+19pUAPgrgewC8o/f+6as0vw3gHe7h1tr7Abw/GOM+u86FMvO8Kq768GMWtmaK56j3vmXowZuG7tpg3LeaRbDMuJ7H2NWwR92N6jzj4zaDd1XdVK6VofN/rqsaSFW+C+O5TOZB0Ra4fpWavw3odG8UVlddLr2eAdhtacbYDwD8CQB/sff+kdbaByEhe++9t9YsF733VwC8AgB7e3tdBe5WyfFS2BBuhuymvBuNG9cdWnGZ2ufiI5c/02DOKyuPI6ScIefpM4dTObuKr1leXbpM5pVhVOju7s/WKesyBi/8OW8tk3cd5nbVl7Rm6jKqg3OEWq9ROY5mjP2TAD7Ze//I1flP4tLYf6e19s7e+6dba+8E8OpEXmmfKRhmQ4/NHHkwb9To3AgZGrloojJ23syyCs20USr0DFk43jnNDFVpZxSFy3bG6gzd1ccZefZs3Bs5zioc1nRL6sbdxuCD17+HToReRJcuvibEz48ixxF/LnrMdHtUTkVDY++9/3Zr7Tdba3+09/5xAN8I4Feufu8D8ANXx5+eKTDrr2Tpsj67e04Vip0DIwwbG/+P/FzjVctuWWEyA+f/zilUzmGGNJRXctdfJKJrPnyelT8yAOfgs3QjObu6czrtkml7sOOPI38N2LVfRezoKrR23ReXV6SvaHae/S8C+NHW2iGAXwPwF3A5kv/jrbXvBvBvAXz7TEZqxNr34fsHBwebH4dZ/IwzEG4AHqiJgcDT09Mtw2YkCh7Ck8e7zhrSM6/xXw1+pPjuHp9nSjsTsilVjsSht4bfLgzXvCoHweky567nHHFlhj9ynFymPsdTb1xHHq8BsNGDCOP39vY225oD2NrdZuRsM96dLLI6q44FucU4TFPG3nv/ZQBfbW5948zzTK7hsp8+x8/rdSW3RJUHW/g5RnWn7BXiZ3V0Suj+6zG7x1Qpjot2svMZZK6OGX8VZY5y5riEMkfD/0cOj+87kBrNCkUZIwMfRS1Vd3EJ7XxbquiPx4+/ssr3FL0DXTkvJRYse+UIu+IDDvGJ5ljBd3FxsZkO1OWUUS6HbNq94K5CNAY7jaVGVCEU180p2AhZqnKY15Hyzxq8i9oy/rmb5n5Zty0it9H8OddLx2329vY2nxPX7cDjfmvXH4ZcrVabMvk7f/rjiC+TlWtHBT7XVVhq9HeyLdXMz6G7C82YVLBaLntFF8LFL8qIX0wVunsZZYYyQu+RY1Cq6pyhembs1a/KNyOW9wyi830X5fH/TJ4zcnZ109khTqtdN3X4GkXGcxkKq+PO2k/rzLJy+Y90cufIzt9DPzw8xNHR0eYY1+NTyvplVF2g4hyCK1ORj9+6C1QIlD89PUXv1y9VcKNzH409P3teVRqnPLPInqXXqKUK7WYVvkJ0fm5E2h7VdKVD9pCl+6gm8xF86l6FbncaRnL9RZ57e3tbC28iCo3ng48YN4oXY+ILvgA2b3DqdLKTP8si02l1MNwezkmM7OHOw/gwfj6yoWsoX/XjKu/o7kdjhvHqiH2guc7Va39eDX2EkCPjdvdmEEKvzeZZDdDNGjmTM2w+d9f43kyEx20xu1HFqC24XbPFN9wt5AU2GdI6GWq7ZbNLKhfmQwFsBHrAHRh7IHqg+fHx8eYY9xjZgVpYej0z7CokDIFzf7216xV+geguSpiZT15KmVGyQunRPevycvlW/fOqLiy/UZ2r0D3Oq5+rX7Vmo5JlpT+Rr7a9Ao12OzXqnDF0dy3atIqElEb3g3Zu7Gzgjx8/xqNHj3B8fIzHjx9jtVrh+Ph4M0BXzX2qx82Qnn+6Ao/fsOPBuBh04UE8XkQR5WsYmPE6S5VBzhi8Kk2WHxs5Hytjr9A6rrlIQ89Hxj4a7Q7eqhepsro7Yr7VgfBGkzG4y3y6rcyd7FmOygf39zNDj/KyyPbzAtn5F311XXvMglPvWZXjUDxTOL7P4wS8FDKLCoJU0ePHA4GZ0mljcZmVw3DIHtczZM9+wM391Vz9bqNkymsWks7Id8YhBqnRaFtmZWjEwGG6Pj87qKx1mNXjEWAsAZSdGvve3h5eeuklvPzyyzg6OsKTJ0/w0ksv4ejoCC+99BL29/c3yxGB68rqwpesP+kM0i3K0Ck8Tdt730L2GKyLARin+JwfgK0FQBmiVooa1zID4TTuXMcfRkY+4idkpHUf8aSOS0P/zCFzGZq3IjDrhnOUru0VzflaoPn+/v6mzXXJdiy0ib0YQmfi+UyWmRN2kU6mEw40ZhzInSC7+/GLL2yM8Rwrl5vq4LT6zIwiKbronG/8nwlV9ZoamObDg4MOrTNDz2iEghkiZpEH55sZYMUDU+Uk4v4MsmdKX5E6fXU8nH82FlChu+Pb1SG7V7VxJeNZ2qmx7+/v4wu+4Avw8ssvb/rpT548wWq1wqNHjzb9IFYIhwYzCFQhe9UovJsoj5Kyt2ajVcTIRlb5OTc9x2MCWZg6ogrBswUzLEMta8aZafmOD322ahPXTVBes+lB5jOeZecd6BvXdWGNTq0Gskf7K9+8EKz3vrWrjYuaKrRWnXfyHNV7pCs7R3bXX3cbVTimOZTMGpkbt/Lk+pwSr4UPQWtk4BzKaLpIG901GJfBxrLEk88YvVMqzWM2mqjK5fo4GrXP86C65smOOfJk2UY7uzl51/5LFoUxH1wndWgzddc0M+20c2R/8uQJnjx5gqOjIzx69GjTR4+XXaowj/tbWUO7MMspkUMQvscDa9yoXDa/Z181dGZwkR+PKkd6Par3Z6qUvooo+LnMuc6SOo+RQVbO0tUtmwN3zsvVgbtOkZfWm9sg0FwX4OjIeyB8zOBEu3KE6OrjrmXyG802ZPkq7XyALqbajo6ONqvnwtiBbcXOPNYI+eM/H921Spkzp8F88U46oQRuMVAcVUF5OiWW5HLjKh8jow9yyK2Ko7y5umfXKgXmcvieQ6/ZUF6P2Y+fyaI3diiajnl36M5pmP9od379ladxlS+NJrLoYomxz9CdhPHZNBvgw9iR8um1Ubjurs8gJ6MEgBtG7ozd9Y81PGdyg4FZ6OYoS+eUpMrDyT4LTUc8Zc9XRq75quIvMYBRF6GKdtTY2Xi1Dlk4H3llIMV8aH1v69wc7dzYj4+Pt1bLBaK7xnDKluWb5VGld+S8fRCv3W6tbX1BVo09iJWSV2NxOZxnIP6o8ZVn5VUNxCl0ZhjPQ8qDoxmjZ2Rj3mc+8eTqkumXcyBRPpfDb8RpntHuoctukZY6+CyqiF+U7+q3xPkz3cnaeFbuuA7U0zQuL/1fee/ZMJTPVXHYg2eenEfjo9yMP16swXlnK6W0Ps6wnNfPZOjIKeFtHUCFZFpmdc+h7wjtKh6ckauhadnKx4yz4vuj+mfRpXP42fMjupN5djefnnlkd67X+ZgZh+aTKQH/VwGHMQei6/v44cw4L349NvLShTa6bDf+c0OPugPKs5veqxSmkteMIVUoHkdnRFqupgm++dNOIyNV3pkYrQOxI29F9ogkWmubNNpt47q5Pru2ERt15pQzGYwMfkQ7f5/deT5XeX2myk/TjJApC+mYsiijQnRXryX1cMiQRT+jrog6Br4+S6OyqryqqGyUj4suFI2d8lf8OnSujGiE6vrMSA9vQ65ufI/p3iE7cBMtssrMCC0L5WefzVDSNSQbuSJ7NjAXiO4cHNdZeddpvkgXaME8c16sxNzHdZGB40OvOQej007ueZalq7MaTjalul6vt5Yqu/fVmVdXlsoh8jk7O8PZ2dkG2SOtqzvf48VVLIvZ6UzneGcjrRG6jwx+58bOpEw7ZpcYvUPEGcrCUeUtQ3S3pJbDtczYM2eVHWf6fqMf12vU5VFZVEbM9cgiNb4+UnY2VA3fMyRWvtRIdCDO5elkoQ4qyqmiOFef0XV14Fl6dXKztHNjZ6EHcR9pBomdB3fPV8rM55mnZN5mQvdRV8QhoTNArocauzMkZ9SswDPKnPHC5XNkkVGl1C560jrxfUZ2t8Cl0oc4xnNnZ2ebYyA7v76sFG2edTv1GeaxiqAqR6cA4SIMV8eR7QTdibGrsCrkG1XGKUyWj/6fJWfwithZH1PrEecz5QHXL8lkiO/KVCQbOTSnfKx0jqcsHz5m95VfvcehN4fx7LgyuWXOIo5h7Dwwx45Hf26GhfOP89Hcf/ZzdXD1UbnF/1lDB+7A2HnEk0kNKEM3PgZppfWXIZquisq8rZITrjOW2/yYsrKX5jVr7JXBZ/eq+vNxRhmdoxqF25yvtifvS8cRQpZfZui6QCbKYrnorEdVNyfrynmrjJVmunjAjo299+v93nrffmtMPwDhwvEKGRWBHOo6A4j/GhpGXopwWb04T250d42PGfqOQnV+Lvs/Y+BA/j67Q/YI5/noFNI52JFCBv+8AWiE3/z1VQUCzZe3lOLwvfe+lR8PtnH9eDp1tVptBmBdOyjfI3RXPcscQ1YWk4tgK9q5sYdQQkky43ZG6ozdpalCbE7rGm0G/SoEVcNhT1/lUTkiPo74d3m4uizJS7tJo7wczSI7cHOnGOe8tCwXYlc/rR+jOaO608XZdpyVy20co+P93hn7s2fPNp6Vp6x0ZZ1jvjJ2d1+RyfGz5FoYciiLdje4wXVpLCsCvzedobE6C0YO/eY4Xx8huxpExqOinU4FLlHIoJEjDn7Oz89xenq6QXiWQ+aEuG7cL1dk111kWN9iSpV3O4531KPfru3B7e4+MJI5AuZ5FMI7IIvrGh1XbbJzY+dN/DLEdorAHjAzcOetXf5ctl7Xe3rdRQJs+EsHaGZQYuQEKsTL6jFzLzPopYY+6pZxvs5YqtkEbVOXxygfdmjZbEvWrdJyZ5Bey67kyTqf8T2D6sCOjX29XuONN97YbKzvXiDRo6vESHGy/45UWIoS3GgxMh59Vf02tyJ71viu756l0UUl3H91SJKh9IwTiPpnSMJUoXvVFvoyUZTH9eGRcn7xxJWRGXuGps7xhA5G/zxe0IofyyV45/GKzMlw+Zmeqv5UsgRuzoRUewEo7RzZz87ONgLmL6TGkVEyBkUcVajtzpWPSMMe3T3HSu145SnEWUPOjJ7Lc4N4GVIt6dtGHSvjHznIDG2q0HTkoCtEdmVk0Zgzcn0++NPwl4Gn2g/RyaNyNKpvmp+LWJSyNlEHWtHOkf3p06ebHTlZoLFTjb42WoUoI4NXqsKgSmDq0VkJtRuxxNjVsF1k4D5tpEtgszK03k75WPFdBKBI7ORZGbnK0D3PdQ75cvkqZ247zocHfjlvnTXgOvNLLYHs/IITL+UNftTZO4RXB1OtTZhxvjNRVJUHcAfG/vrrr+P8/HwrTNL15ix49yos04zBV8rOyK7hJZcRAt/b29tsQugMoDL2KDv+uzTxn0NaDeN1gC4zeldnJz9WpszYnXG5vDgfp6QZb8G/TodpxMW8OEXP1u6HcfK9qBsDDg/QhSNgmbo32Xgw1oXvI/TOHCE/NwtoFe08jD89Pd2ExPHT8D3OeRebUbg+CuMdakWazNhVwVhhslA0UMmNxkcaTqtGnoXvo+jAocrI4J0y6f1RaKgyrhyHlqnGrsaoR20TRV39rxEH39M2Z4SvlkFrPk72txk0VTnO0sgBM+3U2C8uLvDZz34Wp6enN77RHkiuodTIAOO/UoXwmqYydjfvGgatyK4NDvjNDR2yq2ErskcYP0L0KM8ZEJftZOf+K4K6eug9ZxxZ2zmDD7SdRfY48qCmWznHaXU7NC4rUF2jHien4D2iL22rTEZaZ72Xybci1jVHd7qCjpUeuLn/WqC+W4sOeEUMmgl3WOGi76VGPxtORZrM6FQODsX5+pI+v0P0zMj5/DaG7vJT2VRK6p7LED0z9jhWg6psqNqWwE1j5zJHQOGMddQmVX7VtYwyB1zRnUy9nZ2dbY14MrIfHh4OkZ0bWZXAhaXagFkDK6LoLwYRdTCPy55taDZknULja27PtWzvtczgs4hGEZfvZWFslW+lcGxwzvgijbaJzshw+6o+xP2Y5Vmv11itVhuQURm4boDWXctWh8xLe/WlG45UHFURUvCRtZ1GlC6t0s5fhOEP1q/X6833soDtXVtDqXlajI8O5V1lXUQQ1939QHiehuFGc0Lmc41WHDlkzxD9NqP7UQYftXyngDOI7vLKrjvZ6ECZlgtsd6t0my9tCzX2yCdGzZVHdRaa92xdVeZV+7i6VmCQpa+i2BeG7K21vwLgvwDQAfxzAH8BwDsBfAjA2wF8FMB39d5Pq3zCw7IhR1gfR+2zj4w9KlvwXob/el9HaHnxT0zt6MIfreOoIZ2CZHPobs81HfxzXYYlISHzpYjL3RPO19VzhFCquNplimOM5cQ1Z0SRr9ODrDw3Uj6SUzVwqvvY6bv3Klcno9l2cnVcSkNjb629C8BfAvDe3vvT1tqPA/gOAN8M4Ad77x9qrf1PAL4bwA9VeXEIGg0ZBhTGH0f3SaiR0Rret9K5X9xXR8Lz/mHsOnMQ55HHrLFlxs6G7EJ2Ne4MRZYoEKd1yu/ONXqpyuXQnY0sQ/jo3vFXU9npRf0zPeAy1cCcwVfkHJs6Zg7j9d0Ezes2Bqryqc5HNBvGHwB41Fo7A/AYwKcBfAOA77y6/yMA/joGxg5sD4bFeRh5KI96ZH79le9nhq4Nz+Gh5pMhe/C4Xq9vGHvc409WLQkJK2Ovjmrc1bw6l+VklPEVaTP+tayRk9G2YBnGNXbmPA3GvLDOZIiuPGYhtNZX/3Pe7NAcqmvk5ZyJC8Ur/biNU5h5ZmjsvfdPtdb+FoBPAHgK4P/EZdj+B733GPX4JIB3zTClihIKEI0ZyM4j8WrsroKu4Z1TYBTPjB2ADeN1wU90PdzCn1F4ywidoXd1DcjfVqvkHjxqqD5L2n7BR1Y2ozMjuobcALbaW7/my3V27cc8uZ+rRyYz5pnLz0J33YradRVmw3gu1zmnWWBzNBPGfzGAbwXwHgB/AOAnAHzT6Dl6/v0A3h/MVBWJYwiNK5HknabJDJ3LckrDwnfGpCPxwa9O42i93DU1Wq63u+cM251XVKF2Rc55LXE0Ia9w6uzQ3bNVm4fSV4o9koUzvixyCL7V8Y4G45SfKrJw6WZD/9k2nQnj/xSAX++9v3aV8U8B+DoAX9RaO7hC93cD+JR7uPf+CoBXAODg4KDTdZd26zw8PffRqsqPUD2OrlFdqB9dCJ5600FEXnnFTiAjdSZxrNC7GoGPPGcULpPNLL8Z/87gXaSjTjV+6tgdX9pGLl0mD+aZ13g4R85lqWHrNFtsR60bTc5EV8qvk5XWv6IXEsbjMnz/2tbaY1yG8d8I4BcB/ByAb8PliPz7APz0RF5DioqyIsT5jCFloY8e9b8TtIaHzAfz5crJ+GPKjJ3LzAyd85tF2KiT8pzxV9WBy8umuDRvh/AzTkuN4bakCDiKBpm4HVzIfptoSYnrmelt5gheCLL33j/SWvtJAP8MwDmAX8IlUv8fAD7UWvsbV9d+eEnFKqRmJdfR7tugu55nKO94Y8TWaULebUc//zRSTDWWDLlHaF79fzMoM/YMpdS4OYzv/fL9h5iO5ReMdICO88x0RsuMcplfXb3Jjpzzih8jeexp50bg1YlUbeCcdUYjQ9e6VzQ1Gt97/34A3y+Xfw3A18w8zzTT11JEZQVZQq4RM+EpT6ykPE2kMwasPKqYI4TPjCVDzRGy639O4+ro0mRpXZ46rsD3nDwdsvPgrI5qu3wqZXftqlN+2k4a5ThZKl+K7Jx2RFXbKY/Ml6vvUrqzL8KMDD77r+heKSWfq/DcMXtOUYkdAI8guzUBkYdThMzAKyN26VxaJ4sXRcqLOiStt3O43CVqrW12L2KeK3k64jwzWYzk2fv1HgJnZ2c4PT3d9M8Z0V0IX+mRk5+770Aja4OldGcfdoz/fGSKSiuyjjwaG7XmF/dHhu7y08U0zFsYe4TxrJwzTi0zXj3OKLCmHVGWbiYqcVOA/Dz/2Mj5Wlzf29vD6enp1mBdDIYu6R6NQCBzVFEPHogLY9ctrR2yOz4qB688Kf8ZzehqRnf2YcfsnMmFZVXIFvezfLL7FbGBM7+BRuv1eoNI2idVY5/x/Nn5EmPPri2td8a3Gko1z17lwdd0PYU6zsjDOdIldVRDD+MORF+v15tPQ/EnomZH3GeN/EVEXUvz2Pn32W/T1wDy0Gc0cBGk0zuZs9DnIn3kwYoWS3053NTj89SZeamM3vE9cnpVOiZnUGrsTnmj3q7PrMTbScU7CBHW8yIrdQBuDGcU6ge/ukAmjHy9XuPk5GSD6icnJ1uI7t44rOrmyq7OZ8k5zmrPRuAeIPubXZZTxCVhPKeP/3wd8PvIu7D1eWhk7LNKtOQ6yyZDJsfPTF1d2M/vBeg7ArxoKVvAlDlxLktXwvHe9PyK6sx69xHNOoAlVAHdqKydIzu/MhrXZvohVR9yph/nEHJmIMSV5ZyEGrYuzhnVpereVIZW8XobmnF8bDh8faa/qWF0UHSB4h2E+B/IfnZ2dmMg1L0gxTJWhxShehh2GLT2y09OTjb3z87ObuShclC5LTHySgdnuqec9t712fkliKDZQZeKMmXL0EmfqfpgnDaLBhjJXXpVxIzHUZhdKdltaYnTyCKl25TpjB3A5sWo+K8yZ8NzDtYZSSAz773vvien91RXl9Z3xvCfp2ubRa+Odo7sR0dHW/tzKTowqQFXiJgd2YiylV6zvGfkwlj+nxm7O85EMi+q31fVpTrXa8p3VY9oc55LZ+K18ozs8YahQ3Y9MukyZGfQ/ImoOOq21lEn53AymdwG7Z3cHJBp3iGPinZu7DGVoo2dIWY8p/8zY8icQ4XILr0+58rJnIYaevwfvaKrhuLoRaP6bN5V18HxX7WTc/SRltFanWfozYyxR9ooR0fcR8jO/XTmJzN2lY3qoauLcxAqp1GkVz2vtFNj39vbw/Hx8cZz6oYEwFyfdAYhNa2Shs1ZKORC7hn0dfy76TuXJzdy5bBuG/6NyHVvMqRRAx9FWCPnGml4QCzqzKPyauyA3+2Xw3dGdt0zLu5x5ME6oY5Jjd7VQUkNfQRYM+0bMue1Hhnt3NgfPXq0WQ/N/aKlRu+MMI6VEqkyZLQkwtC+k4Z7Wr4aEDdYZuiurOel2zoL5U8XEjlSo2Aj5WtRN57DD5R2KK4Rkwu1HbK7kXb92APz7uoy6naOZDjj+DIe+Dld2JXRnYTx2SjuCNGzPN3/LO1sP2p0zx2X8Opo5GBmUGEJzfBToU/cU2Nz+VTlj6IDziOMPpDeRUKsR9pnd2vbXXg+4j27X3X7ZtK5tNm1uM4GX9HOkf3x48ebqZSYTolFFezFM+/pqEKU6plRn6jK0yFZFdrNhHrquDR/54CWoIMrK6t/JXsXkWSyiDZV4pBbIwPXL9Ywulozr/3pyugrlHZHLUPTx//MYc0A0gxFvXkXpRgPy2jnU2/BnIZXoTSsHA71M8oEWT33osJhza/y4rcpz4X+mu/S/Nxx6fMO0bWr4WSsKO42/mAjVQfARybtTrFxz4BHFq5znWfbz3XZ9Lxy9hW5SGgG8HZq7Pv7+3j8+DFaaxtUb61t7SXPXtcp+aiR3b04ZmHyCOGr0Eyvu/yYB6e4Wb6uIZ/XOSkq8zXmd2norZQpMjsI/YIvh+XA9qYePHhWhdkO1dVoox0UVNRRcEQxEz04h+zK4HuZTlbEsooPrPD22xntHNmDqd771muNGUoEVR5RaSaEfjNphL5L+HFKWeUxgw5Z+DvLj+axxBGx0btf5uA16pulyjFoNLLk2SxdVY7ThQxslNxzTn4V7bzPfnR0tFkKGYrCH46I0B64/hKqC4lUoOqBHT2P0c+gXXWP68Avh0Qah5KZUXLUszQUHCFzFpU4tJoJHbmLxugW/czWrtdecD8+UFU3DnGLcbLIxEVQMw6zQnbNS8tx8puVk5K2O7cXv1LNv3uD7NGw/JZTCIO3egJwQ9jxvBq8E37cf14aedwqjHVplziLLNx2zmKUV5WvczCAN3LHj5JzPqzwHH7Ghp0axkc+PGCrbxzOOu5I76bSqghyJjrMnIp22zK+lpC2VxYZVXQnA3QHB5fFsjHrvm6ZEce50osO3bXhMsSdNXjOh5WY856JEGbCviUGnz0zUtJR+WoIQbwff2bsDAJxdCF8ZpiOF0dZn9ulGz3v0nHZz2P4auSV0d8rZI/+Onuh6LfriqagJY2xxONn+VQNpMjopp0cH6PQMeNF81Bkd8o8a+xcDxcyzjyvvKozinvR3u6jG7wVNxs7L6xhY4/63+Zdh5EuZfllYFIZuzoslsuMnPm+WxrMv5BpRTs3dmUuDN0dY+RVGxeoB6hmDGqJF6/6TktC+Sy/ET9qnGpMS8LaET8zqF/lo/yojKLNecWXvq7KITw71ahvNj5TyeI2TnYULc7qmTN4x+soetTrTOwIMtp5GL9arQBsT8HESKvuTBL3gZvf3NIw2HnrJY2TIaT+1JvqKPIotHTefQZpbmN8I9I8q4UqGV/VOeejTt6F7+rAOHzl2IisAAAG2klEQVRX45jpW7swe/TL6pm17UxUMTL4kJUe3dLgzPBDthXdybZUgdRqNIr0YfjsECIfFyqOPGdFo3RZqLsEpTWdKrfjp+rrRR7Pg+qRR4Ycs44lU9z4ryierW1XnjLeuM2jfOVHz5fISdFzZOizNGrPKHvEj0tfpQHuaDQe8MjOe4+FUCIdT8O5zf/Ue6sROOdQkc77Z6OfGSpxOdwYWWM79NHzkQFWCu/CvixqyQxwVE6GOJx3ttMMk4bwrCN6PUPyiABvg+bqTLQtZuQ9k9+oTZxuaXnadhXdyU41PFgT4XuEIDFSr59JBnBjCga4aVzV/1lPXCFMFVZF+pGRsSyWOqER36PwtkKNDE2rcjgNHzUE5XlhF55m9XfyXUIjA9WyRv9dHqN2m4nOXARTla/Of4TqwB0N0EXDM9Or1cruJsqDc3GMvLg/p8trlWbCdHfM+k2zo/Ccv+atCM6bJeh1FxUsdWaZUTMyjJC9QqjKyKtoSPO/jUGzPB1qu2ipkhOj5m116jaUyYT/s+3EMyzzjO5knt0pbYTx8YVUHnnX94+5IQLt3UKTCmGVnLGrUauicp6qTJq3M3ZWJFeHTEFdtLBU8ZxhVggfPGRGDmw7Rv1oBht95iSX1KFysksMO8t31F4vytBHerkkGr1XyA5sh/G9X0+rRRjPx4ODg42ht9Y2571fbzQQxKP1DgVnwjcWmPbZR8rplMAZJeetPGZOq+LZUYaOlUFXhs7ydN0jZ+xuXjhDn6q+VZ/aRTmz+WaycQ5taTdwCY26cc65ZvmM+LuzraS1ocLw+bVXVgzeaQS4VqxwBMDNEexZZB8ZeWXwfKzy5p8OpOh97q5oWVyXkSfPeHFIPhvGuzwjrU6vhYN2ZVay0zB8pnsy0292z6juVMj+PFR1T1zXiB2pOlf3vOu6KC37LOoLIoccVRp1DO75mTAm46X67/LOBPs8nn8p/7cxdHc+m49Ln7VLdT/j5z7QfeRpRIuilzcjNEkLa+01AK8D+N2dFfpi6N/B5x/PwOcn3w88Px/9+733f9fd2KmxA0Br7Rd771+900Kfkz4feQY+P/l+4PnNozsJ4x/ogR5o9/Rg7A/0QG8Rugtjf+UOynxe+nzkGfj85PuB5zeJdt5nf6AHeqC7oYcw/oEe6C1CD8b+QA/0FqGdGXtr7Ztaax9vrf1qa+0Duyp3KbXWvrS19nOttV9prf3L1tr3XF1/W2vtH7fW/s3V8Yvvmlel1tp+a+2XWms/e3X+ntbaR65k/mOttcO75pGptfZFrbWfbK39q9bax1prf/LzRM5/5Uo3/kVr7X9rrR3fd1kDOzL21to+gP8RwH8K4L0A/mxr7b27KPsWdA7gr/be3wvgawH8l1e8fgDAh3vvXwHgw1fn942+B8DH6PxvAvjB3vsfAfD7AL77TrjK6YMA/mHv/Y8B+Epc8n6v5dxaexeAvwTgq3vvfxzAPoDvwP2X9dxWPc/7A/AnAfwjOv8+AN+3i7JfAO8/DeBPA/g4gHdeXXsngI/fNW/C57txaRzfAOBnATRcruo6cG1w1z8AXwjg13E1SEzX77uc3wXgNwG8DZfvlvwsgP/kPss6frsK40NAQZ+8unavqbX2ZQC+CsBHALyj9/7pq1u/DeAdd8RWRn8HwF8DEG/QvB3AH/Tez6/O75vM3wPgNQB/76rr8Xdba09wz+Xce/8UgL8F4BMAPg3gMwA+ivstawAPA3QptdZeAvD3Afzl3vsf8r1+6b7vzZxla+1bALzae//oXfOygA4A/AkAP9R7/ypcvjOxFbLfNzkDwNUYwrfi0ln9ewCeAPimO2VqknZl7J8C8KV0/u6ra/eSWmsrXBr6j/bef+rq8u+01t55df+dAF69K/4MfR2AP9Na+w0AH8JlKP9BAF/UWovXmO+bzD8J4JO9949cnf8kLo3/PssZAP4UgF/vvb/Wez8D8FO4lP99ljWA3Rn7LwD4iqsRy0NcDmj8zI7KXkTt8j3HHwbwsd7736ZbPwPgfVf/34fLvvy9oN779/Xe3917/zJcyvb/7r3/OQA/B+DbrpLdN55/G8Bvttb+6NWlbwTwK7jHcr6iTwD42tba4ytdCb7vraw3tMOBjW8G8K8B/L8A/ru7Hqwo+PyPcBk6/j8Afvnq98247AN/GMC/AfB/AXjbXfOa8P/1AH726v+XA/inAH4VwE8AOLpr/oTX/wDAL17J+n8H8MWfD3IG8N8D+FcA/gWA/xXA0X2Xde/9YbnsAz3QW4UeBuge6IHeIvRg7A/0QG8RejD2B3qgtwg9GPsDPdBbhB6M/YEe6C1CD8b+QA/0FqEHY3+gB3qL0P8H53YHUCjpJtYAAAAASUVORK5CYII=\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# creating val dataset object & verifying it\n",
        "%time val_dataset = get_dataset(Val_paths, Val_labels, train=False)\n",
        "\n",
        "image, label = next(iter(val_dataset))\n",
        "print(image.shape)\n",
        "print(label.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "swMkv9K73JDo",
        "outputId": "2b30c77a-8585-4faa-81fb-25e64d7227d9"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "CPU times: user 28.7 ms, sys: 1.9 ms, total: 30.6 ms\n",
            "Wall time: 50.1 ms\n",
            "(32, 96, 96, 3)\n",
            "(32, 7)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# view a sample val image\n",
        "print(Le.inverse_transform(np.argmax(label, axis=1))[0])\n",
        "plt.imshow((image[0].numpy()/255).reshape(96,96,3))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "XZQata3T3ckn",
        "outputId": "1964844d-143a-4885-fac7-8feedd7ebf94"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "angry\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f3df91377d0>"
            ]
          },
          "metadata": {},
          "execution_count": 21
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "#### Building EfficientNet Model"
      ],
      "metadata": {
        "id": "ShQ3c0UR3ynR"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "from tensorflow.keras.applications import EfficientNetB2\n",
        "\n",
        "backbone = EfficientNetB2(\n",
        "    input_shape = (96,96,3),\n",
        "    include_top = False,\n",
        ")\n",
        "\n",
        "model = tf.keras.Sequential([\n",
        "    backbone,\n",
        "    tf.keras.layers.GlobalAveragePooling2D(),\n",
        "    tf.keras.layers.Dropout(0.3),\n",
        "    tf.keras.layers.Dense(128, activation=\"relu\"),\n",
        "    tf.keras.layers.Dense(7, activation=\"softmax\"),\n",
        "])\n",
        "\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CM_eIRZl3p2w",
        "outputId": "84c6f629-f9bd-4fb6-87ef-96a8af425442"
      },
      "execution_count": 22,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/keras-applications/efficientnetb2_notop.h5\n",
            "31793152/31790344 [==============================] - 1s 0us/step\n",
            "31801344/31790344 [==============================] - 1s 0us/step\n",
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " efficientnetb2 (Functional)  (None, 3, 3, 1408)       7768569   \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 1408)             0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 1408)              0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 128)               180352    \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 7)                 903       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,949,824\n",
