Inception_FE+VGG_FT+VGG_FE.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Copy of Final_Project.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "ub_xAe31w6sA",
        "DkjvanJfW8jj"
      ],
      "machine_shape": "hm"
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "language_info": {
      "name": "python"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "\n",
        "drive.mount('/content/drive')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BIbvjN6j3y69",
        "outputId": "8b2851b1-60e0-4ca6-976f-5fd5e02987b5"
      },
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Data Preprocess"
      ],
      "metadata": {
        "id": "fKZGYpZz4MP5"
      }
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "mJWSHYEMpJEN"
      },
      "outputs": [],
      "source": [
        "import numpy as np\n",
        "import pandas as pd\n",
        "import seaborn as sns\n",
        "import matplotlib.pyplot as plt\n",
        "%matplotlib inline\n",
        "\n",
        "import cv2\n",
        "import skimage\n",
        "from skimage.transform import resize\n",
        "\n",
        "import os\n",
        "import torch\n",
        "import torch.nn as nn\n",
        "import torchvision\n",
        "from torchvision import models\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from tensorflow.keras.utils import to_categorical"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# !unzip -uq \"/content/drive/My Drive/ASL_Data.zip\"\n",
        "!unzip -o \"/content/drive/My Drive/ASL_Data.zip\" | awk 'BEGIN {ORS=\" \"} {if(NR%10==0)print \".\"}'"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QRaA5V6DHy4Y",
        "outputId": "0f3c6358-1a11-415e-e947-700615bcac00"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            ". . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . "
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "batch_size = 10\n",
        "imageSize = 76\n",
        "target_dims = (imageSize, imageSize, 3)\n",
        "num_classes = 29\n",
        "\n",
        "train_len = 87000\n",
        "train_dir = '/content/ASL_Data/asl_alphabet_train/asl_alphabet_train/'"
      ],
      "metadata": {
        "id": "anCGjn9ILcVz"
      },
      "execution_count": 3,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "def get_data(folder):\n",
        "    X = np.empty((train_len, imageSize, imageSize, 3), dtype=np.float32)\n",
        "    y = np.empty((train_len,), dtype=np.int64)\n",
        "    cnt = 0\n",
        "    for folderName in os.listdir(folder):\n",
        "        if not folderName.startswith('.'):\n",
        "            if folderName in ['A']:\n",
        "                label = 0\n",
        "            elif folderName in ['B']:\n",
        "                label = 1\n",
        "            elif folderName in ['C']:\n",
        "                label = 2\n",
        "            elif folderName in ['D']:\n",
        "                label = 3\n",
        "            elif folderName in ['E']:\n",
        "                label = 4\n",
        "            elif folderName in ['F']:\n",
        "                label = 5\n",
        "            elif folderName in ['G']:\n",
        "                label = 6\n",
        "            elif folderName in ['H']:\n",
        "                label = 7\n",
        "            elif folderName in ['I']:\n",
        "                label = 8\n",
        "            elif folderName in ['J']:\n",
        "                label = 9\n",
        "            elif folderName in ['K']:\n",
        "                label = 10\n",
        "            elif folderName in ['L']:\n",
        "                label = 11\n",
        "            elif folderName in ['M']:\n",
        "                label = 12\n",
        "            elif folderName in ['N']:\n",
        "                label = 13\n",
        "            elif folderName in ['O']:\n",
        "                label = 14\n",
        "            elif folderName in ['P']:\n",
        "                label = 15\n",
        "            elif folderName in ['Q']:\n",
        "                label = 16\n",
        "            elif folderName in ['R']:\n",
        "                label = 17\n",
        "            elif folderName in ['S']:\n",
        "                label = 18\n",
        "            elif folderName in ['T']:\n",
        "                label = 19\n",
        "            elif folderName in ['U']:\n",
        "                label = 20\n",
        "            elif folderName in ['V']:\n",
        "                label = 21\n",
        "            elif folderName in ['W']:\n",
        "                label = 22\n",
        "            elif folderName in ['X']:\n",
        "                label = 23\n",
        "            elif folderName in ['Y']:\n",
        "                label = 24\n",
        "            elif folderName in ['Z']:\n",
        "                label = 25\n",
        "            elif folderName in ['del']:\n",
        "                label = 26\n",
        "            elif folderName in ['nothing']:\n",
        "                label = 27\n",
        "            elif folderName in ['space']:\n",
        "                label = 28           \n",
        "            else:\n",
        "                label = 29\n",
        "            for image_filename in os.listdir(folder + folderName):\n",
        "                img_file = cv2.imread(folder + folderName + '/' + image_filename)\n",
        "                if img_file is not None:\n",
        "                    img_file = skimage.transform.resize(img_file, (imageSize, imageSize, 3))\n",
        "                    img_arr = np.asarray(img_file).reshape((-1, imageSize, imageSize, 3))\n",
        "                    \n",
        "                    X[cnt] = img_arr\n",
        "                    y[cnt] = label\n",
        "                    cnt += 1\n",
        "    return X,y"
      ],
      "metadata": {
        "id": "6BXtpAHyLoGl"
      },
      "execution_count": 4,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "X_train, y_train = get_data(train_dir)\n",
        "print(\"Images successfully imported...\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pkftXGpJLqAU",
        "outputId": "8b2e7577-55df-481a-c808-a53b92dfd6ce"
      },
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Images successfully imported...\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(\"The shape of X_train is : \", X_train.shape)\n",
        "print(\"The shape of y_train is : \", y_train.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1azLL6sNqBzh",
        "outputId": "130b44d1-4dae-402c-8cab-e97954217533"
      },
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "The shape of X_train is :  (87000, 76, 76, 3)\n",
            "The shape of y_train is :  (87000,)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "plt.imshow(X_train[0])\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 268
        },
        "id": "phMp8nSJOR48",
        "outputId": "1f74e56b-fa9e-4411-e651-8aed67b2f75f"
      },
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAPsAAAD7CAYAAACscuKmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nOy9Taht35Yf9Btz7b3POff+33v/9yrlq6JSkECCIQgmnRixI5FAGkI6UphASCOQjjYCNizsxIaN2FEERSgwWIJYBj9QJCISIsGOBFEUUygxKFZ8ZRSqSqvqvf89e81hY4zfGL+59j7n3vfhvQ/rrnv32WuvNT/HHF9zjDHnNHfH5+vz9fn6//81PnUDPl+fr8/Xx7k+E/vn6/P1O+T6TOyfr8/X75DrM7F/vj5fv0Ouz8T++fp8/Q65PhP75+vz9Tvk+qGI3cz+hJn9j2b2t83s539Ujfp8fb4+Xz/6y35QP7uZbQD+JwB/HMCvAPibAP6Uu/+tH13zPl+fr8/Xj+o6/RB5/wiAv+3ufwcAzOyXAPxJAC8S+7fM/GcB7ACeYZj1xuEA7JD+U4b7HNvC631t+pA+vFT2h1wfWt73C7sfpk3vu37QttzN976Gfl+VfR+9/gERwl6twt5fwL3KjWVb/dzcYQD+9znxa+53a/1hiP1nAPxv8vtXAPxDr2X4WQD/6Rj4DQx8xwe+C4MPh9texM55xQ5gSidvEMDX58d0S1rcglTL88MztsMOmdyY3jufRXpzze/1914dY3n7chs/7L3dYTAu6e1uH3lvmfL9aH+fjVn+sUMDtU6F1YfUwrKqBh2ImzL6gcM+gHYUEkw/YnDRsKtGZN2eHY1Uo8nUG5O65IaNGTBMRBmp39lTh7lAqrpjMiqZ1gw2ANsMZsC2bRhj4OyOL/Ydj3Pi5377t1/s+Q9D7B90mdmfB/DnAeCnAfw6Bn7VBn7ZNvyGGaZN7BZ9D4IJwEw/oOABqEeOWT99RYx7BG/yi8SrZZjHcJoAvtIaCdlv8x2+XyP2Y7s/hL/fT3OfULtOv5vv2FYt5zb9KoE0z/r7tlUKg6XgFwhfmaYmJlOJnwfCIXm5oIlWcpQQzFOIlKMiyY2EfgCUwWAW7LoYBRLKh3oMQehGgWAtjW+Q83AFrkXZNjaYDdgwXC5nnC8bzAzbNrANw2VOfOv5Hd5er7h+73svlvnDEPvfRQhrXr87n62Ndv8FAL8AAH+/Df+ObfhlG/jPtxN+FQNXm3g3JiYAc2+pOE24H+BmjT4L5hyQ3RTND39rwJpgtSTLRGbAcKpI2gaXPA53TkTWNhjWiUnnuiX2A7D0x+37uwSScCnJpCWoVLpDwlKeCs8bM85RxcENDUidx9pfeH94uciwGoeEvhswyHwNId4s6UUYsdYp42c1+DJ+rpJzwJWB1G0TuxUvMAxsIWUxQK3AAbg73EUmm2PAi+C3YdhGtsUJhpVhuX4Hd8DpdMJ2PmPbNnzti7d4++YJYxhOJ2AbwNN+xc/81m/hy3df4bu/8Rt3oB7XD0PsfxPA7zez34sg8n8SwJ9+LcME8F0YfsNCuv9dMzzbwFcW72Ci0hT3bEWzhtX4JH8v3/2TXP6+ZPMEbBOlGYtPRHNyYkUolukxGCEeAjFuaNVvWldc/tig9xH6i5cHMRwYl9673xWXN49Ufi9akN8mPjK3l+p+6fe9y6Cw8dCupPz6ZUnwaPXa4WIDwlLSQOMQnFMwq7yRP1lwET2FvotUDlI3bNmavDfDzHIbN6PHwyYGHGaG8wDGqEQlgVza52QcpsR+wfl0xnba8NsPj/ja0xuMYTifgG043lyveHt9hsExx8sOth+Y2N39amb/NID/DMAG4C+7+//w3nwWHdnHCdex4YqJZ5uYcBgmQGlpOo9q1GrE1m/CmFzbBZ5H3dFvPotkTwQYrijW6OzWROXYAwOyXkoeYUvMtrRi3Dz34jS3TGAp6XAJk7qT7QU2uCSyw++b9IbbNr3y/IOI3G5fyAhjcAzMivHCRqrPAChRK4vB3TGr1yZj2gIj2iOEJWyl1Hkl9irehRGN/BxZkxVuRPYcTwv5bwbsZqI4WAmXkvQ24BasybYNYzuF6v74gMvjA7Zt4PRwAbYBN+DqE3NOXOfE7o7dHa95136oObu7/1UAf/WDMxiAAcxtw3V7wPPY8M4m3mHPgdph2AE43AdeMCqCmLJoly3G8sOfdpsGMz98SG0i/3gpx/mYCOBVaUj2AaVkY9qaclg1CdmsKtWjPK8E2pamhltC1unBC1rC2qu16y9d99Wfvn1pKD6gmHvlFYxuiD6haMCwUap86NEj3yWxC0GG8XSK5NZKR4H4lthTS1jsALYSO5K1JF4MO0Ldcizvqz9Eq2FdbhG9G8yz1zYwcIKZ4XQ64/RwwbZteHj7iDdfPGGMgdNpC2KH4zp3YE6c5453PvGM+1MnXv+fG+juXmbwMeBjC3luSMneVOFm94ndBdlb4N6/3FYpUsW9QCLe711G27LN3f6DDM46orr451qhdbPdD3lEqsfjnlqY3xu8lQJXOSbvC+nl3VFkv3YdLOA6DblX3osC5ZVqSu2VcQ6k7x4NoNRZMgLC3mCS1eECiXX+3tBVYlfCXkjbRDM7ED1gmB7M6MZSU4KA42ndvVQwaKAblcuqz/wAA7Zt9RmnE8ZpwzbCSMfpowNlJ4jP6yz9IxO7ZeMvOD++wcN2AWxit4lpDvgV5s8AHHN3+J7oSvXE1QYuEpywFWIt2CvB5xTAyatNM5UqcBh8qWJBXEPMXohJnv8TWVRZ0IKUQl7jN8nwrGo/5FXpc48p2iH5Ut8h/U26O220O829y0g/9GpikPlTSnKRsovNxBZmo5pSM7YDkzPAx7yBUVnFq6rbAdOuKTYUwZZ4pncG9dzIwH2tp3g7iXwYNmwwGLbtjMvpAWMMXB7OeHh8wNgGTpcBtx0TE1c3+B5lm5PVGwYGtrQlvHR9XGI3A8YJ2/mCy9MXuJwfAAPmmNgBmL8D/B3gjvl8xXy+BqHPHXNOODwNIamCz7kQGgm6aXhVpzNl838mNARx+QRwO+iLfKxBG/oaa+qslZJZ2wjBUDjMhHkdFNBqrynaadNe0nz47nXd/CXaBwRHD8Jradi9au7xpReboIxslXAASspVpgMTPlazKPCUsjXXoWo12vCmGhXLFkZHiWlZkWW60tvS7812zIT9GBvGSC1kogROiSrn3N0wbGAb4Uq7XM54enjEtm24PJ7x8PQQ5YwJtz1iTyYwc3A2I3QMwTK2V3nux1fjC6jJwYO9pTV7ADPnwWWQCaOYJcTN2voscQmoQTtWdXhC2V6IptJdJPLLbWddR7C+BmZKMWoVInVergTEsMUTcS/dQnArtfWU4l7Oe9R7++be9d4ZwUu8Bosgr8THmIajPH21IdYoUFMDrUwe3rM9qOv77rsXceIOHhz60dl7PKLcNMTZwBjx2cYWvvNtYBsDY1iq7a2PdklIJvISN769PiqxOxz73HHd3+H5q9/GV/sV74bheQxMAwxXmKeBbpIYDWMb8K2BVwPgTajG0Z6Z1z246mxu2ky/pax7egDMAd+bs9+bnFZ9q0S6w2Hyak2BwcELP1kk1SqjjlMMIswt7qUms+i3LCPrfAlbFzVYmZCg6GvM77XrhYx2F1gHCb7cl127iLk1+yXqIaaC2lcDNFZDNQUztcADS7ALevIkJqJ+Ju98Jhs2wDGyjRvmvHWBWQJ8mOG0nSL67XQqaX4+n/BwCTV+bGGzwZwxjqUB2jozTdhMALsfcWO9Pq5kd4Sr4PqMd199F++uV7zbBp63UxK7YySBhN01B2Y7wba0xGJgYKhijZi15O99Yu5hlZ1zYu57a2b8nqRbB/ZnwPcoT/U1CES1A9TtjpJH54CscAbjWgJWKdZurLdHrix5hKndHUwbq/QC0g2035aFxoiDzbH+duzBobYP0nrek26hHMIwRt7Qz/hXJ2L3ivU7/5iBmqOWVe47tQeUBMmMThiK/YBDIbzdU2gE6EeVOX007rL9DmCktWgMnE4XnLYND5cL3r59g/PpjNM2cD6HNX76jt2vNfaLgNAmZf0TPdovXR9fjU+8K6PbdLjNBOzErQ3ZAHfYREaw6Vuy+koJtaL3gMbbspkMCj3PKIckSEIP6IAL5ZYzucRRGCuylI7qgCV3poR5Scu6ee6Auhpe0gCEwFVlaFvUPRK5p9jfb84i7f1OgkOT33fZK79uiuO8/U7xN6xIFa2l9IBjaxOr9MbN8/7p1AheaGZJ2XttXpQsb8aRc3wbA9u2RWx7xrdTZe/gHfQM5E5LDbZoLWzTj49kB6jvwKbDLIknVfYJBzyWv1SAhBnMdoYeoSwiQLkazAzbCGBSYpcmm3GOwzbYyKi8GYwmALqV5GVYo3sYAmeWNfcE5D7DQ0CvAA161AhuLkYBekbkCZvypM8l/lokOVVSkeqLTqkg1bvCz/tDv6r8zSAaQaMOh1eY6U0pN804qk5YEty7q+7IGxIQRiPxfpOT0pSaTsNm0VRawJduoJpCvSnKYshrSMkwFI7KXW4tB/bSyYTYhkGj+ibm4ro3M2yXU6jrpxPevHmDy+kcgTLnU1rwHdOvSeQThr3r8Ywr8NRwndhlGTiECv556foExJ4qp8+QlDXMltI+LeyL4ayDHo5cE+4YI/2SQ1Q0DiwXE2wbtu0U9UwvTNvstMQujwHMGQELPh1zOvY95v9ujomZ6tuOGHYqUdEiuvcWV49ZGRW7txTMipW8ZyQhU7lwh7sUogAuMjhKUkvmyaaSMyyKAPWjbGQpJrfVvCD11pT2KvoJCQsTokGSxK8wqzqKm7MkLxDd1iL4UPU2bkhYW1lZHBnBdwic8hzubttouKVg8Swl7GfxvY2BcR44nU44n894fHrEw/mCYRCH2RSPUE9Ni0l5MiHSkMBilfH3r08QVOP1d5U97OBKJ/XuRnfjc28mQZrjAKGDY3xOTOwh/UnsicyUrlxSS5++yBoRBBk0Y9ITG6VzBbPKfFmOITQZECFB5KYE5z27lRKHaUgIivFHOAAvjPZB2vJnqf1WxS9p672U7VKOaALFP5Y22Qv3r106yN3R+GU3KdfSc33CAUBK3jhIPr+5sRqbcsUBzcAPyTUJkrCPlY5tYAxgG1sS+gmn04YxBObCsNq+03A11CjdXRVYqv97wPxJiJ3dmbYS/OIVz8gg5lEhx2dGgpxhjIvMa685j5r23MYZ7yJ3Sw3MEPMmskgpY2SZbrGksObfHIzUCADHnHsRfH+jpg6eMcxRvy+cPAwJFKlIRsDgH+17lNc00ZrKDVJaw/UO+st3E1rzHNFEqKwwmc6pCYcF+7Fy7RvV4IC0RagMgFEqWvN1P4QagHBnFScVqWi3uWGoKY2rxR4R+R7jY5IlIxsPXMZGPhDj/ogVVBjDcHk4F4G/eXrCw+WMYQPnzTDsGgRei7+84LhKalkb57GSTvs9rD+v8dVP5Gd39YQtDWyBcRDjMmfqeVoH1cxJlmcNLZZozH5E8FgRNZmFBpJhGBtjoEchy4idAwBYTBkydntkesAx9yvc9yTsPQIt3LFPTx3RYZz3+4Rl2u4rpX23UQX+IpkbIAcA30qY5YdEdR1Dko9ylSsRo+6V8DoKbZF1t/d3tTLIgwPD8VsY3JbrMs6G0rr42ykdO0hnrVkYqBB8R+8d2uv3HnZzzCIGhO0gTlwuJzw8XJLYH/D4cIm2zZiqhQG31+txfh/8NNtU6NELctb08rkDMV6fLDa+AKvuDRxp1BrGpTR7S35AEC3TLELhgPiqKSwIz8d5M0Oh2tND0JsV9Hr7iMQahSgkGveY0zu8FvA5LcIj6hiceoD5HFwdEBkG3DkVcbCgG+laHRLtYFGBgmgcK4wbCvee4aDB9xRlHRxV/+9SMm6U5iNTP4aSHtqhr46lH+PjA1e6PFty970vf2+1oW7bvW75mnrBMxRj2U4RGBN+9DPOpxNO24bNUFO7duHegVvWrZu58HtwGTjx3ITR3FOf5PoExN7+ThsDw8Mj6TVcIi1osDEuLc1wWXJtBsyUdMs8I40mLiO28ISUbkLzjmSwOdLBj0pnTb6Rkj70NyBjmkFDziJdpL8GmG2pKXhsxVVFr4EhQDCM6WGH9jkxqeqnZhD0n1qNJzMg4mRQEkq1j4q6+wctokByB/ELhl4EdGS+96V6agwvTSKLUS9gWu6VZSUUVZmVhKxHpPxr7Spoy+p3d5Qh957gkbJdc6bmB1hEwG2B048PZzw8XLANw9PTJVV34DwcW/rOPad1zZii/AmaE6Ur8Jhqgn767MNIjSaDwmwRfLfXRyf2GJf0OaaK3HYNxgxXasC8NrYoKy0xYaSKWW4XcjwvybMwAl6CRWpUKveKhaRlE7o1HFwgNjCQtffZF/pNK0eO2hhWz7ue/jSpBUMzz7UAc8ImrbvJ4DzQwmdI7nqv4mhx3+kISN+EkLWXa9pVO6J8dCUmO8AWaAJ0wlOYrmpcy9V1HwOOblN3ZF2skrPl+T11IXrbbQnMSqkp8RMmGWL4HGo7WSBEzdQGbGwZMHPGw/mCbRu4nM+4nE8wODZcMXxPvp0M/MgUfWVyBGW41Sw3wqDoz6/BKcSPGbEjB4fxwAOxBVTRo9OF0oRn5rEtkTv2RPTA2US0ms97LYcE0J6qu/irMqNfaTtvJKBl3TCUjo6sxCbcLWwHS7hnSHv4xKwYgVtilwpa8qU0t4SP5UZ9xfVHajiDVmQX9Z999NQYvbQAhniGKiOLfw4gWhb7SLtV/Y20Q1/nWC4dW8GqKtXNO705cMJDGa0HrhJf+Y0vT9bWNyM6Ft7ambEgXZlX5QcemxnOpxMulzPGGHg8n3E5nWLrKBjGzDUdWF2vdkCAbivfNGOym7aSAWnLX9Ck8vr4kn2E1fu0nXDaToAbdg8S8ukZUYdc/BKdG5vHVl/wVGuT96UqW5wZlG9531iLG+Kli3URbJaEdiR0lzIoC3rZZM+9w58+9wORmOEqQ9HIgpaWOa0Rdi1pR7eR0mWkYmuGLY2JEZ80V2QGMmw4iHruO3xeU7u4ltbjJWkAcH82GiiP8Kj5TkiTJh8lTLk3gaFKpaO29aHX4lBfmYry9iML44KSZrbpRoX4tK1Db9rS0RodSdFhOI2B8zkI/OnxEU9Pj9jGwOV8Smme21LtATMbnl1PNmVK8Cm/OEUtQXGYvNhsZp83gQcLGO5en8RAx+ikbQzs3KkjORfn5kZumhLM0rUVOJJGipFStXzmMWi1EaRB1vyinpWaG43BwcwNE/SN6yjh08LgB4RzwCGSlddc1VIdl9rI0AaGZ3hFWvmlwZmJKqPszzYMW6qP4RriYgy09LSInHb3VIRmMFI/zINLlSVMoi1lRDz6sW8ksdy/Rrw6BdApjZbh+lDJ94DRNwguW1Ld0UR0nFszAUKFWgvjFOygL1Sdw4Lgxxg4bxseMiLufNpw3jIyczrUwNq+/qMcTglev13g0P1c2yKMk7z5FYL/qMQe2+1seHh8wNe/8TX81uURX03Hd2fAZGbEWoSiivHNAjmnO+y6Y9Cnzlh1MdpNn8DkPvQdNhtGkdQEyEAAmOwZTippReowD2YqIqICdtEiUkochq6esQ7yqmzrnLPzWA9rW5m9doqNcMpc7uvXWArpULMTqF63ay+0pW2UbTc1Ga8WB7wIEwN8K2bIiMXGQSWervUGZFxVuKTn2B6Zw/tFvNLkwjjlfUvkboxa61tu9/fyobeh5HBii1lEa5rh4XLB49MF29hwuZxCwxoAMCM4FCT0EE4G9JSMLSsxfRAI1DBMWkVkpnfERwsreoReAd/HJfYMMnj7xRv8rm//BObTF/juPvGb19wwbwJ7WinnnNipes5rbWDx/O6K/RpEblzK6hNzxrES+5zY92tId+n73HfsaeEOQBK9RxOgC2r4bEOYqwEs+wJkFAO1hCaXGjJhCuYj3kTIXr+u6QcQsdBsWz8HazdhgCORyICrRG6V3EiJ0widRDBijt99Z89GlO+xAq09FlsikgQJHfbwY7CJUSOgImDks3sSfKRviSthxrXq41beLU/FS9NE7e2bPqTXkdFSaSAuUnJOldKdekjP/Nu2hXV92/D09IC3b9+Wm20bDY/r1DVoGY5N5YHtJtMZZCoQwdNTCDIJ9jVwWwk8bTU3Npb1ei+xm9lfBvCPA/h77v4P5LNvAfh3AfweAP8LgJ9z9197b1nI+fppw8PDAx4fHzD3iefrjqsD++5J7MC+hwR3d8zdMHcLS/RuuXQUIfEZnGJI49SehEo/cBv8yhdpA7W+HGnjJII6lbuRmoHGRre46viSRqtVxUIa5lrHsGMq0UrD7kBdRJbsVnsQqwPZyJlrnNPqTuSJZoZRyJP4bwi+tL+en87CMRJsEvxcSag0gmxcq/gobaAQ2TuHpxahfnnOW8VWKoBZ8UZvbHmocnEdESV4GZR+Wx4Da84UWFrJj6WaDYwtNpk4bRvO59gbrorz0KSm4oG0I/ndYScgnWZKx1wJ/pb5+OH7fdeHSPZ/E8C/CuDfkmc/D+CvuftfytNbfx7AP/sBZeEEw2UMPJ42PJ03jNOGcdmweyy+32lonzMs2440KsWc8/ndjP3pPBJ7quc+dzhmagTXXLU2sc89mccV+x6MYPcrpl8bcZMxcB8L9zBoeUY4jSXARaRohbHOVrGO2OuVWCS6DowiFodRt0325jXHAW81pKRr/R6Ouc9wS8kyT6qTUcIsI2hrDs2MuJpKbRptS9iyEC8ENu6ZXP1IRcYtpleWaF0aSkZYkGlVzd5pF4bYfe+ArOXJTbqbJ8WlI04iuqVScaRGZ8UQYx4e1vXL5Yynp4dYe345YWwjXX+OPefmEy4COrSrEDZbwevoYl7ZVNx7MiN3q3UbVoDmL46Nyef+9V5id/e/YWa/5/D4TwL4R/P+FwH8F/gAYh8AzgY8bgNvLxu+ejjjcQCPI3fagPVuG87lqmgrvQPX3TLS0OF7rEzzmWr8QuCO5+sV767PmO647ldcr89wn/je/hW+mu/SLjBKFZo7y53w52dg57w4XR+lyoaU7XmoCQ+ghwDNvl37YypGUVZeCkcAaqDTU2dqIEssOKkJLV1T4u+iEXrnscwGoAI1WGQjaJTHQKfWCgJx3Xp6oPN40zqAsknMOSpugGquLviIOIl+19OiAmQjkYS/LgTe2v2BcHSvQCtCMd9QS5BzH3riHTWHIFLDtp3w9PiI83nDw8MZb94+4bSN9m8j93BPTwjXfACpouf8e5iVzUUxYCVRk57l0hjrYBsXLY4lNeP8IdX4F65vu/t38v5XAXz7pYR61ttPjVEdY+D+ZobTyPh078FzGh2A3OY9VMYxgtjhjv060ujm8D2s0XNOTN8CgVK9mu4R1GIhYa4j5vbhqiL3REiY6Zg2sU2PVXIAhlixI5QWcAv3m5exhJxpFAE19fI7ex+jhmWYX2LIS7DA4XvVT1eOQbHKusU2sNgMpHlqmyZC1vSjpKhKkUPDrRlJiy4SrFWTFCxIqddM7XjdFe1N2QdpuF79u1XdY/rjcyG7XFjD6DhuOLGlGu/ILaNE+1phSKAk5Aw3dfSve3Btgm+220nuhSq8dv3QBjp3d3slTk/PevsD57Pv7rjOK67Xd7g+f4V9C8KmdkV/cXBaS++brGevfbOBM2YLtTMl5KijeK77huf9FCp9hp1Od7zzN3g3rzk4Gds+HdfnMAqGnWDvpbBTt7mKcMfdw+Dn7tjnNVe75ZQipe3cZ4ZFApAo1l4FtCKaCUES9ScDhYp7A3BVtpMcax89jZNXAjoOkZVBsqlvDQbyLIMqvU5bvKSikq3ce79yTaVz4+bsMW1J+0Os50eOjkYyUgtp5Ff4zVS/FyZmhIIEKi25ulkGK6EwxsBDbTYRHqTzacN22nLMdeMSAD7aEy/zbt0aWz31nbTtJmKpgc7DQ7rH82GjV7cZSqW5DdO+vX5QYv8/zOyn3f07ZvbTAP7eh2RyAFcPS+X1+R2en78Xe1gUHrPBrVrFrw1xplauPCMTGK2+8XkIvADaPtPgh1acHMAzJq6xeXWmHZhz4t1XV1zTaMA55PTYx87z+/oclv7rfsXz9Rk+J573K67XCFTZ9z3ch3Pi+tyegzCqeWO+SkA2TAKDOP8rLZ2jK8K97ARUkZ07kbGCFLU1H9ZLiKUpEXpHYl8YSsKsNtdYKGZlMFZ/Db3nctzzHbdzbkMjuTfLak3DpVxf6rZihceeuq+zpnVy0h9H4tHJsI0wvj0+PeByPuN02vD4cMFpa01xTlWnLQVVbOVc7jJr7YBsueqlFusDZhtsAWSk7Rh8woqNTO3RvJ8Zbso4Xj8osf/HAP4sgL+U3//Rh2elyqP+XeJjgGINarF14H2d7yiwlxhsxJlh7Xq0VL0dp+V43tAgphnmJqgiq9h2S43BqIrmlkOp9teGFu4YY0vjnge3tz0Rjs+STrKNTYdBlKGpzH4uTICq4oKk3v0Pabt62u9eJZDtflp/+VE1goWQSiBMxTSPjIm6jzhm9/jQ3ca2FiHLpQ4M4Njjjh+otEnwg7zCRMEfQeiqqvNYZO55UKVWZY2Nlt6GIHyyM23t0g0F1c3z9QGFGLoRNyG86ZN/hd4/xPX27yCMcb/LzH4FwF9EEPlfMbM/B+B/BfBz7yuHFxVDfkJLttWYlDdubZzyUoJGak+MUqfhosHjqcYzUKelyxbFD69NBypYwh02Ji4xeRfnkmGetpzfD+xzAzehiIMrwk24X0MC7jtqZd6+Z5DQ9J4eMJ/nTrup/kdAURgZ932PdfGINJ7pvbbJBmyfuUuug+vnwVBZBnIUHhLqt8i14IYiWyIuCaRRj6SB/j4yjiOzkDiE9YUQ4aJ9+M3f5arpBIvw5jk3aZsoTNKYzSSQiH47bSdsY8PjwwMu5xO2LdT407aFj3zMWF0GK0GgqvNAGjw9Fl0Omdky5qO2JjGXGAh2xNGuwBSGApOOIuhxbe3BDgN5//oQa/yfeuHVP/b+4u+Ux08KgjS0BxAmZN7q2W+iWW5BKSGem40Vn6gt+Ow6JgHGqYAhZgSJKVR/HDiZ57vYj47RUzNTT8Mb/AkAACAASURBVGztP82QX2Qdc08Gw6A+JBND72k3p6dnYK/v8BbMZBjhMrxen3F9fk7bwSxNYV69tsmGTSBdkHMn5aadwQztrNH4OP5xdmHRGupaHAAHdaIuIdhXpbP3YANNqJnXD3WrJG7JfKi2mMkqIr361sgffZYZPokzpwzDHOdTrE47bRvevnnE4+UBYwCnUwTKhAcm1xO4QTd9ZDDPMM+95OJ7FGiIkw0qM9UUiNvCOM16nzvyAOkTqYKErtOG165Ps3nFcnEOiGXgVRboNyV3+I9bVk144eUk4TvvW2VbcS/z5POKuUki6ME4yDSjRhH1T2SYPsLVQtV6WjCnmTH+07z8zdMzki3DgPcxsI/o0zYMW/Zv7rM1gS1jDKZHBB2ZALUDn3E4gTPCjSfiziL9owTt4B0lnEDqULbDMBLK6DhCosawiEwJ0fXdYTAJ2xte0+lXQr+V5h96FVsqqW5JcCOOX7o5hSUJqXqb7kHQMMnnLW8NhcnfV7uO4IgvMc+JdkZV/aXrfRD5+KvenO0PVhyWcPFxJm1Ni+WswP1OWOltomt6rBNud5oSai6eQfrTi7PER1bUhhtwhGyckA0z0v8dc38Uk+Jc1NzSIJMy1ciIDHt6CSYGdq81ZnB/CM6fRj04sM8d87rnc4AndV73DCFOTWLm9laTkYbziuv1qwxIumK/vktj4o7r/lzTjzr5s/bLA2AxXTAAmD33JGitUPmA1lSh8ikWF9qCyoe7fs+/fvPKD0WskhyOZJiaaVVpo7WpweWRYoaB8+mC0+kSS1IfHuskltPJQsU3TtUCxuaBNGEYQxHelobHtgFAtAxUFCPga7sSF0cy0UjshVMRjkE8oiSypT+9O1IPw2sE//GJHehVbkAQ/KR075Vk0/NkV1AVkvwJtDkbIWfla7UbaE0vgHS0cAPc2YUoTPWrghhcN7OMZaiU6RWSK/9ia+CxNNQRAUOTn+JPuaFFTo5rCpPSnHk9Gdf1OnHdV3sA3GvXk31/xrt33wuvwfUd3r37CnPueL4+w96dQsJP2hrCvTj3PZEsvBPwtGcU3LjrT5uawiXaqnaLpAnuO9S4rZOIUpsOWKGEfvuuvl6QajcGsMPb1sRGWMzNsJ1OOJ/j/PPL5YKHh0vOzQFL37mTwD1s483ig6kP89pcsjU+T7lj1S+OteKZMrLB1NJVFyk+ZQHUMb6A9qn3ETrwCY5/ug1wpPur0wAhFXWeI/bbZptQotV6GrBauYaFrJl50Q+w1k0twWytOzi91pLEh7VRvtyplAv/fvcrn/uMfepQdAZHBnhaI8JMhsJpyIADpxPmsDzhM1T7U1qXyxCY++SFmzDXEuw7PAk/jwrNLjZ6hr10ZXSYYWplf9izZXMPQqAIPZhKKQpMU9qWZIMUc+cyP7w7DGlsIhraVhyceIJZ7CZzOp1SdW/VvEeu8ccNslpNyZU2h1Vqazx9iSvPsaLYFjWgarIVD5bOEyGtgCbv7sPmeH3cVW9AImWuNUtDB6VxiGhGqKkqJEEFnsYPR+3J5goM4Yis0EC11KuM3ujRCgmVRAvvZBWSLw7bRs4at2QKO70IFquTuDCiuTxKa2AYMEMzD8OIXoUVvzeWkcFFoU1kmnnCfk4J7hP7vIYEd8fV9wQZj73OEOLcyGI+P2cg0cT13buYRjjygAzHnHnvnl6O5Av7Ffv1moyPBotkTcKJb2Lf677HqBCkCH72ONzApZMXuO8c+jBy4YoNw+X8hMeHJ4yx4XK+4Hw6B8yHZ12pIdVw0gfEeIAYD094zxzDMpAxbf0lCPimG19TohJmdqhn4R4NKacsU8JPjfjHaYkrIN3m4gynZLcK/ippSobPPh06Unh1r4MWfs9imlA+23l6TQEJPr9LyqIIvreH6nprBVliHr0IAMDDJyLpbH7E9s4Jv+6t8Vir//TTehrLTGFmDUdKLBuxI61vGzqGgS4fj6lD3u85H32eV1xzOfD+/BwLiPaJ5+9+L4KHZkwd5gxD4fU5iYGEn9Ccsq9ALaxxL3i4ixrqjdro0YC6AGq6RU2gjH0rQdUd4QfDwi4tGMAYQezn8xmPD48YY8P5fMb5dM72P4fKnuOkgqNdugyOoU1mldplPDMsbTje6V8y8dpksv6sOVf+2G1qFvOC2nO4PjqxtxRGdYBooV3wpYfrLcVjddF6NjlMIqYk25AcGmlVhj7iobRDGpdSRJHJWhpZFNRuwswmZbqWZ3IPI4uob0g7uHY5FInm4H2+VwTSjKyMW1lHeySkk/slMkotrcvDvDQFwwaHwS5nbGPAp+O0MXwYuJ6CYe08EsuB/bxhv56z7iZmhifDUdMGIIyQFZRS3N1F00pPy8IQUFOV4pf0yBBhahgzyCktZmMb2E5bxbfHunOaxKYIAUWwZLQqr5dglpWpGKAnlEkJ8tv84CZLpi3rFPwmH/tHiZJfNyb5fOF2U4JeH/l89vx29Gq1mFnW0NbKKYZ5xsi2lAcNRR47RidXVWuoKoczV5spGKg+R0Wjpa83Ua3M5VZ1UCbchzOmGSfLiWW2MVAaI613ZZBE7jhKOsgqcw0QOJhsmyXRDxhOmLVNVW0rbQ63lFYDsI0SYYJRYxtmG+POG2KjCoefz6hov/yeM+IFwivgeL6mRV+t+76XZ+D5mloDCX/O1CCuFSjEMGTkEmUyOc9RzsiFIIIKCaZnpANpSOSD9DgM45QbQV5OeHh8xNgGHs8XXM4jXW6O4VeErOaZfSJujOZasoMk8mQkfZCD4ETdt4ZTUzPrE4cUCzzHl1qqCnZuMtIajiCfELfX/evXJ5DsRGaqyj20yi9TroEpFuLLi6rTALANMXQlcGhEf4FfSqM6+MW9Lc7xLrmqN4xx71t6GPQhJ8EidtCt+quTIkG4MUX2fWafp6NckNAttNKkPzBg2MHVdnUkVq7IKo2J0V42i5BGBRYEIgKpMWxILabhxHM4HUHs71KlV+m3zzD4TXeM53cYXC8w91yGPCOue9/he+4kNElYZCwtZXurZ47H7L4lB3cZ8prG5rIKM8PYtjhbjSGwY8TWUliXnchyveUvZ+2io7eENhl+u4MKWUJJcfOanrEOKjiFZiXEV1ymxd80ozUOadtfuj7NhpPo89XYgePnqJFMJQQL/6SL+r5AXW57tZYCYlHkIaDtwIsV1dfsRmmebUrJAoR20HNMUc4MuQ4cJe0pnWi8cvmj41lbZ1CqCSKEQSmW2sa6+3XVGzeWNNpGUkLWFlfl3kwpwUdOJst+hQyEAzbpFVAtypvAMm3YETwlVyxHPp0sVwNOzOvWqwS5EYm323H3iZm2jggJjr7VeXpCsGbIU1eRc/JT+s23VN8tfedX0EHYY63M3wAutiJulCE0ntYyHpHEimJHTCvc8hZtjIobhRvMmYWafGMt29jh2j+ANh28en0aP3sS+ma5rj3fcZ0b+21JVKkMVofD0GS555fqAvmSu44kq7cKspG0wsmNmoUBMZdFjSKPaS5ypsrO52wTmiia67YqR6nDI9+qHYok6iwV/rJlUVdXacoFPI7pPMd7wkDLeJJvWsl5OOV0bpmkp+skFBLXGIsfC3Jy+548BIE+5pONQtBet4BymW7nEfsKFFDia/KAyzT6lTTP4KG5zzQaOq4zwohjKnDtqcB8xs7IqCL6VJOR255dYiPI07bhctrKLz7nc+KglZq8ag+NhZbn+g1ERCOZySa4vMqBVu2Jw8Pa4x5GWgdFCZnJVluIt5Ybm+cQ56wCtAx5XEBxiGY+P4aSncglkv2OhCexx60ATLWbSkyLKKtoaRrc9KAmHO9TlTByVXm85vHy+7kVeVe//JAU5sJTvCU5ArFN2ljNZtK7oJOpjzAZLjwNionZP9E2vid4eETt2utZobrHYE1AZCzpmqv9y7keIeevRcwNRrjF4QhkQeo/ntQapgQFZdPhwHXfYc8D7hM299w117FPg+XWY36lDSf6QffXyDZt21axBfSjG2FUQG5i4W8l2kWyy79R5XS3AJK55pKSEnZ74gRhwfyDg45gxgXXLIDNNOJG4RdEgL1O6MAn8LMbuEuNrUfNJiDVr2m542nMRudCVsMhO9sEksb7gdjrnPeJlI6yYnLn2C4NxTSsX9T7CrWxll7Hy2/uVR2LChZmUCpBSv1uxMJEVNibWcEHYD6KY8KtU3QkaQYJpeFQy/W1UUmIjK