            "Trainable params: 7,882,249\n",
            "Non-trainable params: 67,575\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# compiling model\n",
        "model.compile(\n",
        "    optimizer = tf.keras.optimizers.Adam(learning_rate=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-07),\n",
        "    loss = 'categorical_crossentropy',\n",
        "    metrics = ['accuracy', tf.keras.metrics.Precision(name='precision'), tf.keras.metrics.Recall(name='recall')],\n",
        ")"
      ],
      "metadata": {
        "id": "2Y-GhoMB4SMC"
      },
      "execution_count": 23,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# train model\n",
        "history = model.fit(\n",
        "    train_dataset,\n",
        "    steps_per_epoch = len(Train_paths)//BATCH_SIZE,\n",
        "    epochs = 12,\n",
        "    # callbacks = [checkpoint, early_stop],\n",
        "    validation_data = val_dataset,\n",
        "    validation_steps = len(Val_paths)//BATCH_SIZE,\n",
        "    class_weight = class_weight,\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CRXta9nk49pZ",
        "outputId": "8f13852b-99fd-4335-c59d-b9a619c5960f"
      },
      "execution_count": 24,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/12\n",
            "672/672 [==============================] - 124s 143ms/step - loss: 2.7832 - accuracy: 0.4030 - precision: 0.6459 - recall: 0.1947 - val_loss: 1.6379 - val_accuracy: 0.3640 - val_precision: 0.5109 - val_recall: 0.1833\n",
            "Epoch 2/12\n",
            "672/672 [==============================] - 96s 140ms/step - loss: 2.3722 - accuracy: 0.4923 - precision: 0.6784 - recall: 0.2918 - val_loss: 1.4926 - val_accuracy: 0.4654 - val_precision: 0.6042 - val_recall: 0.3305\n",
            "Epoch 3/12\n",
            "672/672 [==============================] - 93s 138ms/step - loss: 2.2258 - accuracy: 0.5275 - precision: 0.7009 - recall: 0.3343 - val_loss: 1.3596 - val_accuracy: 0.4880 - val_precision: 0.5977 - val_recall: 0.3545\n",
            "Epoch 4/12\n",
            "672/672 [==============================] - 94s 139ms/step - loss: 2.1195 - accuracy: 0.5440 - precision: 0.7114 - recall: 0.3603 - val_loss: 1.4070 - val_accuracy: 0.4805 - val_precision: 0.5932 - val_recall: 0.3271\n",
            "Epoch 5/12\n",
            "672/672 [==============================] - 94s 140ms/step - loss: 2.0382 - accuracy: 0.5606 - precision: 0.7165 - recall: 0.3863 - val_loss: 1.1531 - val_accuracy: 0.5660 - val_precision: 0.6837 - val_recall: 0.4414\n",
            "Epoch 6/12\n",
            "672/672 [==============================] - 94s 139ms/step - loss: 2.0090 - accuracy: 0.5668 - precision: 0.7189 - recall: 0.3968 - val_loss: 1.2675 - val_accuracy: 0.5218 - val_precision: 0.6672 - val_recall: 0.3863\n",
            "Epoch 7/12\n",
            "672/672 [==============================] - 98s 146ms/step - loss: 1.9076 - accuracy: 0.5800 - precision: 0.7238 - recall: 0.4218 - val_loss: 1.1638 - val_accuracy: 0.5660 - val_precision: 0.6893 - val_recall: 0.4389\n",
            "Epoch 8/12\n",
            "672/672 [==============================] - 95s 141ms/step - loss: 1.8421 - accuracy: 0.5944 - precision: 0.7324 - recall: 0.4406 - val_loss: 1.1467 - val_accuracy: 0.5762 - val_precision: 0.6624 - val_recall: 0.4897\n",
            "Epoch 9/12\n",
            "672/672 [==============================] - 93s 139ms/step - loss: 1.8206 - accuracy: 0.6008 - precision: 0.7312 - recall: 0.4542 - val_loss: 1.1932 - val_accuracy: 0.5558 - val_precision: 0.6715 - val_recall: 0.4368\n",
            "Epoch 10/12\n",
            "672/672 [==============================] - 94s 139ms/step - loss: 1.7956 - accuracy: 0.6062 - precision: 0.7344 - recall: 0.4593 - val_loss: 1.1315 - val_accuracy: 0.5830 - val_precision: 0.6742 - val_recall: 0.4795\n",
            "Epoch 11/12\n",
            "672/672 [==============================] - 93s 139ms/step - loss: 1.7305 - accuracy: 0.6171 - precision: 0.7382 - recall: 0.4771 - val_loss: 1.0549 - val_accuracy: 0.6090 - val_precision: 0.7014 - val_recall: 0.5165\n",
            "Epoch 12/12\n",
            "672/672 [==============================] - 93s 139ms/step - loss: 1.6747 - accuracy: 0.6244 - precision: 0.7419 - recall: 0.4873 - val_loss: 1.1601 - val_accuracy: 0.5875 - val_precision: 0.6709 - val_recall: 0.5074\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# turn off the backbone\n",
        "model.layers[0].trainable = False"
      ],
      "metadata": {
        "id": "MuK3KZ7A5Ty-"
      },
      "execution_count": 33,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# defining callbacks\n",
        "checkpoint = tf.keras.callbacks.ModelCheckpoint(\"best_weights.h5\", verbose=1, save_best_only=True, save_weights_only=True)\n",
        "early_stop = tf.keras.callbacks.EarlyStopping(patience=4)"
      ],
      "metadata": {
        "id": "0TA5AQw85Zmq"
      },
      "execution_count": 26,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "cYRNVReA7fKT",
        "outputId": "646d8d94-3b9f-4b9c-c0f7-46aa16945e67"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " efficientnetb2 (Functional)  (None, 3, 3, 1408)       7768569   \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 1408)             0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dropout (Dropout)           (None, 1408)              0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 128)               180352    \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 7)                 903       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,949,824\n",
            "Trainable params: 181,255\n",
            "Non-trainable params: 7,768,569\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# train the phase 2 of model\n",
        "history = model.fit(\n",
        "    train_dataset,\n",
        "    steps_per_epoch = len(Train_paths)//BATCH_SIZE,\n",
        "    epochs = 8,\n",
        "    callbacks = [checkpoint, early_stop],\n",
        "    validation_data = val_dataset,\n",
        "    validation_steps = len(Val_paths)//BATCH_SIZE,\n",
        "    class_weight = class_weight,\n",
        ")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Xn0kVN7y7aOB",
        "outputId": "bfa4150d-17b4-4431-82c5-89626ba0e254"
      },
      "execution_count": 28,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.6620 - accuracy: 0.6311 - precision: 0.7483 - recall: 0.5010\n",
            "Epoch 1: val_loss improved from inf to 1.05514, saving model to best_weights.h5\n",
            "672/672 [==============================] - 94s 139ms/step - loss: 1.6620 - accuracy: 0.6311 - precision: 0.7483 - recall: 0.5010 - val_loss: 1.0551 - val_accuracy: 0.6091 - val_precision: 0.7002 - val_recall: 0.5112\n",
            "Epoch 2/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.6099 - accuracy: 0.6401 - precision: 0.7532 - recall: 0.5144\n",
            "Epoch 2: val_loss improved from 1.05514 to 1.02879, saving model to best_weights.h5\n",
            "672/672 [==============================] - 93s 138ms/step - loss: 1.6099 - accuracy: 0.6401 - precision: 0.7532 - recall: 0.5144 - val_loss: 1.0288 - val_accuracy: 0.6211 - val_precision: 0.7174 - val_recall: 0.5230\n",
            "Epoch 3/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.5901 - accuracy: 0.6447 - precision: 0.7520 - recall: 0.5260\n",
            "Epoch 3: val_loss did not improve from 1.02879\n",
            "672/672 [==============================] - 94s 140ms/step - loss: 1.5901 - accuracy: 0.6447 - precision: 0.7520 - recall: 0.5260 - val_loss: 1.1125 - val_accuracy: 0.6003 - val_precision: 0.6879 - val_recall: 0.4937\n",