8H4Ln/eUU6snCCfZaNIpZ8kZ1NDPQOq6xmpBchpgte92zbCZbqf+KDN5OfA3AMzGGYaUzk/N0spK8BsR59bDk3B2zf7/D6lK0kei5uoQov8/LGU69nDbEmtsV/YNQWqrMCCAOnCpGPu/wmhRBXq4wum4yGA0h2xA03XiP5jyvZBWjbiNDEzSLaC8hgmzJEpCTK1US1R1iqgYbcLgpQMAA+Qq0sM6043SgVQWdLYVrVEOmo3raKC/l17FS5zCQVGYJzqtG0ndmsBx5URka3UbgHjWyUXF0w0anYARj7TSSNsr1W/8Xmj1blqn5izgiv0ATCBtBixHeR1CmlfM7cojsQ2kZG0xk3s+zxWsSUKwOiSxFwbNhS1Zsw7CnBfTO4b1VPj0bP24k7xUEcMdefGRJtBttGh/oaiSjEjCWRD87RC408TmDFrWGuBBD7YUBvvsYUXsKoGIxY+otRyJibDYwMVZ76yoPBLvv9WW6dNbZV3Thcn+B89hwk049Jt5O4rBGEHVJgDC1uhRMAoie5o6W7JsprWS1Ss771TRYOOvuO9RhwowmsDV0NfZrm8J1IrW063neO/uuSLuC0wrJWsFHzOXbCGnh832Wuqm1pwNowJyKH3557yTWUGVtOdnzYAPJOJ0cxEyvNJ4N/QW0ph3JZh1dTcPci8OlWEYNhvynwQW6llwk5a0LqNMru7ZhRmHcScLVzZexYRqiZ383FebrzT7ThBuVY8mtiHR+d2BPTMuCjeF2q6s1fyUV7TthzX4G+RV6SiOd7xnOzSk99R9PVsGlxJTW9DEbxYlRZDWmrRRCcR0srKhmdR1yHz+irqlWIvPpWW2l5ER8NOqpjIJ9H95opUb/jnnHTPBYVOQWrYIW3ZdjIu6a2LW+cDDcNecUVWk8qJmAUvjR+dvyAAmcxmHJQXNyHyUgMuvKwre7RJmv/OoRZ555lgzB3z0Cb8L+b87itSFuzcuNa87bCF3FTWJiApvqVwqSs6JGiF7BkVQWEwAmeQFu1FE2b7MqlwKvhiL6OtK0MR5+se//6BH72/kdrNn2PYgdN10g+99lLLRKQDs/VQIEY5aWWwBMKh6i3CZHho1EnpT+a6qlylZsr0/DcLvc4oLKIpoG88ANHGQgnAtECwRlyKWJWmdEwWfXWBF/TDDa1LOnQp/U1tb/0HKjxxw+uq0Tk6IPU5I1z7VUkgdN4J8iWDKJmrdx3n/SbqFqMK9c4RPne/aoNClN7oFFzNDFy97YYigg8dVuJzHwTZOjPYoCzuNtI5EXoLemdBE+i5/il5C10EAvrkfya2bfA6ZNpGnFsJiPndFAYhnqK3ILQp004d7p94fpkm1c0miZHM8KokUzv1hj0yGAVonq8yHqPKjRFjgBWyyzunQ0qKzOB28mqDQdJXuRfzJ39k3YuA+7VnJ4Kr5LAi/dngIYoApDuODNkyf2dkr6VkhZLVYA8V5CSSCSv7oLLnrit2ay4lzZOqpW6yTxumtcQWp5UqmprSu4ESLNgsYwX0ayEzvroGVF8KHl7g2JLOM6Rxy5NXH768fE93O0WaR/XPjnqaHLTYn7MJHvIpwHYBre4dxsp+VZFtRFDJIsOiOzc2WslVsm+EBU8VaeKi5JpQxt4ws9pgI0F1cwd28HmABgwR9PYbKPXMIMPK8nOkMty27nXHJdGpyIBqggV/8OpTjMHUpjnUU+cI9c/a+3AM/y4ZLmhmFmUPWp7MMDWRUguRE2IJhNsFZSQS2kvKvpwk37RxVebRAP0/Re1R+ppjklXp8xp49CExnHO2ze2xbw2leCUjH0vEUJcM4ax0vhGBs/7wJfaYA7dzIBj2yD4XO1JSo9WMfkcAL1LNsR5eiE6lPsA5hH+jNzcwizO4eQW6y/xD3zYhpM/izj66dvZn19w93/lBzvvjVZXJXjL+PhwQfSAtE8ZaMCxw7DY45tBHFbuNrU2o+fdifBRRGsEukCmObzHtlBI24IEbrBczu8K+3X1Db1Wo7yyNWfPkPbsX39Q7QLm8FZni+vJ8l4+SoKdA6D/vPfA6XXmXgST6fygUcAwK66spZ9TG8hIRfEitpRKy3I0hR23GqZF1hyUjZboXEPAcYrvOXoq0kU0Iy827q2G0+5TnovJTVDayAodP0s38IYq0YS7FVMfgXMFQ8GY40X8YAo6ZBhEptGKAMBjxUnsAeP2vDgoNKIdPloCegqU14ic13h/ElwB/DPu/gcB/FEA/5SZ/UH0eW+/H8Bfy98fdJX3hR3JZ8vnmMnqz2JfUgAtlmhrBOBvIgD/0cUS5fTALXO5HPHaSbQkQNbP8sTIwyGuenKOHn5yNQA142b4cPl0cRuLsIQY67d+pBd8aGyP+HYFUg1J03wyXmhtoHiP0UouZYgE4pgekfBQ4zq8N0Pu0q9miBWgdAdfmOp2ZG15Q8mp99UOO36vI3u3QwtOHuC7gLOnp1o+B7PrY5pGtoaF2BU+jM4BfNjust8B8J28/3/M7JcB/Ax+wPPealAspPIciH3ZGauNGFAXsRDGCsXkFYC2lJuLOpx1ZIkyIO1fVpbSgGuFwguokYvrRL38wgBieSj1SgdqxVMrdMndU1sYXcakwUz8KSPbXk0sCWDgwYSeRoGawrjVZtHLNAChaPBQqlrSShhzNbVOoWphBjWL1DgWaa6yld8oxOxFLYSLEluOA6cJYNtbO+GvKI5qvtfYsw87+0XY5TdVeiP0lZEBsSAnAzyOy1NbNrfeNxI/JnGHbaeqz95ZLE7qeJGYxkTVgm8lJLKeSt8EXwyIKnqO5kYtLcf/DMMZhpPkuXd9X3P2PODxDwP4r/B9nPfGq6RDfriIK9Q1vm8XTvHGpf1HTsm2AanPVzIGOVQCLhmsGGVVyRrQYenvXVd6X+7Z1v/ZKqXB4lw5D8Qp4oT13LXC9Wd+0rpc6qxAqUzi+dN7DkzVbmaaSeQvvT6VeDeMJNo4oNrK2EfV2GwHuZR6OYIReavahnBFCrJOgVUbvwLOgYhN7Ix0FNro9FC2Ee0uViz8FKXq59jEyNSqBY0J6BhCCglLuyxxwIDN6izAbKQ0ii1qTXCx52AvvsEpEowbjlhpkQAwpk4eumwDCdhqX8NqLwgj4nKr9I2jrRmfEF6E23Nl1uuDid3MvgDw7wP4C+7+f6t12f3l8970YMdvj3uzBvJopWDeJ7bp3BiN2GqZrnxZmhdI7nc/st42mTK5jFdm8kbKXIpPohdGYyV+pF9H6UJOXEUf64jvZaFDUZ9pdw/dpHTIOGsX4eoKZimHzUpiLOKoMUAzzhWSDTULaUPfP8OAFYytFCjFdwMtmTRKqmd/0rBaNTaoAWEmS8tWMN6OgwLw2LnSdGSMFuguDzoezprtvIB5gJL/Mie9k0OawLoLdPZKvjvXBxG7mZ0RhP5vu/t/kI8/6Lw3PxzsWPOOCG4HzNb92UarvvUw9NCQGLOBbC5zr2FFLzRANWCsBo+0Qs99MRVry7LDYT4kqCYbsrj6vKQ4gzRCsufcuMRWyoO2aKFEnMfUQGzwAGugMdFF8d27f40wS0iL8ADHGLNU3OmxOGbSOs2uEzakyKS7kcywooZTbVTY80XHQIhU9eaXtTiQYORYZ2NJ7yHJh/Ym1x1FQIxNHT+CMyQkjZfc33CQacFEq2p1uPmsFUxD8Kd9pSCUfZktGgYPGWwgpj3Fk/Bnj5kKRYn3U/sJJbyWF/wwUw+rFXoM+umYkajDbbUR37s+xBpvAP4NAL/s7v+SvPrBz3tbBBWdNYJlR6koTNQlNr4RlXPZJCxwL/RI07AM6VCqqnnNe2rvNOAOl48/lY/ir4cfyXNjGegcla+bnps12CwXCV1Qax8ZoZXqnrVkJ7Nj3x1UBsShY2w+5+KByZan5Jj4rAI3R+VTCiOf80l+1e64glOJ+2RSNe/MBNTDB7pAwWmXB8XszZghotyIF7PhqQpVZLHinyoIW/AJUxFpWLYGb4IfGw2ZchWDpQ2ne0DLfWxMgZtLtZOGFetOQVVjjqIDPmMZZMZ1r0gzyPDv6ap9fYhk/0cA/BkA/72Z/bf57J/DD3Hemw5zd50szWUk6SYSqhHglfHsIJVLhdcRR3NeedItWvTC9bU+ShKsO9KTyb0iVKFyzoe7vUFoppP8RnmpU5mfL4hREW9CcBWPnfWziLAfoMJm47VpUwvuHV0Q6+XXWSdZTUvZQddR1S1qsTHUuRqdJzJbx8AXmq++8NKYwEg9VN31XfPkxKOE/xIAaCtOrYz6QJB6Z+tjPisDJeHn3SYmanxey+3oIzGI8o91yiNDu3uVSvQaiff1Idb4//KVKr/P894cqPCSODJhIA0L5tio6Dji1JPQnDFFlbXRq4p4wkdcjD8PFd2J4OmTjBgZiiUDN2WO9NpfZUJN1OqbJ/GgUJP2g5xOFHLHs1bTHUjJHviyxcqmikhzGXmsS0ANZQwK+y0RpRGGG03UZozkE04SDoJwOdTSVhyFmMgzHsWKdjVYVw/apOrtublE+6lTq0jJbg5RSdkhy3PvaDQd8Fwaa5hhRAQJbDbzr1bTkr0yBB9UcR2LoC5MdhiNkzAw0RAGSPyAyfJW9JQRQJ9cTR5yEDBrxcsibgI8g6IYVINsB8D97pc1Gw5wT7o6KDS3+3aGFr9wfaKtpFP9cQKQIOciiLg0xrw4X0pGI7GD6qxaaKma0wDviVcHDgsSIqtobnwr4IlQnadMODWNEO5f6muXwPZxgU/NKWvwmDSt4eZhBS8mgJxuGIEBSDVzIuLJ2e5szGQaU90BBXt20viQsTGpibS21K5M7svaTDDn7nVK54i2mCFOirEkGl3qlNOfPXzm4V6r1iJiA4jgni5JlGchjT+BD57EmzhVHh8ynUV6r0SxWLrLaKnIt4K81kRwbI8K4/LgSHxiD0hYqncIwmyWEOulmA6MKhenyJOXrk+0B52VdBpmONnANA1DucnQN8LU1WlWRhiJnkPirY5ND2s8WWV2c3Jl0tqeUbnWvgTBZ94bs/VrI9H6Q9+sfa/S6lgmlx5ku0uy9FyPNQ4kHJxs9k473NN6rlDqPwaVqE2wBTu3DItNQjYxUo0miEFCFehaMvGUrcJ+k/AJFxKlCyHUZhNt2CyNxpI5qmGxYKpjiGbWAu9lLQTLRJkP2mDprKvLWJDGCcNDO9gtdL7a3w7ijkt7VDVHcCCGnBrATcDAcn1kYk8iT8TYMHCC4TK2NP6IJR5eUoSQKMljguTyDGaoE0v4mLHOog5RrlCC0rBB8umQSwC+ol/bD3pEy4hmgOKWBu5M7o5jrppyIV8Zya1NllE2px45yHnevD6fWe505OmvQYRcFhobVgi7KWktATapaTkJ56CpwHu6U4ZFhEbFjT4GWeEBDiuPFkrMj20A0Ofee8bpOmJ7qmhz776rDKgIrMgmADUHQDcYVwuWJsXxFdavUZi0qXiGHPeUqDUy9AgJfkXzli2j3Us7KfZsqh1GgczP5bu1eQVTTeJaCzgzYAzPM+e2OCwEL1+fSI1v2TAsjydGuiBKMpOdowZ3KUDkMeC9y09qNeS2TCJKmxBYiQyosgo0opCAOVgH5S7yuhUCVJWZj8RUA+1dUAfrMH2XqsrBIihk3p7sio8KmYsnYJVEljlpz6g4tapWNCW2kfkMi8ZC/33lqbQytWnZJB3IdrNh+YQx4448hTfVf7eRcBnJCJdRzKID2BrT7hXZplqMVXph1eUjj351c/dlHtn1ifZeJEvtpDCqknuNtS1wFsZRaYXpCPRuEVnsB/xXpwe/TO4fX4338oTCsMF8hMWVxiwSFlkx2kUS133LMORTxIbu+hAJTeu5I7Yb4ug5UMbSVqHioaB51Ukr+Mg5uy0pVsm+wkDL75KHjKeQHTPctKH7rgRHxG91WTUTSimib6mRiLPHC9GcnSd+teQqmjIa8LzGr7Qn50SNeQ7sVhCf42GWxlgykiwrhIBoO+j2WydDMWeINF++lZC4XYoBdd8E5Aj7AnfHeRn2tuRb4FdvsrWlnaJ3trWV6Xfpqm+qEBARl3EfgUKvu92AT3I+exC34YSBU0r4rTF8AoWChlLla5jLsIZgDmjXPGE2Uo0dQBHhhtyksgadanJNKcGgB5VQ0eb8TiRB1l1PvcMjiXAMq+whSD3cO+/CIBKhe+Cot8sQCgNsEYbcVCN+jmRCJg1fJLioqaynPBfVXxMcW8umRmDekwBdVRa+lVSfi8CNAxt1ZJwEQ3IZd15eBycMRwUjcU5OBdhcPdWsu3eRLV+BaCcxvSBDqONFhfBpje+4D/XALDaivFYWcbS09/AZ+0A8Hc081aakIz7VNVRX4AUhHiRBO8ULwiWvj35kc1yqSAnpOXrErb80u98pbjWYLayhPy7PDZWDfF20ygXx1yvfUBMoiWTQTQxbAL5cyg3eJCG2n77ZzoJkRYS2ZIU1L7EsnhtIFr5oc1o8dJH1Ltohs5wsY7TWlRWv902xBjak296E3iG86OyNDcEJunCq5IxgdI44NRpDY4F46pXQZLzZmrsaIeuvJ5bt8+X3gQLrd6nSblIGU8w2L/FzIOZbUl2nq4J+dX9kEi9dH3132UI9GszKGGI1qHqJIIzf7KxKH3OUw9MnRroyaGhreDbCFYGsaFMSr1kDwBV4hinqGartsdCEtuQhfezBpzqpc1VWVWgoHN5qscdSXSN+AbSpstqbj+Lk1iAsXbHGeeEJ7eRaNocQIknKiz6UVimuQ08zUnWO2gIL8WxQGNjUtqKutAAAIABJREFUaFdW1ANMWK/Bay5vNmUBTyu4RJkidFcSjjZTcyjmlEyacCTRGTrGHeC0qqd3TedHccKxScZgBbG23FffwktBHUBjyJRgF4dOocZAn4LU7E3Z1mvXR7fGeyJg6TNgOES/j1vFuP6eJnN4qoHJxkOLmeHfdc914cKxKQiSsZj30lOUahoA34WDUp2KrZVbOyCzmE7dIB4uKn0alobsnMJNAo8jq/pJE7kMo0kUmZF5WLR7rhx+AtgHfyu7QxmLeP6cQw4XxDKTCmMH2hjXRJZ9zf7BAbeZm2BEYFMdzpmqJ9HfSJSct9Idb9EXS8vzSPCxfWQkUOJlr5KAhzSbtoDN+F6gmjDQfQYULZvRJfGOxgciwCIghFEHHgu8k9nT7lCeA9GadLgrZjG5QAjFO1MEqU/x+6Xr0+xBJwBbWb2v7164FsWw5ikEbiNYW8nvc70mT5GgMgBH2Ln8Zf3NHu6o/tal9Jr4zh3FNSEUez+0kHvoUwK0wcfb4i8Tv/RctRbEaiq/fAPCBBsmSzuzznonErKs9EbtoLjRoZtiqEIrZj0I0lhBiZLYrFjhIijzkmRbZYU1HKWNJglflZAFuJUpmzAFPlOMhIyHHcbliGfK/7UdJne9IImk/tK0c70+/u6ylvyZAJLPTYvlNyUUV4HVoLjV9mApDxYG0AEmrRFwKyXSlimXLoNQHl7oaAseXPb2Ed9rKSEO7vgZNfYhi17fuXZfLd0gSZPLNDbUSatAGnjoH5+lXnodoGihuie1q6uH8NIdjIQ99tHNUHTNdhImLI+bdSC+qVUYLDUYR00Q3FFHRoGa0R2C6wfZ1uAos+q2nCLpHFb7gYIvQQjCR/iPMrpFznA8bvgVoaQc6MDQigA7c+Nd2hBI1dUumUoqY6wye0MLqpFx37qqV/kZhms6Cbm9Pvq+8WXlPH4AHBtK+Ai6B3FDCT5Uv95mOUMsma8kC5ojGgFD2K6SOWBubegCEMEwhjpq2ro8lzKCkc1siRd6FrEbdztBcfzE62Zhs1tilvzFMhw0iX13wLHn3Dx3vHEG1CCPfGsWx8AOjc5uZqdADiZVOshIhlVMIo2ReUyJ2lCI8A6gD4QE4kwfVsJtcHq4SZxm8Zp6TLsMTeBorclUQIUijWhPRmaV9vokpqIp67FVgtP8zQey37k32BrwlezLUO3jakoyuLI/UIsSnGlYJFYnLtRRXrXlGZqw0fvzGXbwJKDXrk+ixquq8uql6orgh0pD4XlSesusHjhKYCsW32XlnaHULI3li1KbNZtIi/5hawZpsKrJa5naYpEMif1VKhnBjfFSjEdFtZbMI2PEXXpga276ztVDpgyvpfsqPQ3Lg/V+Ibyj1d0Oie9cBpRrTmTY7ThH4lU/kmYoI0GD5tDBesY0lbbKJyOwpZpqyzEvFF4rUyNTX1q7DnK5QZfnx7+sy9tF+D5SAj7J8U/8eNPdkFnHgQMYOui/uDAo8ayNTSBRDTjPHosCb5rg+rbGrDn+kdAp8QA1vDTx9Y45zXdZzkoYEbZKBKyzyxFW/IqCcqyjZ6mgpg4eobVDouSs5uk9D3WMISqz9/O1+HZTcf4/vDfU0L0By4sBa4Oj7tNBzpNht+03l/BGa+6n59KXhb3ycNONaOEJKEnHfeBCK6R8I1dRQiFs4sc0qxlZbzba7ymdCZshP2pdveBvhdFmm44EDzA60cXRVFYIaW/G/7NN1kJMN0hdtH2g1pNUiO3rofGfktgRIztG7KedkPIDsQcCcTi9dgSx7OTIXWEG9cDJo4ZI0spd436CmrItEkBnPJsjrNxefCmcJtbqai2v9QbyAOpM7trnXbBvGFIFbzWMxD5qtFoc2gGJqTb3AY0mKiNKWsMcg+5I1A4UVWNJEA7LzHrpoUipMTMM1RdYEPkMjFqhRh1VWR7AyR7Y0geO5eSKLbbHW0spgqMP3bZU7+N0GxL8TvZlBhMBohN0Tm2GqN+eVnhqVS0S5FnNmT3H1nP1ZDOt0hhgC2xb1ATsqjlOYm6bTy/maQFCnCwBJIxC3wevSI/Ca5SOT3giTF+L3ledKc0yAdDBJtCbRIwkPpd8jhurZRa45j8USRtApyUyk9hM3mkyI1pjMSAeq0oVuwJypIyyHxRBUGFgexadWJC1K6Jaboc2iqbe0BApVuKIFFeusiUB1uCSWzW6phRVeAfRHIqK8oxTkVJTQC9Lx9UcGKBGHlYrj+oQDr87h8qbl9PzEeHqnUIte1Y6H7q3K0y0ncxajIQtuUPk9xEoyz4y69fpHMAnWgizADsx2kHrcbwZ2lGz3HDBypJuMNjMb49VdEHsFueLpwpau6/CS+UrrRy+7B1PM1INCFLdoq8XXMeNlCQ5XN6bDm+QZYpmC15EoY4xcwNkQ2ktQchh0OKOqdGGCd2/LliKL9OXMrkVq8/WJ3OYnmfhFTNsxOW0wZJZgml81Y24UQKlNaVhPJ8K1JQ2yDkl5IBCVJyFexoZK8DJqunU3OgISNlfB0mMZEzBJLgJSabSxUQqCPKvLsCq6Zo149SL0viGwJ1jgSqbbXRQPek2EF7FZoqz2yHmRBmxhANbMxMl6tp/b2DdFOSF6xNK9kXRqcFTnk1i2kZLBs7/SOQ0joWhOjsrfhcbiQaWhCu1GyARWJzHafBt8uyUlhUSYSsCNPtJ1T0T6Pri2mUmrdRRN40/qT0wxhmO3Rp99qQWblVMnljbFXvDbXHj5BrdHSJZRfshjEs1z4g2q80rsC6NLRaEmm657zmdSIhaOTgLljzXMeAQ8JsOzBleBMxZ7s0xJJ49iR2IqQ+f8eya8BvMYj3Fyg9U22TWc19Ll+1Ce+i+FtMvgj8QuWMhPOFzWBvQok03mWTFXpxRGbEaQRtrJfQsKh+BP32OfDXg7vUJVr2hkQsKxE6iA8OxOwrITtdAYGim2MuiJGv12m0Bd0lfW0pVP3AwolKxpRU0GgYDyKFy3pu8lz8qXatek3uRJmhGEv9V8WehbcSpvPXHC6Hoo56YGE5NQPKJpnIbipVEfNQVZRzVW1HMC1YaQjHNpczkokYGfUcVLwBSpzm+tkL+gpuRIPK9d/+MzxTeQuxd6vq7rfLSS7cKKCKB3mufYnT1+2gG7vkPCmcPLkpISQBu6nrf9Uk2r+h/IwgJYUTpTSCsgmeAjiPuv1ma6p0aXkvpYojTNEAFjsgQnl8iBdMOtCGllps6Sma08pqD6j1LJ2K5Tdm2SInd0X7hUNTDam/CdJroySrI5ylVimhkK1NqOWb0yWcZudfcjtSO4Nin9zptzzEgAWwAkBZ4hgUTrMIM9CRlODBre+/2hdOaF8a6qi6yOC3qOdYkTEMg+JSxpMIwmy8w/od+BfoOhs0Ya7FtDFPVvePkl+O0RkhHwrr4sptMfUSjKyxAa5OZp8SAMpiCccd3qBhmHRRGAXcyK5V2rJ1j32cjHljk3etDtpJ+BPA3ADxk+n/P3f+imf1eAL8E4CcA/NcA/oy7v3tveal68ygjXXBf2+8QUMIRuaKKhBAX53utaKoqUxs7shyZS5XBxbj00ovDk6iJazwLtXM7OO2gfG/bAMANAZPhRxoldqKUWaynL4IXZkbruWlvBUFwdAN2wIzJdCS0mj7tFsPzpNmEC6dBtfsTe1wUlXjGMchyJGpOJV5DiJoOB6UJtI7JVIl1kMwt5PJH2StQ40VBwGeT0ZNVENdHrOU3sSfcrKdEbBuZkow4ZFSLvFx9j8vYsWEjTp5Bq/EttlJXqU550kBCtk741VEfdwhfkP6V6/WQm7i+AvDH3P0fBPCHAPwJM/ujAP5FAP+yu/8+AL8G4M99QFlsezVwkfILgSMZArBCfoVsGVFWXVxqiE+4xFCDz0MFepVUTxu6irWltV1VfRrZTQi20YUmvdyD1VwkCiV5DmPNx7UTtuCubtXY2xhh+dZ2lFYu9dEvu5XEOx5IyYyjCLzU33Q12QAGPwaMQTcoP6hDNHubpUM78l6lq6rYK/frdvVUQhHclifVmwJAj0l/ALrOWIpLWV2g1Lvc12DfXoKTvepR8EJCsk3hBIGBjHPDkgdY6L8VCq+R/IdsJe0AfjN/nvPjAP4YgD+dz38RwD8P4F9/X3ndwdhji3un6zYCgBA6gN5RVaWldMyEW6o6DwmY8CakChPJYh1a18okwkqraLBKAACFCAV02g4kFZE+8m+hdjIvYV1/GjlQbVNCpEosrFHaLd2v9m2pIVgawODAHCaGNoI1pWhxWZHyoxN6LguMk1pGBT5V7IDvLbVSf/esiHAcFUHDanL84gzqVoQMgEU8Rmi3shtrjHJAoSJXvAYyNDZJS+3QRuwVl2lbOxDJTqAqAzqMWQGvxi7KbiYpMe7SIRrVYiyT0buDx16bUfUP3NlGQa2wg8yncev160OPf9oQqvrvA/CvAfifAfy6u18zya8gTnZ9vRwoMqasyjlNqYRF7ELMjh4wP5YoAgAy2GuKsn8UChf/MCEMZSrCjUsHzrIylS5Lb8bj/e0zLe7oLZRJpIUA2h9n9ULgNxX0IxVcdoCZqdLLdC09yL56YsHqDWWO10pyGhLdsjr6yifdR76s3eZ8FyY4jgiSobttUf8tYFZxQIR+6+1gLENpPw6sRrsm3gKu9ZhY4lEQO7/bu9EeH+l7MRshqLrJ6ZmgnAqJssEQJx2gL8HK4NnSP5hlC7WanpXmeJDlpWksmPLi9UHE7hF/+ofM7EsA/yGAP/Ah+aI9erDj1pyzQig5n2l43DRcqeKFjrneqKonUyK9hg6Y1MnvIUHdjD++0yhweEvKZmH0OCjjaO7r1U4afW7aONYnpgi81L82pfsiJLByzWXuxi2mK4cBLlykpBv9ZSZBNZ4eswp9k22WnWMMOULKF6hx5ZdlI2votGvl5aiEQkTe0C235crwWv9CaVzNTJohFaNzHTW03UcHsBggRBBoPWs0W1l3CMe6yHKrARW9GP1DMQudvtQ0bgtt54Iznt484e1mGHcPT43r+7LGu/uvm9lfB/APA/jSzE4p3X83gL/7Qh452PHiM8M8Q1WKdWHrloj6ze55EXlt2pDJSt3qlSmg0amjun0Zq7L4sx5vTlrBLpkvNEwvouw3QBtnrIx/4bfO8nz0tCARoCzrVTotsC544CVemH9FZIqhJOuFONCy70bVZd9ze6k5sWcxA03knrtG9Fl4nu2blZgbTNhEHbg4nOeoc3xCvZ9Dwm3nrE0qJyWzZZle2B4fnmvnCF/9JB4QZmRrkV4P61Q1uAzgiDiHylWMuUa+V7iZGEBDNWsEWMYaXQ+ZqRL7gXnVWJV2Q5wWe01OR5q3EF8p5eP899PlhNPphLd+xjcfT/jJueN0fpmk32ugM7OfTIkOM3sC8McB/DKAvw7gn8hkfxYfeLAjedkRCq2e3PpSO2eryG3ocPRwN/HXd9LNUmYNzmrooDW7N6bsekrhZaQHmbtJoUnIlsTfPZTpCVTSot1RAhI/fkZKz3omDMkO1Sto5aPCiQt1GtG8ntM9pUa2+qQ1Le4zf0ZuDX6bfFs+t85XSKzIXI00+RLVVRimMs6R7WAdqhb3lIWWoLQKVaBTu9j6c1CTQULTgbP1Xuuxzn37aYFVHEyoQYvtD8tVxLWC+7ZtOJ0GTucTHh4e8Pj09ENL9p8G8Is5bx8A/oq7/ydm9rcA/JKZ/QsA/hvESa/vvYKguP6cgQNUx3zRwOs6PDuGNpZvtH4f7wPcvX96D/ay9j3/tSqpbi+klKedPQnvIPJrCA1oez1f0qUjBiCoKu+86VYbmYMgogOwXsGvp46yZyEFSsQWglEdNKCIcp2rHpnwVm2zkpwrOx5hT4tVamAXuHEHEKfDBJMyQsm9iMvTFlDAXgScEXTRCjPY9D5PRK6mQbs7hii8aW2FkOXfmkpIO5SRH5FRjXW3ITVreqsGZhurvYmHhtbGDH02IQncgMv5jMv5gu204Wtf+wJPbx7xxb7j7fd+Gw/Pz7g5gVauD7HG/3cA/vCd538HwB95X369DMDJDZsDwyc230ELHxy5+IcGs0a6dL+n0lviIPEliVWk9zILV5+nS2CEWQ05iWFaowHngai25L5uMnWo+aw1onCDi+LCC8ySYVSQicE3SbG0LUpszq5Sx1Br9MxgGdUv8go8bpFWdVfoEISj/dSTVnZHxLRTQ6m1gK3ZlC8+x2DOaCMjygIOEzxtp5n6zDlusEwu/4z9LCa56cLUK8Z+OOqkGBoHgaoD1msSdMwqai9hOIsYfJk82prtwGnTa3HkAqlSGwcXZLS6iKcHVt3DXV8sRbZ6KMSuw22hwTy9ecTbp7e4PDzg29/+SXzjy2/g8d1X+Nb/9X/izW/9Jsb2I5qz/yivBluO2mGVlcE00TIghXwhLloNTmIu/uqa2pZ703Q4ZNA2YJlALFqf4lbFQhxLzHld2YVUijK/Q1ZBaSELNNbHKYJWBrC8hNof1Mu1qo+dHqVfQcrUOlq/gWgELd2SMbgDDHCJiXr2c0APiK8emx9gJ4yOfLM0lHhHlGlbxk3PsAKNvTvWc+c6ZC/oHCT7EhCGZKyuAGe6TEEGIi1dho1fByKnzWCMge204Xw54+FywePjI968eYOHYThfzti+t930UK+PSuyOkJ78lDqMdc69EjWSGHhb+5Ou8/EcCBM9TBmG2kvcvPY2q5VKmsYa1k6unvfcR1bbV4Oe1DxJuL6Gr5I6iDO9rVXP6Jvldd9xp05F7VxpX3AgwoQWI+kJR2sGOYt48hRWs4puJKGX3sL2E+bmSIPCsroLKb3JkGpvAEyAi1+E29GYhiyTmtESMjFa5ffZbWbnlCEz0/JbJK3CGoQDx4nEZQAk4s29TwxWyJdqbljj5Eu5M7S2Ks3L+2ErO+2Am4GxBUN5fHrCw9MjTtsJ3/zGl/j6F1/D+XzC27dPOJ02bNtYtIaXro8u2XcSO5rgV0XzKC6prsW7WciEDjUHZGTFtcRBYLFFPx38MTHlRBNqCxmt5kC5BrOAIYfKLZZ7ttPaj1xCDoCe42aHLjbHN3mYbivBWIk8B42EfnBngUwQXeYK1yZ0t9SE3DFtok4rt46qG2ygiDd64boP1isQaamuAx+7/7GCjlwoPQLEAIYHc3RcW4uKsAt7x4x4/Gy7AKiY0BqJ2JfJX4K6DwENCarTJstEsWz64OEZZIhBnEbO7krsPIOtq1XbVFZZ95Y3Ywxs24YxNnz9a1/iy29+E5fLGd/68kt8+fWvYQzDdt4wtoHteZMz6O+ol3l9AjW+ZWnT5w2Zv5AzUjWHve1Yh7ys980jyFoA3LQjiZhSgYgLmrxq+Oupts1YFmm2LG8roTuR/Mjuj1DQQq37273yLhBJSESkRZU5VmG3v3XV1fH1oR2toOYjAXMfPdWdWNxTix3D7k99CNsb+Bh0kQufubb7ZVxfmeJaLHSaV5phDaYJ+K1WNb5Q8i18NaVULuy9v5Pit7HhtJ2wbRvO5wseHh5wOYf6fr6cMTLuW70QuIHNen38raSTvNrl4+tKLapbbo3HiyQ8ICTzl4EEZWFXfVy1B2UtwVnT4iN2g16Mw/gW5pKDixMDJiqotNS20kqoaSbRG5B7w2WfuU+czL8VAYRmhFAc6xEqRzU/fMTTeqfd+1Zjq74ZkNOPQ/03FBKEWNtFTUq9LEVOp6yQXk/Ye8C6FrXl6j/fKd0pCEb+bLWtmHNqW2TcLWvTl3+Hga6/GgO482yLVABjlEU78GCVykv5qeXReNcMw3vwKGBMQIgeS4Xvtm04n87YtoG3b9/i61//Gk7nM7729a/h61//OrbThsfHhyXUlq7g6MKCMTfXR99KOneZQJx7FS4ZbiJQS/tXgXxL8Pltoy29ZB4B1EQC0ZF2F43CI6AniG0rzuiiYZBJDg8XlrlVuRRnLUi8w28tyuOxS5WIO00uiNBuvuZhZDiL/KueJx+EbEhdzMxg8ETAqVZ8sUYq7aqN3qBRX1ZE1gBBEcWiW+1N3DpIvbDHMTFqajPmFrCHg2b1YASyd6B52mMcnpxhzhk74gCtMjtZ7VH973asx3/RcSqDawiYVeBAfIKGdEoYvQmGOIqRMUrQ3Nr2UrxegrE49kB5WI4Ev502PDw+4LRt+PKbX+Knvv1tXC5nPD494enNU8YVeK4pAHrvecGpV65PYo23+tvAMXnTbJTSYdXOWrrdqm+miQTAZRlmZQsHSSklBdnh7X0f6gs641K+ci12VvOtZfiS9I6ESmKjqvuK1rqU/hIeKLyapUhlS857pcgzayZFTcpqqS5EAsbAUh02a0OhpR3BLF2U6xBFHbQRJOdrK7gt4G48eg8VAItwL94mSNWnv2Ih4ANvOUDl4BHR8WRzLQOPYDidzrhcLjifTrhcLrhczjifzzifNmxjFC9SgbGgyI8XsVtt2zQwsPmI7XDd6lyEjlUbcAtP6DRPHzj3iWM46hEd0z9Ma7M3MdwShd2+0KlAhdCu7/Ss19qhFsm2ciAYJGLEBhrcDKlW52YYEs/tyoy6QSlxE3ZltWZb0jdQ+NR+ce4N0KzxyFhRC0MKHtWfldVJKIz4j7NG89geKbGeaq977wPQynYYQz0TaDDUTK7uo7UFA/ed85S+VkzQRkwHfOxx1l4SPDWtdVpHaqP+68loQpvgdmTDgM1mRcaOgeoXG0rpXXybQ0WEBBaiHLL8vA6YhNX26efLGW+ewqr+xRdf4Ce+9SUu5zPevH2Dt1885Zx9w7bN1ARqAEoQ1jLi91D7R5fssY5c11JHIwlAIlavBI+dViY08o6ErjbXQKiKrkvJ4XfXpq1irCRkEX7OreteCb1kFvr88OMWQ0n+qnwshNpllGpeUtqX2vvISF8su8FIsqxqt877V4IlvlrjisDvOFPv/kjQcFnr4022zybM9q6Ve7t7S2bPJalh+VYVNOfZAxjbHVcqu+dUe+mJsGaWkzDzhBXLj61tTNoMG1UvbKKmE5ApJBwDs6zkPIih9j60FiCU8vHDm+FbE3yt14e1zYVGNTOcz2e8/eItHh4e8M0vv4Gf+qm/Dw8PDzhthtNpyH4FSR2Zl+CZ5rWV9Puuj0rsJn/hB7RMoVaulCV+Eof7VnePFTTelykQjIlbSnDh+CkZm+BbkplmkDIcMUcDNUh0WXWjXQGKOMt64E3oN+1jiSVwlRhdsaxzUrpKMty5d8l5p8abK4XhwhAWud84LuRm3WRhKQuL7jjhtWoyJutuc56rkWlebaIto1+0pgUZX0t4koE2+yoBotJa2n08y40ER0mrY2WEmWB8EKyV2r6dThhjw+PjBY+PD2Ftv5xxOg1sm6WPHcU4FmaS5VGjbDD/mEl2S7W2wChGkJkhlUQADvakIQvxvGxdS8FIJPGD/x7J8wVFXRZXoc1NtfWx5LVstCEZEQ19JsE23H4GlLYsgGs7URrE9Indr7EtM7yYW/hVQ3pwUQmNZur3bmglybiGo/KV3GeTagWXC6EaJXS3mf0t8rAejA7E9IJJqL8LtKKdPnL1nBUMHCONo2GQi9BjryAnrozzIuoOPYXRRuHY9443sBHTQh7pjNBru/1uwjAg0xZG8gObO2x6R1QcHN/eG35FitKuqsuLUT8W6rCs3pnofIpAmYeHB3z9G1/H48Mjnp4e8eWX38DlcsHDwxmXhw3bFjC1IW5U1oOtmE+sVNQFS3j1+gQGuib04qpJfOUY8wxYsJZ5lAeHmTSkqAqQ4NnqwToibe2KI+UBaOkLtI+4H+d95irrcDIUwSKdq4ZhKn9Mr/IBANOxz4k593wdyHw6BeGNpL062sp0ssNw1CTSZCDmYmVOizU3VXCVntkQSsQ+RKMeHnotYlckHkk3ui7Ebi3VGWAT4ztAP36cA5+eEg+in/A4F6CUub4fGxmiAYy3H7Os9DV+Zv8vc2/ya8my7gf9vojM1eymqk5zz73nvYPlN7AACcmA7gyEDJYYgAUzC4kJzdiyxADjP4A5HtEICTEAAUKyxMgCGXmKwBIDRDNBRvjJ5mF43Hfvqdp7rYz4GHxtRK5dVefa1Car1l5r5cqMjObr42tCUkBwViIkz9WMNMYXAeKO0rteS2F/CLaKiEXIDNS+y1zNUXmIu5SbFyxLwel0wLt3b/Bwf4/T+YR3b9/icFilHdthInkZbNPATEId8N+smx9B+Fcr7Jhfbnjxg/KvSIIfAM7MEg6FIyROx+Bek4iNPHsnPyOfYvSuXLl19NbAndEANL2qLBW0lAkQIG7gDEnw2LsiesPzdkVTwG1KPJZlxeHIKFSwrBUgsb5WzbM+D83tE2yAnscavZefApHNhdZy1FtyTLmWfT1sDkyEzPMWCpJeRelhw7TTgIwhdks7ocgZR5S5RhGE687mAeTPTt31szkEBUBNcEDRvYzstp3q8jKSMW8ihnpu1yzFKOI9DGa1Fkf08/mIw3HF+XTC6XTE4XjAui4pRNcgdZrGjMkTvGbW+anjVWq9GQcWl1kKt01N18OAytnd180ilKQ8MXsCihLQNJAIA8dsgkIWgSOKwyfQjWQuirO4ZTZBzO35iuf3z+i949oanjcpmbwcVyynFVQK1sOCZanSdFME3xouH57RLg3X1vDh+RmtdTTuuHYxWR1PR9zdn1GXirvzEXf3hFoKlgVYqjnH2LjhHND02IF72/dkN+zKJWQHQGajW3w6kNSAkYUHHijCc8ysnhgdnTInNP/6XpS7i6RDvktSwSxrSNzkc4k01abCMTN6IVDpLrzYPnxUseAISYQZCmlwKhMOGxKBZW+V8WuKNLOI+S9K1sKqCdOdKf2L7+I4ZEk4jydB6OPxgG+/+xqPjw84rAseHu5xOKyopWBdqtYHzEQ0E8ZcjjmvTxCVT3F14BWj3np6ce7oIHcljmJiqRELGrOr7BE9ywaZD82UUB4u3JMdUbzCKLNz8+vliucPz+hbw9O24cN1AwNY7w449I4kN5DtAAAgAElEQVRSS3BRBrA1Qfbrhqf3T7g+X3G5Nrx/esZ1a2jMuHSRD87XBibCsiwoteJw7OAqujxX7Voi6vKIkeN6/3UeSU+z6klORI3LW1IMKNdVAuIliQDx8CNDGiUYbmiEipo6aRqiPPCmxH3HbSzRtZkKOnexpBvhch3f7BLWZSUYJrKb6MGkqfhJrxVi4xKP03XzVBzlPKFRNl+xV2abHUHA4lroe7EtVf8VXnC0FFnPw3HF6XzE2zdv8O6rdyLKHw9Yqnnqsa/VAJuUn5dfSE/LN9wQUdPxxaPe3IvNETcMat0oORAwoYdvQQ2chvaqJoZpGH5zryaMG3LI/dK/VmG094Zt28CtY9ua6NvM4tHVxIvterlKGaJC2LYrnp+rAHYXw09vjOu1QyR59d4DZHtI6891ZmxNMrJeLhc8PVfUKlhuOiBVIYQj3xlHPnAlhw8DToy/GQHk9Dk3rEjv6nCiJUHTYrdjflnkHEfL/uwo4BTEUQx3caltxdklhYSRFyLflyckI68H32Q7R0CR9WaeBkrXBZ1T+BqmMxEJSg1kzl5IHGKO4td+/3DG+azW9uNBxfop5txVKYp5yVbF1Mvh+ZT6m9b5peOLc/ZOGvlWGK0wGgGbcWt0rXE2go3ZQgHbc1fjUZBe1b9G/ck3VnTxohpYeLibdd1ZAAHcGb1v6J2xXa94/vCEtjW0a8N1u4I747Jt2LarXPP8hMbC5VHJ/GfEKMiEUioOyxG1rNg6oWPR/d6mBSsZG3d8eH5CKYSODVt7Rq0VD3d36Oczaik4H0+o6+rzAyBxS0NonalkQLZ870IsyTlGpqg+A0QutpLGUOfQdFlEvdrvtdbSqnF6N5EbcG8xpDukBHNzYitGSzFWQQ103E0FMDFeDaYbu/tDKUUNd2Re2XBK4rCkEED2dHbLudNKn5ZEGHx+bQ5jvJ4mS3dUHt/c4+3bB6zrgrfv3uDh4c792tfDCtt+yyRS2oHX88tWeAoKlOAZo4ev6f14+XiFQBi4T2+I5BycXa9L0qYjvE8OwUXOyVoHQ3mnwUE0VSyktPS2/SUN2PUBcB2tN1y3DW3b0LaO1sUS3FWf771ju17wvF1E5CdWYgQUqmJwW1bQeQXW4OzuMUeyBSScfUNhwvUqumUtFYd1wXFdgVrFyOdi7I3JHayDBrij8w4oto4yN8/qDmd2oTqsG/04GJEZ+3ylJq+93L6L4mQ6LsMdc9K8yy5Ch4WyFrDq9aq6sRC0wlJ2qhRVnZxWU/QxHzxKRN5L03unH+1ryCMcPyTiEJinSF8KDscDzndnHA4rHh/v8fh4DyoFpRavQhNPyI9N+v+uU+NHR3Kim5fdOl7HNz75Kgfiybfs6JLxOQCSPXaaknV2qOjqD8Jk0LIGkzhm3EKRnpWrcCOJ5uKKUlZwKeDa0CuJ+N26L4gFvXRIwEZTomJptoz61xIbjKWzZFptsvdeGCA1NnEDeAO4MNr1iuvlAq4VbVnBywLDZHOjNWKFNO78PbzzoPo2DZcNMKvTw8bxODvNZA5pnG4C0OzsY37x2rBZqYMy68IMBEjcp4UZiIjEDJEurGxVJ5RO4MLYwKiN3DWaidULljV/ho1+FOl9qhzznVvsd3tsPDrvLjxZzDkVHI9HnE9HLMuCN28f8fBwh2VdJBy1JoKS7Aa+VVeCiTmRBkCpTztVLNOMTBc+cryCU83tXpkXmqOqcQ8IArrYrZY5YoC67TUToLXag8Mz2GtWAzCuARNHI+eZaQOtibjIgBQ/4AJ0ceUU3XoD9Quod1ADUBuICLwRGmR7bevC/QuJl5RwsqKZQCsKs7qGMsrWwBbN1buHfJpgi1JwLQVPDCx1wbGuWOsiaYRrBZSrGWcbKL3vedsOg69AgC9zwLginBkqWZpARfc1K6mFotKFVfSZW6f0PKaojEMGoSNFBpWiS25ykdg7LKpMdy7BYKyNZeuyMzYqaFWlLbXct8Zom+7f2/YmmVwHl+98/0YRhlW6MINjNv56oVG32APLWnA4HFBrxVfv3uLrr7/Cui54fLjDw71kel0XdZYiRV4jnDrPtgNgxCBKrCfGF0AdfQAswjakAMrzuj8+G9k1u+x/B+D3mfnPENHv4bcp7IiX+0MqnztHB3xxeDwxcBAkrp69x2zbRQiGnWVXAVzlN6RvutUG845THkZVUyYzUKqKq0lhSv016SDlVvWxmWeVtd/1u7nOgsWxA72Dm4Rr9tbQtyaxAl2IA5cyGLKQP1K8Bk7GAeyOay766jkeG3NnkdT0sIbO8W8DmSPT1LeM536dd4j83bLWwjg7k8+VMHBxxiGoz31RYl3YvS7ZqCDP7VtfFOYmcLoxGOmpLr30r6ijTMXhdMD5/ozDuuB0PuF4PIJITDhZeiCfYPLzGVfN3rN/dqgRmaNnhPoUc/8pnP3PQ/LFv9HvVtjxPyGifwdS2PGTtd4CPGw7R5EAJj4qiJi0DXadfrAm75o0V8sMXAIpjoAKOW1r4iDDjOtlQ9sauOtOWQNqXXBYTyhlAViNbMaZuMH4HKNq2xKjbRKDjM5E9i5eYr2h96bdFSTvVLBQVStziwyvjdDVSNGbSBxEYtF/vm4a7qjVQdNCC2fgm4vuCRnINEMaGMH4bi64hMK2/0tOPGPayed+XAx5cyIWrOxFwDUeKohmuxW6dlBEZ9PaGMRFq/YQeu/gDhCEw1PrWqJKVCXGBk/fZYQeBGYrHK0EIHcx9UPKPoub6rJWrMcVpRTc3Z3x8PiAZV3w5vEB5/MBi+Zy9/JOBYM4bnULncsTssg0T/A41yYJ0M1LPnl8bq23HwD8swD+TQD/Gol88U/htynsqPut9pKJ1MohibuY3ihLkZwfkK5RgGWTOW/oZLa43Y1qHdfnCy6XK1pr+PD+CU9PF3AHWiP0RjgeTnjz5oDjQX3VtanGguwiSTSAF7UKVzALxxdfccB8ypkbmAt629C2qxjs6gqiAqYFnVZ0KmhcsLUYcweAwmiV0bRu+vNlA5UL6lJRlsUJUK7g6oVLUnSWV4Eh+2YRXNZPQZTB5VZdli1KMciwryQGsQgvB/Ig3R84H2KqWcaDlssn2ZLVz9oCs3q1KwEvtaF3BjdGgRBxcd+VPHVba1L1RiUn2c8BmFsKVlYZjBERZD43pKGvoiYejwc8vLnHsla8ffsG33z7NdZ1lXzuh1WLMFo+Oh6A1kT0URc3/UvGbc8M20ZQ5JlAD4Qenz4+l7P/WwD+dQCP+v0bfGZhR5pqvQ29BEb5c4CWELlu8I0J0dPlY6PpehXxVIfrraO1jut1w/VyFQesrYA7oZaG3hgW6DKElTpYWn9pP46pG2ZlzvH1xuy8yJ8lTDeVgpL6oQyyq1+9VE3l8ZEZOHzu48NogR8/j+7I41yTEtK0WReDs833YQFi7iNrLkC3+oV9u9lmk4mFrwDFXrvNKSBGTirkHL+oIxGxSkCeCcPWa0IRjmf5jkEiiKUUt8Osq7wkeOWIdRWvyUWzvJZMxMjmz5rOXD4QeDfxU8doPv85Frnp+CSyE9GfAfAHzPzXiehP/dQHzLXezKDkHqlAWLIpR4yH7dcDAyDjtThz05OB23oWG+KA0TbGtgm1v14Z103E48vWcdla6OiF0IixoeHKG0IkF1/4LD8YlbUtbdZ+dY5ord5YOHcX4sKFUHoHCqm0gemlkVos1uTOjCaYjtbEwUdUkQWtargkVY98qoZgw9ZNuK5muTECPiBqQR6d6etMoJ7JXEIWfeX5CJxJyJ0QOkvz+bO5PztHVeJlRM/ml5XTxqZCQTEXYC7gKgY6KlW25lpTAsHg3rTUtK5jA7LE4fvlKFpWqqIUwmFd8fBwxrosOJ2PeFTOfnd3xumwYqlV14EMhDx1FJEkwQCgQS5pGWzw+R22BHGRzz6Tq2ExuU5R0uv28Tmc/R8D8M8R0T8D4ATR2f8SPrOw43xkLm06WDdkB2Aace53knIGxArexhpOGe3ae1fpbbsyrhdF9gvjehUd/XJlPF3Fql5qARVGI8aVG6hLOR3T1xp3DyAxqmz9qRSipfmZc+voRQxrrTVstaECKF3y7TS2ZJWExpYnTzpOXR2QFOGpM1rbsF0BcMd2XVCrJimsBFLdv5hOnri52BMU4jJHS9Kk7xI6Z1Ng1RRBMlZtM1O8jlSKiX1XxQ66AYAjZ7cQzeL9jgWXFNcmLwwMwjg+EcztvhQhkK0z6lVUt7ptINXre5PsQtzF5m9BSMXUH0V0c3ddlwW1FNzdnfDtN1/jfHfC6XSQbbVFdlcOa01iv40vtislXDnTXvb5HKeFVLBLOwbOwfKOBw23GAn+HE5fPnUBM/9FZv6Bmf84gH8BwH/NzP8ifsvCjsAkuVMQsmyrHKnUPDEf6288I1vbxZBG/i4X0SAm5x7G/iwjvOxujWRUGWg4r/dO/bAnDEeei90zJlWARWzlLq90WXREP8S/8VtSRhzwjKAaZ87nc7MzH6HdbxamOQ3xBnwPtH16UPT51jMSA6CIGR+KTFZxZKlVos9qKe6zHlFmySoODUUtRYOQqorsoo8fDqKby7kQ292dOY3Rx0M8Ejef32mpdjM8j9QI543J273fPv5u9tn/An6Lwo7exwIJZdS9DMv64rxgEIPMIQXCNYOhR6NW0qObM4WIadwEwWirKJ1AvUZRRiIsZcFSFxEVlfJK1lsJYhW5tEwTDbDKIYwGkGTKBalxKBNfhliNewerNZ77BqYq7VMX8Y5ZM6ay7+OSBqp0TYIhCC72hHa94goAC6MtKwqJtFBKYiOUjGZKbAwZAUR+NEPMAWbYRdsRqcMTzweoDfMM6UZTEyEKxMoqA5TYz/1gWLRfTiySl55sLUFAVTGfZWwMYGkF61rRuWO7VmxX4fLbBSB1ny763FIkQGVdJdHjm7cPOJ2OOB0PePP2HofjinUpWFcjHIRaQm0KgyM7knv/ndvzQEhl4OYjQog01qE+Rr09GkwkO6LxCYb4U+uz/zUAf00//+TCjt4rgiC6uWImZB92znSCahHKDIYnjxhXniJ8TvfLmQPZwQC1gtLF6l+NZxOwlIpaJOBEJNxwbJH9HPIF870nRQZGB6gB1BzZiXqE3bJwDXF/7+Cm/t9at5zRQUXCZ6G+4O4tlkItpUgiCaJrbH27bti0S33rUviQACYR+dgIqvXDuDTnhAqhW4r1ntPYRKA0gkCAbsMFAdDGQ0pJRGaIrsvLlWIf7FmAxNebn4X96OnB9TlmFJiFGIaoEmFqJPCi78zgLluk18sFl2cxzhK67qCYJYBRa8HxdMDpeMTpfMLPfvYNHh7usawFp9OiW2qmdsh7dQRPhrZiS2fwkglbOBcFEidkTgg+RLu5KzIHwsOIr807Pnq8jrsspdRLKlbJwoSInfPAlyKVLl2cdrFcjWDMiuBice+K4Bb2yBzGvsxNYpvDZmlytLE9YoxzOc+p4z8NTwhYVuNBWI+Rod+FNMfGdPALr/DNZ7XSyzNElFWPLZMQGO5aXGHJPmVeB6B0tFXOhCTuMvy+bNjzaAOHYHJKQoVCdXJ2FHNtNgJCktwGwyIpx7eB0zBFbNf7/BdfZzfnshBAMMBLQW8VXAjcl4g1cGSvOBxWHI+SImpdK5a1Sk64LPpPXRyWbwCMERrG0Jc9ZgaM7SFt3E1JBDNfs2txPL4ssisRKwthPS04nhcQVXBZxBWFCWbtMWMTCJLBVNMWFQCkgL49y77qddvw9OMHXK9NCYFMaykLal1h4mwvCIKilLKUgloWdKgOzOKgwV0IBxFjKUXE5NJhO7XFgSlRZPHqCYzUD0yEtm0CVwvQFyFyDCC8iRhcRFyPDCmG2PJB+JFY1i+bGPg27kC9Yt0Yp7sT7pYVh+MBpEkLQaSqhPSyIlvsg6Ayurq1Ai5iJw5OPmYEbUCQKr/IkK9UUNFU4D3IyI1LPWLMz7lE0OHWk06WccsUiSB+7t5qZu8k8icE5FbRzwcwM7arBDf5OhGwLAvevn2D+zuJUru7O2FdxdJeqkkrId35HJKNJ7h84HRxbq6XO7wEs1G/B2c8NCF3TMqeTARRTgXfbh5f3je+AGUpWA4F63EBl4pOCzoVFJbtE1KAjDzp5ukkeeMLgN46LtsVnZtEpj1d8fx0gUCuTN5yINBJQgplWy/Ed1+wovHFrBqcS/AsXjQFrpt1ImyUNSjAFgn2mSEBLX4GQOvupUdUZZ/Y5WyC66bUYJS7K3J2pkB2jgq26FALfgc9X3HdOurhgLKsWI9HlEooSwnHGO3SQgULhR+6exkqshsa3TLu+pizVIpcSSa4LRUpSsi6dTjGJfBg1Com6Q0YkQyRrL7xzZ4Rsx+poROwmzRAUCkSg0QCFjdkbrmEFrAuK7766h3uH+71XtO7jQjx8JhoNHPgeG6cyJzYPhVXr9yrkQLZR7Qe7wx4Yx/nSCBuH18e2SHbHEWtprbQgC66ApHWrdO1SQKLAU0XhN+uHW3jyFSSJiqSNWZOQiPAJpWQNdFEL+p0UzqKFyZDnvP5Vv9RiIkusKoZUsBA9qg6W4is0RUB1qaGxY7Mt2S8rpdxtNkpEPXaiofIti4JNqAW5aL53IxAVQSSsdVQQvIN93m6wYmHCWA/s4ujplDPhDAl9Ss9wyIS3etsoiQM9u2n3mOLTbHJeh4Ip4hiwTwZ2W3lPRKvFPQanpmAcPaaw1AJGJNfJMQ1XE4NULoqXzPOovNw7/Ns/HW1CLsGdmfmez92fOG88cKtLe/WQRMrSh46OAcy4C5srqMKNBz7qNdLx/sfL7g8XXC9bOhXAlrVrRfZ+6y0YBErlHhVAWBmXMG+bUWNgU0QvV06rlsD9ytKvWDrIoKfiFGV+uQMO8wSsdbTfnlPeqr4u4vHW+kFBYzeCurWsZDkoNuUc7XesTVpRTYqVP/uXV7MQC0SegugaTorIsLz9ixUfSU8Pj2Ba8GJDjiWI+pSUQFUYvUHIFRfkAS+ZEY5DPYJHgB55jHTb4NFOoxNI9mwM0EoHNm9weBYnDvCgSyZs1t7uXWv20cA1bCLOHJ0OOOweSileJZXu1eYRkJuCjNgESudE8LB09LnbaSifnUmkE5bTL0y5qTqybgvHBx8iHrbTfLu+OI6u1kwl1KwVEH0hSSRZGWg2vyrJ5mlGerC4pWji0fc5XnD84erhDM2QIJWCioM2UVEFyAUY0yHEJSuzi8idgPQsMi2iVZ+vW7oAOq6CPdIsrup5aFRxmYAQ8R9hm7McQOBUbv6YmsILLWOjRlX89nnSHO1uB81e2oryXEhW2/MwNbNA5Rlx4GAw/MRT9crlu2KtS8gjbeuJG0SxPnHnZaz+K3cMBvl9Je0fknCunEUGgHer77BpQKoDdlrIjAm3kGlBOPIptcb9pmoEH0MmSBERHIVYVS57G8gWSI6hPEdQQQdUf17vGeyODyBdEAURC5+zdw8n9MRhUAR487T+ZnM/VXqswOA5cUO54bkShiKSAAEyOONfSvHFKgCUNFtFDJDTYhIBizGa4jCaYY1dpxJX5JNDlvfwBujtRWtN3Ru4h2X9k/JRchxPezb6LvNIqlwF2+urjnQ7V0HRQzVX0yM1y1AJoC7R3/11iT+ngI5Ls/P+PD+vSB1KWitS8GB0chwE1CkxzegZkbwWWQHfN6Ns2f+mxZxeGggRyI6+c25Kg3XxBB4vlhVHiVZE9vMHHXfu6SOOBen1D+Mn2FGuWxnmKYtj2F6YsgCCdGh9p5pNyYaNFUl4gmyM+Tn4PsrpJIWiDavp0qMSh2my8/RkgTRM0ESKnJl8WnvHWBagEXFuKWrkUvLFQNgLAAWRXJNNIEmYjw2MG9ouKLRBQ0dnTa0sokDxqWBqIAW4GG7l4QTaKDKIhIXBhVJOEFFHGHEtVTIVmdg64ytdVABFhLOTq2htQ1AkS0zjVtnNKjhQR1sshivPvpNxBDuHdvTEy6a3baxqvNtQyXgeDph++5bvHl4FPfPlcAlapp3B94xko0y5DiS6SknyoTYn7QbTQzORtXxkriO9ohLFudvJwPByYh2xu0RRBLCUZLobS3g56xWnqNfHpsRBUqERY1GhuykyCY2Jg6JYaBl05wSkNNfgs2WZBuZ7EQKA9NLjCRFMJpbcUyGXvcZ2P7F67OLl5S8inLJMohkAKbvZA4TetqytApXr7IoJdJFWJCEOZhIk0WCS6QlgIWL+4ta4u6aPpoJ23ZA6xsab7IopIhY2BfEd/F1xQQVCBZDT72jl65liiS2vRQNzVTDHPyuYMLiWWcZdRRQGEBv6NsF2/USIj0Lsv3m17/B5fmCh/t7tOuGfmT0nqCRBGDZ+2uHIMooiibAH7i2LWdINc7ZS7YD+LL72pDruEgc1z7n9Z+fmf8mGWruU0JggFzlkp+UGNn8Zg6eRfOBCNi5zP3D5mD9tn6NiK7Ebeq/If4sI+V24LMYVE6ePxPaPG/46PHlrfHkqCAczCjnCOYwpAEAy1Mu7KvHC0FtBVYptGgGSmdctgYqJNyYCFIHVstBW44zguSRA6vhjdGaGPO21rBdG1ptajgjQOlyIdkTzwBHgOdQo/RPus/qDRflnyxUlbTP5gPfmWV7TbecuAg9a10km21jbFfdllLRvl0arh8uQGO8//WP+MM//H/w/HzB/cMJpZxRa8FSCdCcaJMQlZAMAfxICB9s036QlXI7i60VjUQjc+tde+o5N7TNw5sB/sjR4NeOhIv2bzMjdIROorr+Mf0b07v328Xo5Oc+9InTiVGxs4p9pnKPxFbnw0Vb3nXASGvu8xyp+zGEfwVk7yjoKKwv0u02Q1bVtzlGp5FkXdM1NfC2ScLHNGuSqonRuojPzBamyhqmWHFcJaNML7K/z2BwLUA1pxuODDabGMwuz2LxL1yxLgWnVdSCWgqWItJCo8iCbiXLwITC6jsAgnvR9YbWLiJZuGnf7hHO3iV8Dr0AZVMpSB06OgTRL09inJRDsulct2f8uEkkHHWBguPpiJ/9/BuU+gscjgccDxXHWhMcxcamI7yL7BgBWlYDGb52JNpEZsOchGkDEmEyqxkmZmSw9sAqgtvPmUwJgoienkdihCvrzBS3KJG2Oci2l4hiM1dUROIJUJwfRpEtHoHotrOUkd1ZmY9toJYhFVl3M2UgDETC1SIFpY8dr2Ogy501Squj4NRrp/XuYmoyvLo5pqYYwZ27ic/cPX94rUWJiREVinefWOE0zFIcoqlO3baOVhsWImAxAEqcHbG88nn/V34UJBbu3nSLrsTv1gPW3C8a3mphupV1J0G5u+wceHS6GBZpQy8dH95/wK//6Nd4vjzj/vEO162h1I5llUCgqGVuAA/4bCasjKWaFfA9ko92ttCDA/FGONhzRTtP+esIKD7XKs25pGjPTP3z/owUi6bfTW10jk+BoLk/QfysPR4b3Hc67ktnCORJQaJTk949IbonArlxjUlinzq+fN549VXvSC8qspfuCAjkpAjm/mqMsBs9sMg55d4MLQwofFoMeZukfF4IaIvFdIsgXlCE+6Kq9FA8cYQsiCDnpkkjllrgNXhRxKPN6A5buKk8m6irXQKxqKaGqOsMk1Uw4azWagIY2RdvXcT3UmQHnkkLTJQK1sw/Xb0OTRpiZikz9eEDrtuGX/3q1/g7f+cPcToe8O7dI9blUbekMHlYCkLReGYPyzuoG/XirHFmBzPmPUxyQnq/L51wdSDNo4ceO2LaJVlsDk5tf7IxbeSO44nMgAbCNQ/Y7pkRdzc7w015YM7NjYBFMFKQMNuBghGaJOKPz8lkZX+8ArILkjR9bUVcULuSVfcLV6OSIwEDvYp/eyexKBvAFgZqFd2PO4PUsr1tV2zXKwAxBq6riK/cGYUrKjMqVyxdkL30IllZOjmh7b3jer0IwVgWCU0tmmyyF/ROmuJqc53cVRUL4IHuyTOgaRNMhpCCizAnHBm6AxtDfQBkR2LhCqYVXIBeVvR6dQooACztUAf4+YL2K0ZRX4bWGafTEb1/j/u7e/BaQAt5/1xs13XKiJN57YgEfrFf5ycmQeAGfUgCMAZkSWin6MHRQb9WOXpCYAyXdG8pDG5IIvrYn7hPVRiOhBb7TlNQsxvzkOdnCOUFAPSQBr3pIFxB0AQAWGHDkm8S4NGMecvy03z9VSrC0Isv+V2PNHF2PjlR+TVuTCLA9sDtKcwdvUmaPMnu2j3feRKo0j/4frzDAmtdt96SJ5P8GlqZdZDj2XaVE6zM3Q3Vg6s7SAxskcKtlpHmqShhjNjnoS8QjzxsG0oveH664MP7J8nS83xFa4xaOVxP88xTGqMvhgHjuJa0+xTraGcHyWAnJhj0jljjlxGLhGfr7P0LpEgLpetm5zM20nTtNAaKbyPB22vmL49lwv6PYh+nB+/Vo5iTW43YjtJnPiodr+AbH4dtWu2QeLo6AFn1W/P3LuIvDYbkCoe4qy5FA0iI0BzZJIcbE4mlW6m2ZxoxhCcDWXHUBTf0dkW7Aq1dwerjDgaoV3VlrSDNLivcnQBVBabMbjKe5GzjREB+GMR56YmVQlLqri/DeahmEPA6cVeN8Prw/gN6a/jxxw94/+N7XK8r7u9X1GVx0lUcJ24gCvawzfuz0/oGwWTjPv5hdyviZ/a23WU067SjceCFI8kZGs1nGXZjtyFz/iBNQmfdeXg3C8MYbTzzdO2oHANp20864qOU34cABQQnz6dddx9lh885XgXZTb90YxlCBZopKRs1MK5RSKqHsPmrKzXv4nTDBWhF8sz0BmwaQspdRG1xuDEFyNBR9uMt+IYJKLr/jr6hXZ91b/so1rHCqttX0a37AuIF4CY7BurvLqJg0Qg285IjtzO6KIoUR2+cnkUtkdpw9uqB8EVcaLPAsBNNGUAHtucr3v/6Pa5PV/zmV7/BH/3qNzgeV9TlHuvRaoknQvOi/hmfGKPUdYtlDnzLkYHH8wEU8yMARzgeBCTmUYUAACAASURBVA3B1eLndzhPCdFNBYAgMLnaQgCVYWzDOAm+NRsIRyPTVQeZeAT7/IUtosfCDKqKbTnqMG7E6nd/bqJvyqRs0mwdGISXqi3Z8ckcdP9fHjyv1GdJQbZwmRMjPqtI74kWUuMW/DIEFtBIZLLuaqBse+OWUcZ3AmyRkAV/46jxfR4L+98dz9e38Tf7e4uCT/jtZ0fuY6mzG7a2Ybtu2DZRa24e7o+c+5n6N4vQt167MX4O/5lF85cG+NEfd1cNYr+L0IYqMSaXMPO1u3nfE8J5hW+sKIZ5+Gi3rRO36N9ugn/S8YU5O7nMTqhuDa+KLOGUMU4YOVknrKUAVfPIaU5hYs1CSuzBH4DmY1Pc7I2xaRZZyTwi+9mO+GRZRsVI6O9McF1gA7ApYjcAjSVIpSsXV+Ojbx0SQTKcMaw4gbNbHZnnGFfuoJHlWr1GvqfteN9WnPGIOADPKb8RHPNDaozr0xXvf/Mebdvw+HgCdcvqqvLDxMHCaOzsTH4fDFc3gC+b4W9c4eDsHDtxaZqvovRMCnjQ7phUmGUMRwnnhHZR0a6NFMVEZtabzE5gVggx98VDXaqZRZSEoayqUbhN6S6McXEb3bA1MKJ4mnG9ZyZ0hk83dgym43MrwvwNAL+GQOzGzL8koq8B/KcA/jiAvwHgzzLzH360IeV4BM3ySVUypzB57IdVfxGu70nUfCiLZL+QVFSFQJ2xUff90UIFpVaACHSJrba2dRALsmMpwEpDWiIZJwLJFUmJi4zakV3npLFHy5UOdYUlud6TXZfQNZPYDq0mKWmqZWRdJRIT4WN7DhpaS04EImiGvElSaZGs/jkTLOsPsyJ8Ay4fLvjxj35Eu1yxff0G1EXVoN6jmijgjh4BWgZ28c0+73i23xT7374QWd9O+9uJnjsiZK4pozJ4SL4JLgob97Z+m4KYkUljlB2ryXqlzaW+6S9mQzCirKsKzjci9SNatBb0MgajONHojsLhwruXEfQcQS3402zm/A07V7r98VPE+H+Smf9hZv6lfv83APxVZv4TAP6qfv8JR6yqU249P1Mpml80EsCgevKj/0tswgm/rVKC0KwLjs/LVnm7OJSqCFr4yBBfOGYEyZfS0KiJkZMonT7Pj+Gbn2TwncV20VrTbUryy2ycNDV2ay1uHdLtJKrOY9t/yDM8EZbxKfOaYHfFrQ5RgpOMyJPaNT3vUwrH5ygk9jCDzfnukGZGVeFTylqiQ+P3zzj+bsT4fx7An9LP/yEk6+xf+OgdhNECnhETIkpmXu5DYYLVUZcEDNac+dcDFUU5u7lWMqhU1LrIvnepKbY9IapZy7KFHNB0Tnotq4irZZWpNxA02q2KoYwq3LXTuTiHrz0r+xLHIZmHnS44bAsmItPU4NO7qA0srsKexRaUJHY1+JFKFYXBRXchiHFtz3h6eg9QQ9uukLBZFW3ZLL0sn+kGclE+k84Dg2Q8bMI5QXdcQ5YUCLbTMhEfCssLDVaYj4D4jmJTKtudLkidHVqjifKl+0K+1M8UqoUjLyMs5tqMb5AmxmQ/MqC+FvZInReaPTtNqkBySU6BXWmP/qXjc5GdAfyXJLLdv6slnX7OzH9Lf//bAH7+WS3ZAsDE+eDjVnJQHhiLYWKrpTIucJ+buFKJh2V4gaoJpYhI78iuRMWQcXjZSLU9KhacyIgAHIlKIbcRQBC+KuFJkoMFtWQxMRlgHQGCxOiQOXEkhnrmkYj+SuAk17kCQQIiThZsKl10RiJwaejE2NoFz5f3KLWjtWvo+gYvhORYkgCXQuIIiXVk3zdQasYkXfv9jxntKIna5PdkMT4fWUwOdQy3XjJB6YmZ2wfznw10yYsl7nOl3ojRzJPlXI7cc0Ow3m4obGnMyO+7JZXz0HJ8Ht18P3Z8LrL/48z8+0T0HYD/ioj+56EbzEwvRN1TKuz4i1onkVTxTEeXwxH9KpulaSFtgBkQ3ZLu0gM0zlwzeabJmwzrSXUfpY3oyfzStjk9LwEmDyvGQUtMg2BDzGF4E+/KAq0Sjh67AqQEhXBDSkCaU87jy0CT1ml4lvGqWWSeeDqnG/LnF44ZfglxYtbuZ1XuVluGt6NNi18YIX6K/O19iFbG3uRdG2t6JiNORPXk2NrYR5fREv2Zu5sJaP59xqmXjs9Cdmb+fX3/AyL6y5DiEP8HEX3PzH+LiL4H8Acv3OuFHf/B45GZuu1ge85Y5hJAmSgbK3eRSBaZCVrEoMHEoNpAXZNgqGMJF2iVDsJWK5Z1kSgyKIczZOuE3gmtE3oror9yRVFX1rxDkNJCwrOwFqAuFSiE2jeUtgC9o1QtEsFiELRQXHGHVSrcWSSDXF9dV8+QvyihiLB5Bm9N/Ad6A7cm4jyrMEcZlyky5LKI7611FWa0IqmVQoLVcIOXe3bx3XTekU4myOYByjkTEwpUuYWylloqIzz8uf51Ig6COaG8JEY9ykfBNW0NbTuLXu5TNq3nWHT/lKzejsDmNZnGUExaQtQCBI9pu6wz1uubBGXomzE0cru1yTKkvhcvZrnR45MGOiK6J6JH+wzgnwbwPwD4LyA13oDPrvXG6tcu3RSEB2w7IvvNW5BMM5FHveVIUySXhVAkX6VW3IRnra0q0teqOdiWCqpF0kkXyexiwXO9k+YklwcZ+BeQIjzSPnsku0ABSq2oVVQFKgVUi6sLUoYpwnXFNMDD+4AcgCNVAFh6Z0ihyK2hb+K844CWXi6mcDzDs+aq7604ECX7BVFCeg0QSoUWM4DvOFYagqstnNSXGNoetry1ovq5ZZKJ59/i8cHhkqRlBCcBvI3fKuSO5EA7ntvI/WIkSCgwq0FUt7c9AVMtOb00eaq+i+lEbECVLfgqtZziG+z5eY6NYGQpwUQLdv3L6NnLCP85nP3nAP6yUqQFwH/MzH+FiP5bAP8ZEf2rAP43AH/2M9oaqLUZ18LiEDQuiybB4ZUbqBGu1gpeCL0BbRPc4s7qohqLw/5cDSZQ6kiI4n9EjN7THSS55qMbab+b4MhNRPJeimzFmT6RglNkLQSRYiltEiivoLx1+ziy1FAZaeBwGBAr+F7AvdkzCtZ1xfl8wul0wrIuQ2olOLDsUXM+y+kkax98gfNqcxK3d63eeIC16fhrazY9e9dY5vefehrH/KaujndTTLiuzyxq5/Y4nbfPe3WCZtPQiJpuIE13MXZ9m++LHTj+6LA/iexa0+1P3jj/fwH405+6fz6cELFyJs0/Z9ZquIgSQFKSCFa84D2j8IK+MPrGuPKm7wSmLuIuCjZUFEjK5q51xgXBRZJAWVEXydq6XBr6BrRGuPAmWWuZJccbpHu9AlwJRAvWskiRwN5Qt03E+N6A1gRGqArZoY5aFonEstVkQMUDL5NkHIVcJCOVrcndhMui+/5WJ4+h/vpABErqYzRTaoUWsFwWvH3zBr/4xc9xPh3x+HiPZQl3WWDsntsSDIhT7Om8IWg3u3HNED3DDMzsOlGwj4kARmgNeLIv+vgx3Z4sDjNlGCyS48OyJOM8dUL0gchm99d0Xj4bqVIbDsyZhlzasseMXq4GA6nXeQ0QLll5m33goi8cX943nil2CdT7zcF06HDyV3IYUzFTdd2yAChAow66SHUU3hiL2tAXiHceuKCx+LMzqai/VFUdFhA1cCMQV3SqIGJsJbbJ1BKmXJ3Ec8/SVHdGuSxS/QQIY112l9UsK1YUwfmwWxWFgMVmVVpSB4hAeFJWFz41nBArWe4dnkRPX0rB+XzC27dvpDrpSSvH0AQrExfNgXB2jocLbl8fwS/pc753uDVRVOW8nGEDPF2/h+zhDNtcRx8HM+YLyDFu90WrmWbkYJ2Rm6drFc0tR4MhvMPRLG3TNDKef4oJ5GHOyQnGXuIYjy/uLpvnOCzoMXGZggImzpGtf/ym4jgUeUsVkaA2EcuZ2bOdMnVUInNYHdxiJZimoKMPrrKlFHRF5kzBu+Z9txx0KBBRuIrPHRUV6aH90sAXHoDC4I+cODhAGGdPsyaefIFiAsBGBaSlrKv6HJH07Xg84PHxHsfjivv7M06nI47HVaqSJoFyxFvCDFO7Y8dSKQFcUlcoI85N1j2pIeN6j/0yzA0KlPs8eOn6FibFbZliDX1L1xmCD+x632dMP6dYvR3mcaaYrmOb9yB5m25ANLgnb9GBKOaFnAYEeXoZ479wRRilnEQeYVYoEjwwBccaAIPTItlZIlC1wRbgtEolVxDWa5P0zdyx1iocVVkOE6RQYyUABRtXVKqSg30pABcwF9S6KIPt2FTsagxcN8lEuy4FtQpAl0Ws/q0Qlr5IJZhO6Focuqthr3ezmuuyGKe2+kTGjRKZlvuVf3MdtgiHDD7DTLNpCCgVePP2Hj/88Auczyd8//23+OabRyyak09KTcMa1b/JwYXCCciBjyU4KC2qI9LgUEIZOGmAQzYRYJYaOCENsTc/EBVKNyQEc4DnIDQ2BvttjuGxuRxcamFuxqnfejEhi92JUKadBfOn8F+97xxIX8LOEq7JiSSmPgfrNqOgXNdt6/aTaC7Hq4W4+iICAGLrIot7csM4hLSEJgELclQhEnVpopN3llpnREApzo0BeFUahmzT2QIXNcoZZ+dSVLUT7i6W9CjwYFTYrO8MTpZ4oDQtR90tE08aA6VFdcRORj0yYEsqRMLoLNRasMw4b/qMAhxPKx7fPODu7oSHhzuczkcstWjJZlcE/LbYu4+Aj6HjoWD5QAZ9l2Kd98s4cbz04ebWEYXaMN/pXNsvHY2fxg0pARc5FsazSA2qA4PRQcyM3UT4aMFEEPb5McemkODg+IrhMTw1bnPGw0jGEcZceh5BGuf6peMVMtUEcBrhtKSPnIA+rkbsL+r1hgDyRrF2HcBCoFUBr5MUSNDoNFY9mGjRyDhGoRWVCY0K2rJ6eSWyWHkm9aSTh2xtA0gqfpYqZaZKLS7GFyuV3OVd0kpT1F1GEuncjzwW14Ajz8TwilsSrCgJJFLrOuF8PuLx8QHruuKrr9/h3btHnE9HHE8Hz4s3A1FGqayPjnAUZIZvX7A/d0uZnGnI8InHM4pggdw7hXf86CoApc2OhPQ75JhJk/UhIxynK/dDyvApKl5qeV5AsnOuyQPQrWMOomUE1PRdQthXANu2hvczk6pbxxdFdkbkj2PBBy3qaLFue08wm6xSYz/TtUnDcqlc6ItcGqFsuo/ZCrgx0BntsoFY9PTDctAFqwB1bNsFvV3A3AAilOtVUl0DWFiqrAIdl8sTtlZxPB6xrgtKKdiuC67XikZAbRW1Syrnyrrr2hibVXMBlOhIFBTpbkTm8BJRa777On4H4OTT72KlAmslLOpT8ObdI3743e9xvjvjFz//Fr/zu9/heDzgsBYtYcwOROMxQ/ELi5nFVjV8ThaZ2004sH6cE9m6p5Lrehi186u8zYzgxk2cURjCfERS9HdDVMLwPU+Z9y8/n0akk3ORAShpCGNlWXTVvETi9OzFtqNvxlmo1Kl96sTKKCmpFi8fX56z6yw4V4d9vhEqKXfAbZnDQnEAmCI9dYAU8QkALSQBKroaeTe6qIWcigTKMHfUYg4y3Y1/xik1NF681wAR44uI8KQBNqUE0lIxIIP51ni/x5cBBoedhuDq7I5a8/Q54atJGaUWHA4H3D/c4f7+Dg8Pd7i7O+NwWDR24PZMfxrRldu5BHajXwk7BulgRnHC+H131e0+kiO6PmGmJLuvWX+fr6Ld39zErb7MHDTPvSG/r6c/GYmbT/cBKbDF7jcOTnGt3iC2rgCUrs/RKgofPV4hb3wgt2IEzP8rz6IEe9hgU5IH4v2wwtIBWoHlWFAasG2b1oLr6NSkhBMIW2/YWkMpUk22lIpaGKWsqPWA3gm1XNCLGKKEk3ZJgNE2EHVczxu2rYMrAUyoZdEuSOIMWD54M3ZZCmpG8ssXX/dOkvhCfA5k7AMwOEfSgTJr5VMlLFXm8Xg64PHtPdbDip/97Gv87Nt3ON+d8XB/FmNiBiAYEXyBA9tzHXhhd4z6/QCUmcvF9XYO+XMiCvm5TiYoPNVmbnrrpuCkxkGiTx715hzebtsj09jPTEnhCOtuxKk/ZUb2RGjcUJnuyUTD+lzUe1FbTGpH6hsDIIunD3jYT8r+eBVkB5FUcCnFxZDuQLVHahFd9FdzNPaD/TyDUYlwoIreCy79AjyJJbyVDVdI6uVr23DZGmopqMuKUhcUJiz1hFYZ6BVLvWqC+oYrAYSG7drxfL2AQbg/X3B9buBFxNhaFwFQekZOXgEPty0oXDXLrSC6lWqmLhS6WsQa9wTcFEAKeGadQkaoCpbDgloLHt8+4Be/8x3u7s745tuv8MMP4jyzrgvWpSRgnDDHzg9cjZxUBYfS4JuJs3NGiHx/4nrxHKTPFIzT7A8U15ckAofGQbAswu6I5dwwGeA48Um7juDiMNJzssTovHgmYiXBYhkR18/rnAZxNIKZnndD1C82PncXDsqSrQbwWeeQZpGzLf3/DdkdTymdGsj9cITeNIpE+0Zt1iH+8yT73OR7fG4eE+92SysNQLYDa3o19c0ucCcVyB54azK5Ukm2o2s2mBz1hmG5p04nRLHQXUsuKT4FyQCW7pxbNLWhVMK6iI/+8bDi7nzC3d0Jd+cjzqcDjqcDavaDBzAqePvP4xnTgQM5rW8EQbgRxmga+sjVJooSI00DduQYCL9c4MQggcwAUjsgCUS/hdSfRvjUF1+IGNMsxscQR/3fkdebnpGehrMhU6VBDvP6KaF9f7zO1luOKXbRaiKVQzB+slkKtKbPicMAsr+8iFPN4XzAQ7uTwoxbw/PTBdwZtQJA8z1r0iCFgorCq7ryLiBIsgp4tVWTv4HWGp4vV7QuIjUVjZMri36GiPTUAOXsxFXagorwLEExgjRpDMl5gtK/QgVVbQSHpYIYOB4PePfVW5zOJzw+3uO777/F+XzC/cMZ67pIUFBR5yFrSdklDRxq5iEhjs5cP8/6HDpsLYzAH9dEO8HljJCaehIBOmbI0rwBth+dkZrSHDkX1bHewod0zS1GcutTRrSMkrdmREnSQBDlRiFoWe0JAE4zzkmMz9TLaGIJl2upOozPMs4Br8LZyRFMei0RT2LMoZEx6sRETFc2bAV1FWuwnqvqTcfA6e6IZVnQWsfWNjw/PaG1jtIB5i3FhQsiFl684KSnhuYqzF0LrnEXi/p23fD0dMGydBwOBxyWVYhDqaBStYsqUpgYj6KSAGvp5i4RaTY6Z1UZTBS8SGwLS5WoutNhwVIXPDzc44cffgdv3j7i7v6Er799J9trFVgWQWwjEo6Ezungc4iY0QAxinQRI8InUyeRW8z1p1FsxkQQMoeNWE1E9VcK/ddEY0BLZFt7If3EY0aumESJEOuJYhwTEcuEdX+k7d2ErInJJ7w19974LdzAzThaoi8MECJTEFgVVpcso3HyMDhzaIot6xuWrN3xOjp7OnyustXCfrHPL44ima7TAsg7gar4wAPAulSsa5XCEpvGuRKQ4TCQITRVys/yFQes7HLvxXWmLJiFtJLODQijrSpHCxKWkxkiASipjWHBUgsOhyMO64Lz+Yjz3Ql3Gsl2OK5YDyu8jrx3Izj1wFsGRM/zkThgmp8RF9g5/8zK7DnTotyU4h117X4Tk/35mWfCP8+IN4wr/c2cL49/QPS5X+lp+VxmMju6cEOUGJydhuvpxvtMOTDOxTC5005Enu8XjldwqgnAzkaZWGcbUEoIoPHRunR6Xu9QTmISgKMoAUsh8EJYCvDm8Q4VkmX26ccNzx82FBQVb4UuVk1D3asZYUyVsMh7iaUDA9t2FUlh2VBrwaGv8KguQyxdjwj8ic9Fx9W5Acrdwztdo+AIsu9aKkpd8earr/Du3Rus64J3bx9wfyfBLN988w535zPWteJ0XFHNGGfIrt6BPtmOyBPAI4jWzO3lWgPI4CNZA3O6GzfCl3kiEgYLmajFNRFOTGZ5dnoyulyl0Ui/J0+4WZKhCbYGopSIydAdpQZMiD3zfMwEnCwIBoN9IRDWIHmcH8Nhnx9KLskKpxYgH+nHrFMzId4fr8TZBfIT2jvQDEAoMjrUigWGRnbm1fcja7d6RremwITlzR3uTye01vCr//tH/Jo/yHkiELTiahHu2WrRCFIGOaLri5sa6q54fn5CawsOx4P309bFLcWG5B0SVNODAIClNDRTB6F6YUexJop3HgpJLr1lxZt37/Dz77/D6XzEd9+9w5s391hrwd35iHWpoLzuJAZKmduRO7NyjPBKiyw1OrNpWqONKF/kUJmQIucmiFXOyxSBsXZ399aygcw8DLM+Sqk9f/MTNP0dkX0gahp/QUNbFgcQbc/EwBA3fmNvw8ado3ujoTS2ZKOIp2ZqGffKTKmrmZU5I3hGDJeGCAjjy8ex/RXF+EnkMmT3kyq+BbnzJYyFoTRI8sn05bbbQWIgW2QLb6kVtRag3+iDzxunZoMoyWVKfMxPXks+ARO3Gj7HmHPxyOHaJK3YvTFMQl0WrIcV67ricFhxOIhIvyxFxuM6oQHmzFVMMkyIAENYvgFze+AJALcrsgozmKRCZ0335MiwPKPzPTfBlqbPPP+YiFomQmmcPukjtnvfg2AllUvBjDHdv+tkEmuIb10wnCLcuiRUlrCLZIeyfB7e50/p68BrIDtBPc8igKQWCRSxErk+4UQeux17r/ZZjFaxqmaAgjsmGM76lwL02nE+HbHddXAD2iYllwFGXVR66AyUDqYNoAYqLGmvilX4IDBvaNszwBvadkDbDgAI3K7iX9+blpsSwoAOLQddULkAXaLxSgnEH8NNWTm17L8vFbg7rXjzeIfj8SDbaocFtViFm1AhbLgDLCPmL4A6EMKMYvO98qt9Twjh1w18FDcz1mQAH4A7EZv8/PS+H0ecN1gJM5b1L/nRKxCQIuCwxXYrLBg2h2ETYJcGQi8zwuFtpM9hI4nrrF3vu605IU1PUm8AB15WnAEw7KAYvLPhxMAt98crIDv5oCXeHMN2SxlETpnkksZgudOQ2DDpZ59EayMJBlSFu/fScToesJ06WmM8PzX1kmPQIm72pQmyy7aZOrjoJBcSBxj0jt4uACradkHbLgARetvEv77rlh13qbLKBCsRVXoBNJlG6bFGwQlNfSgeAlwrcDquuL8743BccDouOKxVkm1qDr4B6fKUT59JgdgmiZBScCMD8BjTzlM7hvgz8vsokpSR7ws1nfzdED43MSCKwQ4EqQeEGfrFjkFuvVYgEISYRjFMlTGWPFmmfxu3TwictoddGrRmM2XSZ4ufvBHdmL/cDUP3IK7yO5dsnMw3GWefZ2J/vFrJZvchN+RNyO4XDpCRgMkRPHF5bzeoPoDkd5w1+oHcDkCsqrR85jhn7C3DC3c1lHSOWPVR4vfP4/kR+PN1DlwwwM3AZUQn5usWJac8b7vnhZTkCJs6cYtYZBKSw14zcaFhEPA5z4fLBZOIa+u1I0q5bZrvsLHOZ41wTNdOQ6IbP2Xikq/fLZsTyf0xq0IjASTsY5Ffvt/tFamzs7QDGhXMjx2voLPLvu9SKpZSgVrAtUqGGCiyE2Dpm2WRWSrCECUpQAxXvthOBLPelgCbVQoQmUiz16pITppumRmNOxqzuLT2IltrTV7ir6+lXzqw9YbSGNfLhutF6sj1BkgCDIbX4mKn1zomHqQPz67o0S/sPpSkBhmrOgN9l6Cb4lKP0wNrNFGIgAvyOcRwfboyc5T90ul1gaw0X5D/0v7emw4gpO7QE6bkSMCMZzmTzyD+D+conncT0fOzEkEZ5epkA6T0fouD25STz3MQMfJ3aZRielIbsR7yjLyDYqe9rDhib52IXiA94/G5hR3fAfj3AfxDOv5/BcD/gp9a2NEHlHKXa6SZIXul4B/d489ZAB4Q5xSdOCvnBADm2TZzBwdjIkCDTfJEyUs/u7860Flro1uqaUv3SwVFQ157Fw+mdu3Yrk1SWXXbEklppGFaeNJnraORRjUSYlCIfSY2kiF8SUDlHid7PTmPPeOdqwwIMmB/5Z2HvzQ0OOm8uyFMWIcRp/ynGS4Tco6qQziWkK9U6reNxUTlaYz5+YwswVD6bbrwZgcT7ZzHlW4fDZ6B6GXwOsptJ1+7PK+2vtqXzM0N2QFLj733QnjpKB/9NY6/BOCvMPM/AMk0+z/htyjsSMMrUb/08it98ijNdIBjIBBSnvL95o4dEWkWzNRcVmGuq6KKy9yzZY43x3p7hdjfcx9gDDxUk6yqWP/ZZ8LeOMn4qYM+FWM7OaTWAGKAz8Rhxkml4d8A8MP07slAXrv8fdApd4s4qkymds1rObR9g4s7CBs8JITKV/mzdgPK47lxPbIqmPrlc0i5q/tjFoIoYDvajhnJ64DUZ/+WHz/2aLhmxJ9o42PHJzk7Eb0F8E8A+JcAgJkvAC5E9NMLO6YOW9JH49KG4EGB1c1UHgoL68vI0lkNWY4gZmySCQhdWwsm9I7WNYilReGE1uG559sG9K2AsKJSRyNGwTVqNfYGZkLvWj4ZBM2NoYOrkrgSBVQ3lM7okKT2jKbjLBpFpQUnmGFuktCxiwGYQCjiVFMW8Z5bD1jWBWUpA2d3gMgsKPlcB1BMi5G+jMK5cdEwFuXLwwGI0zkMLWTEdEB24J+52S1QNXVn3GnIz7hFmoyGevdS0z4TRPtnDlRzmq/0Pp+ya4Ozl7TrYcQeykDICYLRpoh6g6soM5LblEXyCtJsybaTReg3PX7i+Bwx/vcA/J8A/gMi+pMA/jqAP4/PLOxIudbbUifaFjQuD2vH/QjuUTbWGQnOnO/MPlaWSS0qg7CL60IA5NU7y3szRltBWCA5acXJBbDEFfIkS/1toj/YIu00uyxppZOSRXqjwmXk6JY6mkcEyhKCqTylVAEgA7wsxhqFz/nVkYArzWsAVEgcnP6SX6ffbSsI1vQLGLVbvj2Ryf146aD0VFCukQAAIABJREFU7BBr8/RQfJ+kgqGjfoMhOvbIfsugOYj2L9gxBlITbbvu7r+GKpmROF7TDgPFOZ8CWAAnhS2Bo77tSCL2x+eI8QuAfxTAv83M/wiAHzGJ7JxrHU8HM/97zPxLZv7lV6WmCUHUP9DBWumcqvapyukzWTZacknNDV7+8AQUEwWOicqTyz6xWuzIQ1uLl0gKhKX03aOSHABNHI8Fy8/O3M2tBR7YwKCi5aoWCVoR/4OCpVasizjSmDNQKUl8BzSIxJ4W8xDzEd99nkydobAnmOzE1lEYKNuEy8tcUv0fxZoONoGZRQ2zc5vsD21OgJ+a0KElmwYmpyBD6PlZZJAwsZsdck7r53Ai7ZYce57m44Y+A0oNmc0lw+ggltsLGL8nVc7KdxUqKXx5hIlbx+dw9r8J4G8y83+j3/9zCLJ/VmHHWwcBqGBUiE94gxQotIkUapbEQxVH3bhN5CmHrabY6BWWgCRkeRfrKzOKcvgCLeJIBbVUcAF6WVDLglYYtTTUsqIXoJSOUqzWTPNILRHvWX2i2d1lw+UWvh/uUbKczYRArYTDsoCogCW3FqhWHI8H3N2dcT6fcDweNBGFqAroPeDKqX9Wi4rPRcaV4OIqUVD8riTPIdQAcSTlPDgrjVzaPgfYxW7TCI2zpX26LQiGivMoWZVQZ1NKn/OdCkeMJEbfQCRpam/acmQfCEtuw9TEUWqY7tbTAQM0tR3nyRmg3JmQ1+aY4Egu9+v61opWgDaEhe+PT3J2Zv7bAP53Ivr79dSfBvA/4rcq7BhHUEpOdmv7zlIYj4GiCDly5QCKQag3mEwzSmmCi7ZPwyv/LhS7JOo5GtviewJf7cdoXJs5QuZ4mFJrmV++5Y+r7mFIytkXrFUy4lbfjdB7p+flJw+If5PmBxEcuPsM6QhW9Ml/dmle6JsHDX9zX3NbuRvRrdi1EEIQUtIwz8Pzo+3g4rY+maOO/fA7U7tl6G+S8uZc0bnf3jeK6XRphMbxIa6d+2Zbz1GYczTWfuz43H32PwfgPyKiA4D/FcC/rKP8yYUdIWOEK9Gal03mJTl1cFjWI7FDbJfZMVLqG1xCs8B0NmNcH3K/w5aUSLb1KqSeu+5xeEh6BqRhYtVHvm2QtEJV2iBgWQsYVbfplKwRo3BBL6zjNhJWATJdXDh7KRXLsuKwHnA4HLAsC2pRUZ4UxAjinUcyUeSAMxKjnZaVOIAZ4XZh9EBekQlxeABQkF5rxIfg6aryXAU3D0C/JYJbm3PCCiOWu+7sxmjPTsQvaNZ4bX52Qjwr/gkSjmoOTyUaSSm5aMiEu1dpDD6zdcSuMe++hOipHes7QBLcNVEG784nWPfn1mf/7wH88sZPP72wozQoPuNNuGEBoxMhOC27IQ2A11YT/qOTDMBENRF/dFGQNDK220Rd2FrX14beNzXGSd3nQoy6CHZXbpKZtgGo6hNfRdqom4SPNu0vGOh9w7Y9g7nieFqwaF46poqyLGj2nE1cdLslxOBIRUW0gGjVmu0VhSTV1OF4wvl8h/PpDqfDEYd1FX94CslkZ6BRpA1TUHZdTVwlkc7M0cPwhgRJI4FzJNxJjUagaUBaf7OdAxq/U7Ik5+fEyYwQtxB1tFtYYUwj5AOy+yhtTiaGoR2MJJJI4cLkBDpvBWYR0iSNQFhOsJp2TvR7KRmOU3MD0ssJsRcgiCvMqIUQfV84Xi8HnXH3MBm5vSpdoJ9lUDOiD0fM9XxaH6VlmPyVOyJXktIUMuXKHFhcf6BhEez+aFOeZvhRClAh5aQ8JXUSyULTtIepSJiMgrVULItUYC0mwhMhtmhi1Pnz7JY5zqsh2rw448WWQomN29F0gRGIpEqEiXL34H1nBqjPDibp2mSJHgpbzEg/dCRI38QEX+gT6z3sa2Pjzchm8x3SiWN3AF9+XtLz4e1MQ991LiF6+m0U//ONGJ79sePVQlylSITsETbS74gxmBsrICctPlwQNy2mfbjFYZh0S0wQfWsdrXXdFy9qKCvgbmmxWA1pktWGuQJc0bYFW2VcLiyRcl1sCXXrCosN3DeAREpZilLdWkDooFqw1UUKVQDYtiaIxCRbf2BQLyhdVIkKS0G14Hw84vH+Hnd3JxzXVXLKUdKP03ueE3IEmo6RBiS0TOqMAbuthPuyx9Zc3gIdOD6Eq3Ghm8A3M8N4D0SPLbGkKlgdeUS/Ir5e/paQ6ZBzuGVkmzI9QZqmYd4Md3Pudo/kpwhiAUHSZVl73nYoPztOnp+j7x7EtCMwuh6ZGDpHZDgX0pRd/PdCjP97exjQkCI7sFGybdgbDVMTAoF6sBGnbbhkcc+HecgxA1tjbK1ha4zWCK0X3V+vYA6gqYXQa9OIsir54HlBa4xSGnoDqInuvxWt+cYFaFeAGJU61iqdKiA0FDQAbVlQAFxRsJFEtDE39djrsrXXq2TPqYLo67Li4e4OX719g9P5iPPxIMY7R/gA7bwbMc/aKL6aMTGQhIa/8U0kqO7tsiKXJ+qELIQBKFP8kAtjmNU6c0hrM74kkdr7O7LBkZlRECaYuawM24LO4Z1gOH0IbMbIGEucFnLhfuzJdOeIyDpOlQwGr1iKvHr6nkV0qwhjKahtbJF8JBEpsxkhIboBPuGzRHjglTk7SEVs2lvVwwhBfnKPzvkm7AbL6ROD1ZfYK74ljSJRfiUiJe111yqusrWIEYZSWKq3r5Vas7HJRD7hBiaed8iWiaWohgCUTQIDZm2tpWBZqm63LZJZFntDZP5G0wenmfYD558pvVM6O7b1kvPM/htjUOINWJP///DTsMSjOjUNa0T0QYWLAQYBSCOkeTwmPUT7irfDc/I8BvlD3JvG4A2kdbcL3AfCT6WowWmcA39L46Y0JzQjeiaWN+7Px+vkoFMOkH3bu1qpfRzCKpBHQDCDFgJQOA08JW4EggOJe6vkfN+alF5uKEpolGyquG+VaI7LgkMBrsRAO6AVoLeGdVlAAFoTy72sZRMLf9M0U735ShYUVU+qxu4DRA2lEhoTGA0MEe+b9v1QK9bTisPxiLuHO7x99yh12o4L9gkXEiLpMCgDwI351/8JKUbEGAhGnmznKuzSlHE4U4P85jI3ZAQyOHuZMJH8muE2vxfp3ozgzt2dq2O8Km3LzYg+PDcR0jDDWSvZeBBzY+MfqYQAnl+q58j7xsj5tmw3JZCahrm6ZUcAuY33owiej1eo9caO8Gz+7qyRaJRKPnFalTQiX8rMLGwt7AcY7sq9vYv/ydYY2wa0TtgQipKLSyrzVyIs6wGEqsj+jG0jtK3haVlAYGyb1GeXXYOG1hnUWaz8vQkXBwANQ7VwXBHhumy/EdCxoaOhgLCxOBphKVhPBxzPRzy8uce7r95iPSw4HleRDGzMFGNmm5sJ+XVg4xo4FyOYKByoE7sZfvgyhCzExUxxfOOZAax2LnOz4IwUYwEhttkSl0yLnD9j1+uM+KnrlJB9FuMNeRQGvBVOyB/oicxNR64uc3yTSOVlsHHY58S4fExZHcvzlNQbO5/LL9j6fEz6fZW0VL6QN37aneH9FbGgnM7ma7IAr+9srzEubjZG2SE6cXaRDWDK4rknstA/Lk1A49n9uTaWRJoN2SzjjvrVL4sksTwcD1gPK5a1YlmqJ5DMSJ55z83jlkfVCxfTjd8dsOw0jUK9A79/zgicmiP/eVzFfK8RiVt9G5A+Ef9RSs49i69Tm7NqMyPkraCcqdXdb7e/7E9PeTtuX5Mg3N+nNTHY/Zw4dju+LLLfoHiWlsqAJBtmbO+XnYSxW1wTqkbTQxuQZLAaVNYbY9vEGm/llghK7U3dtNA2AOj6nF6BtgCtg/oi53sFuGgiCxPQGIXE6Nc25QNd1I7WNeZ9E2ceiYsnMcQdD2B0ySt3d8KyVnz3O9/hd/++73E+H/Htd1/jfHdArQXLAlS1vDq80n67KM+gf0sTRU500rYakAoc6F+fTOO6CbMGJLXvE2eynyjAd9Q/RxT1LpO645ZEiBO05+i37P9eUhCRX1HCf94t+oRA+KGNuNFF+hswm3X2kscwDCJJPtasmcu96GR6lmsJwfmDEUX/bm1y+H7/DSKVj9cp/zS9Ii/XLDaZAU+4ZKC4tGSGH2KavJdyG/K3d8kZ31oH91h0LzTSzY5gyC86IPUKahXUFqBViBNykXBFKmkxGYQKdEK7StuskXStM7ZrR9u65K+D9LEsBcuyAmAc7k44P97hcFzxzS++xR/7vT+G8/mANw8nnM6rEkVGoT7awBJwjMfM3cg/zuwzc2g/48p/wmp9cHH5U4nxgDRlIjjjug7v+Tr3C5isCUNj4yhdwrLrp+faOS/EmBxzkOCE0sU+m0ocMD1emxmej0TAAtEzsitjchfFib1zPDsjeqgS0YfhezY+JML80vHqFWH8IJuCPBFp4UmMGnng+X1oaNewor0lq/B26eatsUxm8Js96ZHUtyAsRKTbeWJp5y57+b1zSrChj1LoKbrXslS1uq8LDqukiV7XFbWW0NVMaAu8i27PgH6Tyt9iCzocfYYTO7JzYXwayW0+D/99RrhZh/Zzw3WUujZdn0J1b4vXI/GI70ZUJyIwPWfio37PjQcNn4N+ztIldvL+x5q7uRyJ3uzbeEkPeBnJ7XgFzk7OOMw33bk42baYWS5pXBgSa/k4rBHqDUEHQgutuNoaemNJ46yiVOHckmIRExqzcOlO6oBTZY/f9HLT56FAyAChYNs6nj5ctAeyZde5o21isZcU0WLUqkVywVMhvLk/4au3jziejnj3eI/Hs5RyOiwEyZgHSKFJI0VZZL41zzEJggihI2fRPUPdUFzSrwWGHQDzw9cEG3kxyH7XLyMRyMiF4fOMNN6RTi6iUkYistUy+4ncHRJEqBtuoNN18mcN1rSgePIcIWQeO0Hjlf6ivXurP5SnDnOWNAXOsuoVc2zkx57NiImBEz+dgmntP47wXxbZXbpJCA8AxOJLDROkkRA9eKqc5aE5H/BEKmdjCvfIToPOyI5PFq2WuqYGNZY8dFzQ1e3VnHoABLLbioHQto5Lv+pZ8bvvrAE4klNaQjUV2Q+LBLbcHY94+3CP4+mIx7sT7k6rVGGtunfPDLHEGxDspZw9i7FuhU4/IPn03RySOdlQAF0bmy8TWwe0jXb23MiAd/xlIDS71vQ7i1rgSJitoZThxBC+OPKxEQRDJu8jpb6SN/cSoshWZRAJm1OfhYGoTYN36jnC7Ci9xlmi+GzPGa/RphnhXfoTjlcR4zPlD9DR3wLyppvCiDFS+HifJ36H8CZKJ9Y1NwkYEeruT79xQ+sNLVVks8qZwgHyM8XBRi1CiOi9rN865viLtGhGLcnyb0bEceZemEybhKwr0u4umk/MLdOIdgxKKsPYGwVR5zL5xrz1vD/YOxKwELrDqN/nMVDc71cEMbGrzWA6Exp/n6ot3gKp/JkyF8aNC3YzEw0OMEgxL0qD9T1LGrmJcMCZldsxFHl3683jFZxq5CggVCroJGkdZdGn2TNRTSGCdbLmEMdMvQlZtEqUl6HpoXWyqEwLkkNqO668oXPHc7/gQ3vG1jY89SsumpeuF0g2ChapQQpNEIib9k2BQ8tRc1EikDLERrZYscyvB6nAuizV65FncJexGwBHZt0skQ7I7K6uc8xzmhT/mnRwCvQYIarE1yknNOmyDER7p16yP1eyCcszs0c72RpPiJp6uyPudm9NVzNr2i4ORM1TM8wbkls7xdplzKQKoyTeF58jbycDlJGd1F8dv+Woi1wJxccwTmggtWkGHREmDDJ7zs6idPN4Bc6eDCfQhBG2yvKLXPXSquxZvs93RnhrKTdtnH3Ivm/GO2tdJ7hxQ+OGjTdc+hVb23DtDQ2sJR6NwoiPu7VtTi/M4hoLfURU5EyiJQlioQBUC+qi++m1RN+1c1mMxPgrBpRwhOfhe0g9CfB9XtmJSAAm+bwwG6GgfIHPWqxOIhGcppfyHPOubyDbwuIIX+E0KnJVdaImQ7iLw5NLa46r0pG4fy+421SU9HlGeEdA7xuNjUzq9fDDDKdW9itJcfnmnHaciIfxZ6XAkZ2UmHzkeDVrvC1mXqhptLuZc8SdSLycz0JfUFj+f9u7mlDbkqv8rX3O6/c6rSR2VIh2NAlKpCd21IFBB2IMqISMRAw6EcGJgwiK0A6cCTpRMxChiQqC+BcVJQNFMANHja0ZiGmb+Be6RWO67dj2+7n37KrlYP3X3ue++/r1O7fvu2ddzj1n7127flatv1pVtco0Okf6FayBAHSlEFnh1+UzddBGMqLOmDZC+L1ZHjxIFXY/hB03LFOHQLd5bCMaivExdEENNlPt+FB1RfqPbD82LSHZBUwQ+UiS1ReSS8k9tOylqJsjfOwzCsINmZ1ypVzGShsSIeQTXb3tVHlxCdIn+431LAhTJWlIl308wepDuYqPvLRNNz05LrPPI0v9oXzyetvVQKwg1fbx/t0W2FzgPDupr0oWQ0hzkswyriDRyB4MAJT6QxpKBA8AANfcwmCt6YEOgDCTPqehRowuswHU0aihTTMaZvTtDFzvmLYdmwm4hkmEx25COw2tx02Eh6zL6RAjNRi+awANN6cnYJom8JbQNxOwnUDXtpiubUFbW1orWAkTN53mkjSxMzIhbbkMDWqBGatEyO1PvxjInmrileCMyHnrSwDCgZiyJesneL3teyKLNjylvNPrKqR4ZNKc9xTrNAAq/hiB5djWDBRKdcqzBUXO2JWugIw+iJDlIZezBre7kyEVftfHcBRlU/Rx4C1XO/pV+pjAE2mcBMsTZ8LFMXtSiLEHjCKAJGrlmcPUdA2TrrMQiCk99nBUrtnJytHLEnKZnemZOjp1YOoS4hYM2gLTNZIpoU7iHOjsmloKTz4Aio4S7W4+ARUQJpknANME2uhnMBtzX2a5D9T+9f5W5ObRSp4+4+HtglPAiTwYsGr2MPkTAfq1fsaKaf+YgJ4Si6N85zXmQz5ZuTlzGNPSwJzRxBrvr+KtGFHIMOSTcUIZM8t38/BDfqUVXwgclI8nXrFyY2CCWugKcZzB8Be0nx3LxhJhVAojckamjrSciDR4lztLsIrW9VimeKOYsixsHt76jq4HOHQn3hh3xt46u++uPYmUq3/MXb32qT1D5W3cnsdeKEko9Wkliki38BolvNV7cn9ljMRRtTytF+nH5LzHl6DtdNwsWSKMjDHfwLHjwWSxe2dz/gNDLNqpQq4UH4JQSI5Ktyw2Fdr/NL+9HAYBA8JSraKf7Hbe41XqPvbVWExm7Pz8LkxucJ4TYd4POdPN4H0AfgHA7+ANnPUmeY4x2M2Mz0IrSU9KUda0sSFw9S3OvSQJWus4OT1Fmzvm1vzRlKZ3OtRLrtNsnTsahNntu6uDRHYV6fRb0mJm/ufNLwBJxBlT71MswClbF9UUmyakLbCRtwzn8xhXG6Hftp49RBwQ89KJWBNqht5INBd7BibPLg8L9lHUlASa+SV6KXJNX+W8XXdRooVExNkyIdeUKZvszHHGUQZNvgtQr8/BgB24ofmMMjSrhzyk8qGTJ5J7PAjLaFsVwB4XcNiJJygt0hcWeM5GAFBceRgEb89+OE8o6ReY+SlmfgrAtwO4BeBP8QbPevP/ZAhbE2fVgPRQ0lTvRWpGbGsrdUfvHU0jyYbfJJmkJiTcjLdlrfGH8tsTRzmeQr6b2QVstoGNaZOWRtBHGXMlLVA/idEtBaWU9jtNY1BkvlrvJVD9T9G6878/2AGpEUURlRiAFYPub/DyMSBjqcXd3irb4NKQory3LDtk6Hob5bWzmUkS0pAzarcY2aFex26ssCKj3WtCLPX5kPc+uFcz/kMA/oWZv0D3edYbDX95jOU/HDGEkHKpbeW1aKk5yJkZrTXd6Zb2gXNCqGlrauik+pzMiJf8/URXQKfdkjTVGsZfYpK0Uywf7zOpoJuIPKbchkii4kzye+FvGTuUYsmFB4Z0piLfamt4KlRWCD6wR+mqMjwK/ku16osIc7oK8AULLd6r/VnoYB8KVnACKMOnZ3Z4Rx56Vedcbk/iTG1HWJVBt/YkqlcFkNNloeeQekX3L9qx/qmmf63jXTkd987sPwLg9/T3uc56q5DIievH8TYQVQ5ZFQ4eTUB5agLxAtTR1zrabpatpQ3wweIk43kG0KeO3hva1EQwUEPn5kzdIBGvewca5zPdAUy68YXS6D55vdlCT/EEIvY1BcLgwHYCthNhMxG2G8K1zYRryvAWC22tf7OmCNPQHnJY32bpWKw1xyWX7DxTd7xZ+uyUy0mr5vGhhN8v9u9S01u9h/XhgrlRLCTWorhDMHrIjB3lOsM7rkJrZuvYauihp916JsejWV4loIShzMpTejT8BiYopaVoT5IwMWShErSyrPSpq35qA1TTF//QCtzVjDcgOSDiowD+aHx21llvRPSTRPQcET33am/5CaL5cV1frvezsyPlHwKT6ssMO7CxJ/TnvJPphDQezyZ7/Sq/2fNZfVLuLwT14NF1L68761I1c57Zq14cVrlhiq8c6A8p0MGe3qq5rGuKimIzgavWXJdQAdkazkuJkaub0w9tM/wBmeGTpWazEXmIkARl1H1sM5SOEkWm30tMrNAsVuqfkpTAG9aepVd2pWL5Xib29X5ag3vR7D8A4O+Z+Yt6fa6z3pj5GQDPAMCT12/ESM2lrkSZ5ZG6lRnzGucCKgnNmA08kK2O0f3kMvUmR5ppFFliEGRTCtkxrqrL2WLI9ebra2XJpW1GyRpXHUsUHSYly/N8rJWf0modNQzezRE3TbEM1rRSIRbBXkFVvkjhDJOw9M6AOaUymk2pIH1zeZN8bBRmdhI+eT6/VidyCpkjOBo7VKtmWo4BsfhyLV0FLtsdY97hGWXLIz1OLnH505N3QesuASpZln6AVStra+7+QsVJ3a0X3VSHbtWioqCJLLDLvXXBk+Hcmh3AxxAmPHAfZ72JGSznSS8+xvyS0E07c9Lp7UR0xjTSeVlXMgN97uhzB3eOziQGpg6iDpo6aGogagDPEv/dmb2BuIPQMSnDVwYOppto8kMcPNSzrVkmHe2zLrYNSYBoRkS0lcMdyRk9Pra6L6wRWZIbG3RMQAghCNayXRGEnPwlSf6sOpii4xDOANPiHaCmH8Up2WBnKW+AGJoll5zcd1PJRGl9xjyFlvCqsu8iFDx0rVMuIRvWmcGD0SdjeI41ABNiOJW/ad+HIix0yHMbOmh/VSMmyvW+iDON4PdMKgT1sXX0qA3OgHMxOxE9BuDDAP4k3f4lAB8mos8D+D69Ph8shNCSwBhDGxbp1x9y/qHConrcA4x5/JpHEmTkKb3zoZTKV6nMmBGNv6lSwl1KWWqf1Bbk/HJwJCOiuzVjz+hv0a6133uqn/wK43uMlUf7b+0vKE/RUaasfS8rLniPRi8pc5H781whba+PZMRJuC4Tj8NRl8YG+4o+E1HnP+vtJoB3Dvdewb2e9ZYIWTSZmiej5Le0MKanOj52fgqz2WbiLd5c1/3r3BjcTeuLbBO9I9NxTdPNndEa0GY9Vrm7inQkimJjv5507V8H+f0spEogDqKSXWg3OXyBXZvLtw0shCjYZUFBZ6iHpEKzAZ43SQYtjKmgeHMEUk6T5o1XTPfSsWy9ErrTMOD9mZx4tQ4mdj2+cG0sxKdi7aC47cggsqWphrjQuDlLuRdr+MBTQm6UW2tgAjulG4UDB3786Dpi7wXT9OGXIWX8yVHoebswiPS5TrzowLuroguJLmsNJduih2DwIAU4kcnWPqPn8G7qAbBOXsaMrUWgCh+SE8XhgawLZcBoyvCtxcej0RTFXrWmG19MaAxZNgtj3qqvbLqtLJOFWWC0+Jic6QRdYc+Om0XUlczwKIaIIzRbL1XvOAcuNHZh9DJ3jcToXN4fp6UmppQ1OaOzc0kY2b5fgbruGMwF5vRJr5agEtoroy8EKIwSY1+LNKSzQVmbUBFTUQeOPJ3vq7cxlakts2k/AggdRCOzE8rBDxlMYFGut9SDCL4NpNqe+5n+4na9ZSmX9A+nBGdNJOxrEjPUMdfL8HJFUaSwz8GWTjh7xxHjvXyfF2nzfquqc8f382UK1aX3B2N3T6NW7iycZmt5LGsW+dVyRsf/uvk9asehViuXNOKOU/0GQRRMWTWtO0xXispyEcPvNVjH7hJGayvep/Tfu3VRbqaUs3et8fLyPBVMcFBmF9LvkJPKOzboaGxnYClj0JRf0G+ueSiGp3TPRFyfG3anM9rMmHctzHEGqDOYErrZhAMknUp7aEgq5i6ho1kDXbDZy9ICCVcFMG8gDrgUWACqRU1jayQ52Q2n57D7IV2DQya9W0M4m1a3FEm0N9MYmQJWhOXKWJxWPmDbI21EO9qsXkvHp5jEydOsVYiR2srORsvJf+tMCcx2st+I8r2iUUis11A7zy1zLvtQIg3FVnWvny2KCia18jyNalnSDN2yoKgO2043JTBb1m+rQKE0H11hQm3U8ONQqV76pCf76o8zFeQF7HqTztygK8MTNmaWyHkogzZD6uiVKSPoCzpm7q0rs0uQR3Z7Ocr3nBgliKSQwAaMbja2Mvqk9c5zAsbwgDGsS3GruIcinnzOW4hpk/IadQ6CAh2S8WpKzf+zbJlWS0iCL4xLVguLpfdzv2jwzYHmJKovkC0ExyIbG1ptYg+D45uSS2HwB7i1kvo66saep0mMOtOUmYKUp4XhiWSzos0u2G5DUtxk7KXWhA6gjF240PPvUo3okCxoOefAFQ8EuDPQaT0rCrImsdJSnAvoSIU8F7LP6z33w8XtelvYf0ECvslg6ATvgkQ0buy7yW7r4Ru4Lxufco3l9JkRcz3T6x5oMgld37JqQtw0rTcnmcnZ5KwD66SoKLQGsmsNpa9LY1TCk1ENndXlYSHIq1m9UJTh2irhn4Lw1033Fcty6Oq1Juy7Oz4Zf2fBs3BO5YoSJ9QdMluoAAAIfUlEQVRHQ3yxTB4VlAxye8zf4EQndLegy1TO2OhMPFTpbanConTOrEAmXEOYcHlvPy6Bi4hBpxFaO7puAYVLafNOhklSmVMygPOOhPYRE87StNZwcnqC3a5h13T+2c5unNRI7OJ9lwU3cFPeNDkzu2lfPi1+x3u6caabp7jHXnmT9mrKi3YzaS9zrxNJMIJpQ9huJmy2coZczNcWNevttI20UMslay1zUoWikeEAZwSKKnTiNKzLi1NqQBZUi3WIhSEAPWtDyzVjdippo2eL0KCcCy2+AatGEHb9sxIjTwY07oChj3THWMZXrkT+jis2nGrxdlS4D9NSTYM7R2GS5/yH+fNhWxfSO5beKtNdq0PXoxCYNmBdT3HWuP+gzF5GFWxTKUHIQcJBEqC08QTpB8WYK9/vvWOeZ+xOZ8wtmUnGeKqRe7cVdsrUrB9N7Nfjc7cGgtFlri7tmPNQVcYwesbUpEyePIfG8NMETBvS89cTozvk8W50uHmSbZDhS2QtpRMNK64YvprPJ5Ytv8TkrsWqsMlQlFb+xSjrtHn1jVH7ZV29wurW5wg2X9do2haOu8VA6eQact25xsgGWtwVwWfbks3uiuOfcp2ziJIsbXENm3IygTqs5Iu8wmFHCCPRFqRVZneCWWuQw2EddMzY9I7t3HDt5AT9zh2Za59sDtqO2cyTafCpqDJnpRJWCBx+ptv1O3dw/c5t0OmMbSe0WRgU0wZEMxjAZu7YNtHA8ywbYHrv2J3u/Pc8y2msrTXM805CSs877OZTfX6KXd/pkGHGnIJOhloxZ9GEDcm8/IYY17oc5PxIY9xoE7bEeOz0FG87uYNHJ+DGHcL1W5OeBtNDZlMmqcQKtqGIZBWejW9DCZp2VgGbl7canaizUJhAfRB5Pr1Mt2U2TMTN5M+mRHv5dz30gNNGHqeSlGtqpwlrxWeeQrP3ihA06rE2EHQakVyeVR+AE2kiWFMWJGdIAYKbKVa85cXaS8aveTo+M3OS4ts616aH15bqAb6yEoiVpu31W2g3b6PfuiNLvPfAQZl9YsajpzvwzZuYX3kFp62hTYR5m8Ye6q00GjXh504Jzohkl3p2xNIjN0+w+fJN7HYzmDfo6kG3c9iZgbl1zE00+zzPyuCM3W6nTj09Z13H/3PbSdq2w253CnbG3+lQQISC1Kj7WN5AQkVvQTTJYY79FNM04ZG+xaM4wWa7wde+tsXXvXwdN65fx2M3b+ArX38Um4kQ8eOyNnKkyH2OcrJ32CM/OzJtzj4Tn7MnhMGh36bNjTlHA95bN2ikxJyqoRaMZexINR97MZJSMt1zHuSCFMkpGMOQ2Ingu9z8k/KyZwmJdT88vE48qRIiAmnsvNz2bHOsoykEbpHEhfHhwsi7ONDqbbdFSjaz1G/dxumL/4n26mvg3bxePg6u2YFrreH66SluvH4L280WbUNom7T4wvduZslIsVFGr/OGbe6yMIY7A7dPML9+C/PcAGwA2joR6IyahqoS4bDbzaHZdzPa3AA1x93Z12Zh8DZjN5+Cu0S+mecZdpBEtzG7ToP4UIVFG2y4y/p5nrBFF2anhhubjm3f4PE7t/H2m6/j+u4Eb+uneKyd6Dr5PO21Yr46bXLgZtAGzmDOaOx0FjiNmYbqFV4fy6YXBwJPTEv51sBEWlfK7xXGweJ3Jvh4t+LDZEMoj9gFl0wjbf9gtTiejPGjLXFKDkFmU2pdaRTAhrXS3mwpUXzHPOHSULG2eD6qtBIe+u07aK++hva//wdueWdphXvZCPPWgX10twf2yNplRveY7/nKuM+MzwUr+a8VOd5bpHmT6nnXct6E8vk8aR803t9AUQes0gg0mpwPtDCiLwG4CeDlgxV6MfDVePjbCFyNdl62Nn4jM3/N2oODMjsAENFzzPwdBy30wHAV2ghcjXY+TG28nGb8EY5whHuGI7Mf4QhXBC6C2Z+5gDIPDVehjcDVaOdD08aDj9mPcIQjXAwczfgjHOGKwEGZnYi+n4heIKJ/JqK7niBzGYCI3k1EnyGizxHRPxLRx/X+40T0V0T0ef3+qouu6/0CEW2I6LNE9Gm9fi8RPav9+QcabvzSAhG9g4g+RUT/RETPE9EHH6Z+PBizE9EGwK9DQlI/CeBjRPTkocp/gDAD+BlmfhLAdwL4KW3XPR+PdQng4wCeT9e/DOBXmfmbALwK4CcupFZvHnwCwF8w87cA+FZIWx+efoyTSx/sB8AHAfxlun4awNOHKv+A7fwzSCTeFwC8S++9C8ALF123+2zXExBi/14An4YsGnwZwHatfy/bB8DbAfwb1I+V7j80/XhIM/7rAbyYrl/Sew8NENF7AHwAwLN4Q8djvaXh1wD8HGKPyTsBfJmZbefFZe/P9wL4EoDf1qHKJzWE+kPTj0cH3ZsERPQVAP4YwE8z82v5GYtauLTTHkT0EQD/zcx/d9F1eYCwBfBtAH6DmT8AWdZdTPbL3o+HZPb/APDudP2E3rv0QETXIIz+u8xsB2l8UY/FwlnHY10S+C4AHyWifwfw+xBT/hMA3kFEtnPysvfnSwBeYuZn9fpTEOZ/aPrxkMz+twC+WT24j0BOhP3zA5b/QIBkP+hvAniemX8lPXrDx2O91YCZn2bmJ5j5PZB++2tm/lEAnwHwQ5rssrfxvwC8SETv11sfAvA5PET9eOhdbz8IGfttAPwWM//iwQp/QEBE3w3gbwD8A2I8+/OQcfsfAvgGAF8A8MPM/D8XUsk3EYjoewD8LDN/hIjeB9H0jwP4LIAfY+aTi6zf/QARPQXgkwAeAfCvAH4cohAfin48rqA7whGuCBwddEc4whWBI7Mf4QhXBI7MfoQjXBE4MvsRjnBF4MjsRzjCFYEjsx/hCFcEjsx+hCNcETgy+xGOcEXg/wHunHE5tZLuMAAAAABJRU5ErkJggg==\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "X_train, X_val_test, y_train, y_val_test = train_test_split(X_train, y_train, test_size=0.4,random_state=42,stratify=y_train)\n",
        "X_val, X_test, y_val, y_test = train_test_split(X_val_test, y_val_test, test_size=0.5,random_state=42,stratify=y_val_test)"
      ],
      "metadata": {
        "id": "CGp9mEz5OfsO"
      },
      "execution_count": 6,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "y_cat_train = to_categorical(y_train,29)\n",
        "y_cat_val = to_categorical(y_val, 29)\n",
        "y_cat_test = to_categorical(y_test,29)"
      ],
      "metadata": {
        "id": "z_T6UfbsOuti"
      },
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "print(X_train.shape)\n",
        "print(y_train.shape)\n",
        "print(X_val.shape)\n",
        "print(y_val.shape)\n",
        "print(X_test.shape)\n",
        "print(y_test.shape)\n",
        "print(y_cat_train.shape)\n",
        "print(y_cat_val.shape)\n",
        "print(y_cat_test.shape)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QgiX6mPZPDYL",
        "outputId": "dea18807-7b31-4baf-8daf-bea95a992950"
      },
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(52200, 76, 76, 3)\n",
            "(52200,)\n",
            "(17400, 76, 76, 3)\n",
            "(17400,)\n",
            "(17400, 76, 76, 3)\n",
            "(17400,)\n",
            "(52200, 29)\n",
            "(17400, 29)\n",
            "(17400, 29)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "print(X_train[0].shape)\n",
        "\n",
        "image = cv2.resize(X_train[0], (299, 299))\n",
        "\n",
        "plt.imshow(image)\n",
        "print(y_train[0])"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "hUDaGbCVrThA",
        "outputId": "5d40e5b0-d265-4c85-9859-cce287f580ed"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(76, 76, 3)\n",
            "22\n"
          ]
        },
        {
          "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": [
        "from tensorflow import keras\n",
        "from tensorflow.keras import layers\n",
        "\n",
        "data_augmentation = keras.Sequential(\n",
        "    [layers.RandomFlip(\"horizontal\"), layers.RandomRotation(0.1),]\n",
        ")"
      ],
      "metadata": {
        "id": "F8yHxaDFL1KZ"
      },
      "execution_count": 8,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "from keras.engine import training\n",
        "from keras.models import Sequential\n",
        "from keras.models import Model\n",