            "Epoch 4/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.5530 - accuracy: 0.6542 - precision: 0.7574 - recall: 0.5390\n",
            "Epoch 4: val_loss did not improve from 1.02879\n",
            "672/672 [==============================] - 93s 138ms/step - loss: 1.5530 - accuracy: 0.6542 - precision: 0.7574 - recall: 0.5390 - val_loss: 1.0885 - val_accuracy: 0.6010 - val_precision: 0.6989 - val_recall: 0.5100\n",
            "Epoch 5/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.5180 - accuracy: 0.6609 - precision: 0.7583 - recall: 0.5451\n",
            "Epoch 5: val_loss did not improve from 1.02879\n",
            "672/672 [==============================] - 93s 138ms/step - loss: 1.5180 - accuracy: 0.6609 - precision: 0.7583 - recall: 0.5451 - val_loss: 1.1061 - val_accuracy: 0.6014 - val_precision: 0.6759 - val_recall: 0.5315\n",
            "Epoch 6/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.4848 - accuracy: 0.6624 - precision: 0.7631 - recall: 0.5536\n",
            "Epoch 6: val_loss improved from 1.02879 to 1.02766, saving model to best_weights.h5\n",
            "672/672 [==============================] - 92s 137ms/step - loss: 1.4848 - accuracy: 0.6624 - precision: 0.7631 - recall: 0.5536 - val_loss: 1.0277 - val_accuracy: 0.6324 - val_precision: 0.7093 - val_recall: 0.5545\n",
            "Epoch 7/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.4298 - accuracy: 0.6773 - precision: 0.7712 - recall: 0.5748\n",
            "Epoch 7: val_loss did not improve from 1.02766\n",
            "672/672 [==============================] - 92s 137ms/step - loss: 1.4298 - accuracy: 0.6773 - precision: 0.7712 - recall: 0.5748 - val_loss: 1.0634 - val_accuracy: 0.6182 - val_precision: 0.6862 - val_recall: 0.5446\n",
            "Epoch 8/8\n",
            "672/672 [==============================] - ETA: 0s - loss: 1.4200 - accuracy: 0.6802 - precision: 0.7690 - recall: 0.5764\n",
            "Epoch 8: val_loss did not improve from 1.02766\n",
            "672/672 [==============================] - 92s 136ms/step - loss: 1.4200 - accuracy: 0.6802 - precision: 0.7690 - recall: 0.5764 - val_loss: 1.0872 - val_accuracy: 0.6124 - val_precision: 0.6854 - val_recall: 0.5469\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Testing Phase"
      ],
      "metadata": {
        "id": "7uIFr6wq7-p3"
      }
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "2hIYMC-fv0Kf"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "backbone = EfficientNetB2(\n",
        "    input_shape = (96,96,3),\n",
        "    include_top = False,\n",
        ")\n",
        "\n",
        "model = tf.keras.Sequential([\n",
        "    backbone,\n",
        "    tf.keras.layers.GlobalAveragePooling2D(),\n",
        "    tf.keras.layers.Dropout(0.3),\n",
        "    tf.keras.layers.Dense(128, activation=\"relu\"),\n",
        "    tf.keras.layers.Dense(7, activation=\"softmax\"),\n",
        "])\n",
        "\n",
        "model.compile(\n",
        "    optimizer = tf.keras.optimizers.Adam(learning_rate=0.001, beta_1=0.9, beta_2=0.999, epsilon=1e-07),\n",
        "    loss = 'categorical_crossentropy',\n",
        "    metrics = ['accuracy', tf.keras.metrics.Precision(name='precision'), tf.keras.metrics.Recall(name='recall')],\n",
        ")"
      ],
      "metadata": {
        "id": "bEUv3tzZwimn"
      },
      "execution_count": 49,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.load_weights(\"best_weights.h5\")"
      ],
      "metadata": {
        "id": "8GmU4g5-7pyP"
      },
      "execution_count": 50,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# Create a Dataset Object for 'Testing' Set just the way we did for Training and Validation\n",
        "test_image_paths = list(test_path.glob(\"*/*\"))\n",
        "test_image_paths = list(map(lambda x: str(x), test_image_paths))\n",
        "test_labels = list(map(lambda x: get_label(x), test_image_paths))\n",
        "\n",
        "test_labels = Le.transform(test_labels)\n",
        "test_labels = tf.keras.utils.to_categorical(test_labels)\n",
        "\n",
        "test_image_paths = tf.convert_to_tensor(test_image_paths)\n",
        "test_labels = tf.convert_to_tensor(test_labels)\n",
        "\n",
        "def decode_image(image, label):\n",
        "  image = tf.io.read_file(image)\n",
        "  image = tf.io.decode_jpeg(image, channels=3)\n",
        "  image = tf.image.resize(image, [96,96], method=\"bilinear\")\n",
        "  return image, label\n",
        "\n",
        "test_dataset = (\n",
        "    tf.data.Dataset\n",
        "    .from_tensor_slices((test_image_paths, test_labels))\n",
        "    .map(decode_image)\n",
        "    .batch(BATCH_SIZE)\n",
        ")"
      ],
      "metadata": {
        "id": "wbhTkKZz720H"
      },
      "execution_count": 51,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# verify test dataset object\n",
        "image, label = next(iter(test_dataset))\n",
        "\n",
        "print(image.shape)\n",
        "print(label.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zyap8x7F78ef",
        "outputId": "c830290b-9ae2-4b2e-86b3-f4cc53f23c4f"
      },
      "execution_count": 52,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(32, 96, 96, 3)\n",
            "(32, 7)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# view a sample validation image\n",
        "print(Le.inverse_transform(np.argmax(label, axis=1))[0])\n",
        "plt.imshow((image[0].numpy()/255).reshape(96,96,3))"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 304
        },
        "id": "WVYiSJJP8Ci7",
        "outputId": "87922a4c-bf3f-4cda-8fd6-d287bd8f85a5"
      },
      "execution_count": 53,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "angry\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7f3d59092150>"
            ]
          },
          "metadata": {},
          "execution_count": 53
        },
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# evaluating the loaded model\n",
        "loss, acc, prec, rec = model.evaluate(test_dataset)\n",
        "\n",
        "print(\"Testing Accuracy: \", acc)\n",
        "print(\"Testing Precision: \", prec)\n",
        "print(\"Testing Recall: \", rec)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XUfHy4Cd8H_W",
        "outputId": "0f77647e-6bf2-4600-e010-a2d07bd18d22"
      },
      "execution_count": 54,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "225/225 [==============================] - 10s 30ms/step - loss: 1.0290 - accuracy: 0.6330 - precision: 0.7080 - recall: 0.5539\n",
            "Testing Accuracy:  0.6330454349517822\n",
            "Testing Precision:  0.7079772353172302\n",
            "Testing Recall:  0.5539147257804871\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# save model\n",
        "model.save(\"DetectFacialExpressionModel.h5\")"
      ],
      "metadata": {
        "id": "KVCWnuG-8K2V"
      },
      "execution_count": 55,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "# save label encoder\n",
        "import pickle\n",
        "def save_object(obj, name):\n",
        "  pickle_obj = open(f\"{name}.pck\",\"wb\")\n",
        "  pickle.dump(obj, pickle_obj)\n",
        "  pickle_obj.close()"
      ],
      "metadata": {
        "id": "U6RB1MKn8YGM"
      },
      "execution_count": 58,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "save_object(Le, \"LabelEncoder\")"
      ],
      "metadata": {
        "id": "vEApabee8rgB"
      },
      "execution_count": 59,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        ""
      ],
      "metadata": {
        "id": "pfnJg0lhAG0U"
      },
      "execution_count": null,
      "outputs": []
    }
  ]
}