        "from keras.callbacks import ModelCheckpoint, LearningRateScheduler, EarlyStopping, ReduceLROnPlateau, TensorBoard\n",
        "from keras import optimizers, losses, activations, models\n",
        "from keras.layers import Convolution2D, Dense, Input, Flatten, Dropout, MaxPooling2D, BatchNormalization, GlobalAveragePooling2D, Concatenate\n",
        "from keras import applications"
      ],
      "metadata": {
        "id": "v9jPeBzoKVRt"
      },
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "source": [
        "## Inception Feature Extraction"
      ],
      "metadata": {
        "id": "ub_xAe31w6sA"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "base_model = tf.keras.applications.InceptionV3(weights='imagenet', \n",
        "                                               include_top=False,\n",
        "                                               input_shape=(76, 76, 3))\n",
        "base_model.trainable = False\n",
        "nclass = 29"
      ],
      "metadata": {
        "id": "zg4T_uJgiU02"
      },
      "execution_count": 10,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "inception = Model(inputs=base_model.input, \n",
        "                  outputs=base_model.get_layer('mixed3').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(inception)\n",
        "feature_extraction.add(Flatten())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "sGTPI6kDo8q3",
        "outputId": "f21812b3-e755-42d0-a027-af45bdf608a2"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_2\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model_2 (Functional)        (None, 3, 3, 768)         2146976   \n",
            "                                                                 \n",
            " flatten_1 (Flatten)         (None, 6912)              0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                200477    \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 2,347,453\n",
            "Trainable params: 200,477\n",
            "Non-trainable params: 2,146,976\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "file_path=\"/content/best_model.h5\"\n",
        "\n",
        "# checkpoint = ModelCheckpoint(file_path, monitor='acc', verbose=1, save_best_only=True, mode='max')\n",
        "mc = ModelCheckpoint(file_path, monitor='val_loss', verbose=1, mode='min', save_best_only=True)\n",
        "\n",
        "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=10)\n",
        "\n",
        "callbacks_list = [es, mc] #early"
      ],
      "metadata": {
        "id": "FtsLuRymdmNp"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=16,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pdrZgABMpEsH",
        "outputId": "4d3d6dd7-34b9-4d5c-adb2-e538bfdffbad"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 2.36362, saving model to /content/best_model.h5\n",
            "3263/3263 - 20s - loss: 1.9428 - accuracy: 0.9655 - val_loss: 2.3636 - val_accuracy: 0.9595 - 20s/epoch - 6ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss improved from 2.36362 to 1.02367, saving model to /content/best_model.h5\n",
            "3263/3263 - 21s - loss: 1.5258 - accuracy: 0.9751 - val_loss: 1.0237 - val_accuracy: 0.9820 - 21s/epoch - 6ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss improved from 1.02367 to 0.63740, saving model to /content/best_model.h5\n",
            "3263/3263 - 21s - loss: 1.2984 - accuracy: 0.9800 - val_loss: 0.6374 - val_accuracy: 0.9882 - 21s/epoch - 6ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss did not improve from 0.63740\n",
            "3263/3263 - 20s - loss: 1.2905 - accuracy: 0.9817 - val_loss: 1.0510 - val_accuracy: 0.9830 - 20s/epoch - 6ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss improved from 0.63740 to 0.53139, saving model to /content/best_model.h5\n",
            "3263/3263 - 20s - loss: 1.1724 - accuracy: 0.9850 - val_loss: 0.5314 - val_accuracy: 0.9920 - 20s/epoch - 6ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss did not improve from 0.53139\n",
            "3263/3263 - 20s - loss: 0.9395 - accuracy: 0.9870 - val_loss: 4.8568 - val_accuracy: 0.9447 - 20s/epoch - 6ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss did not improve from 0.53139\n",
            "3263/3263 - 21s - loss: 0.9068 - accuracy: 0.9888 - val_loss: 0.5440 - val_accuracy: 0.9898 - 21s/epoch - 6ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.53139\n",
            "3263/3263 - 20s - loss: 0.8902 - accuracy: 0.9892 - val_loss: 0.5650 - val_accuracy: 0.9905 - 20s/epoch - 6ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss did not improve from 0.53139\n",
            "3263/3263 - 20s - loss: 0.7710 - accuracy: 0.9911 - val_loss: 3.5857 - val_accuracy: 0.9690 - 20s/epoch - 6ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss improved from 0.53139 to 0.48804, saving model to /content/best_model.h5\n",
            "3263/3263 - 21s - loss: 0.7151 - accuracy: 0.9919 - val_loss: 0.4880 - val_accuracy: 0.9933 - 21s/epoch - 6ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss improved from 0.48804 to 0.31213, saving model to /content/best_model.h5\n",
            "3263/3263 - 20s - loss: 0.7654 - accuracy: 0.9910 - val_loss: 0.3121 - val_accuracy: 0.9948 - 20s/epoch - 6ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss did not improve from 0.31213\n",
            "3263/3263 - 20s - loss: 0.6770 - accuracy: 0.9922 - val_loss: 0.3981 - val_accuracy: 0.9944 - 20s/epoch - 6ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss did not improve from 0.31213\n",
            "3263/3263 - 20s - loss: 0.5794 - accuracy: 0.9929 - val_loss: 2.0405 - val_accuracy: 0.9806 - 20s/epoch - 6ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.31213\n",
            "3263/3263 - 20s - loss: 0.5477 - accuracy: 0.9943 - val_loss: 2.2349 - val_accuracy: 0.9809 - 20s/epoch - 6ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss improved from 0.31213 to 0.14368, saving model to /content/best_model.h5\n",
            "3263/3263 - 20s - loss: 0.5156 - accuracy: 0.9943 - val_loss: 0.1437 - val_accuracy: 0.9983 - 20s/epoch - 6ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.6456 - accuracy: 0.9941 - val_loss: 1.1933 - val_accuracy: 0.9889 - 20s/epoch - 6ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.4736 - accuracy: 0.9946 - val_loss: 0.7234 - val_accuracy: 0.9928 - 20s/epoch - 6ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.6064 - accuracy: 0.9946 - val_loss: 0.4185 - val_accuracy: 0.9951 - 20s/epoch - 6ms/step\n",
            "Epoch 19/30\n",
            "\n",
            "Epoch 19: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.4365 - accuracy: 0.9957 - val_loss: 1.1352 - val_accuracy: 0.9922 - 20s/epoch - 6ms/step\n",
            "Epoch 20/30\n",
            "\n",
            "Epoch 20: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.4382 - accuracy: 0.9963 - val_loss: 0.9176 - val_accuracy: 0.9909 - 20s/epoch - 6ms/step\n",
            "Epoch 21/30\n",
            "\n",
            "Epoch 21: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.4324 - accuracy: 0.9959 - val_loss: 0.3220 - val_accuracy: 0.9971 - 20s/epoch - 6ms/step\n",
            "Epoch 22/30\n",
            "\n",
            "Epoch 22: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.4198 - accuracy: 0.9957 - val_loss: 0.1585 - val_accuracy: 0.9984 - 20s/epoch - 6ms/step\n",
            "Epoch 23/30\n",
            "\n",
            "Epoch 23: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.3549 - accuracy: 0.9965 - val_loss: 0.7052 - val_accuracy: 0.9926 - 20s/epoch - 6ms/step\n",
            "Epoch 24/30\n",
            "\n",
            "Epoch 24: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.5267 - accuracy: 0.9956 - val_loss: 0.3162 - val_accuracy: 0.9965 - 20s/epoch - 6ms/step\n",
            "Epoch 25/30\n",
            "\n",
            "Epoch 25: val_loss did not improve from 0.14368\n",
            "3263/3263 - 20s - loss: 0.3300 - accuracy: 0.9968 - val_loss: 0.1748 - val_accuracy: 0.9975 - 20s/epoch - 6ms/step\n",
            "Epoch 25: early stopping\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f56767c6590>"
            ]
          },
          "metadata": {},
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "inception_fe1 = [2.3636, 1.0237, 0.6374, 1.0510, 0.5314, 4.8568, 4.8568, 0.5650, 3.5857, 0.4880, 0.3121, 0.3981, 2.0405, 2.2349, 0.1437, 1.1933, 0.7234, 0.4185, 1.1352, 0.9176, 0.3220, 0.1585, 0.7052, 0.3162, 0.1748]"
      ],
      "metadata": {
        "id": "XP2jh6m9aKod"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "dNqIma43pcxI",
        "outputId": "6b7ccda5-6a17-4db9-92a4-893cc9068216"
      },
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 4s 7ms/step - loss: 0.3275 - accuracy: 0.9967\n",
            "Accuracy for test images: 99.672 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "inception = Model(inputs=base_model.input, \n",
        "                  outputs=base_model.get_layer('mixed7').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(inception)\n",
        "feature_extraction.add(Flatten())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "KeGhWXtDQaWM",
        "outputId": "ae3202c2-128f-4a2b-cfc2-9c9142990358"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model (Functional)          (None, 3, 3, 768)         8975264   \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 6912)              0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                200477    \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 9,175,741\n",
            "Trainable params: 200,477\n",
            "Non-trainable params: 8,975,264\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=16,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ptUuCh_DR_IT",
        "outputId": "a136d2dc-0416-45f2-f348-d7c7738815ea"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 0.76080, saving model to /content/best_model.h5\n",
            "3263/3263 - 45s - loss: 1.5454 - accuracy: 0.8785 - val_loss: 0.7608 - val_accuracy: 0.9481 - 45s/epoch - 14ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss improved from 0.76080 to 0.46017, saving model to /content/best_model.h5\n",
            "3263/3263 - 39s - loss: 0.6996 - accuracy: 0.9570 - val_loss: 0.4602 - val_accuracy: 0.9695 - 39s/epoch - 12ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss improved from 0.46017 to 0.38998, saving model to /content/best_model.h5\n",
            "3263/3263 - 39s - loss: 0.5066 - accuracy: 0.9726 - val_loss: 0.3900 - val_accuracy: 0.9809 - 39s/epoch - 12ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss did not improve from 0.38998\n",
            "3263/3263 - 37s - loss: 0.4387 - accuracy: 0.9795 - val_loss: 1.2342 - val_accuracy: 0.9556 - 37s/epoch - 11ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss improved from 0.38998 to 0.32309, saving model to /content/best_model.h5\n",
            "3263/3263 - 37s - loss: 0.3812 - accuracy: 0.9832 - val_loss: 0.3231 - val_accuracy: 0.9843 - 37s/epoch - 11ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss did not improve from 0.32309\n",
            "3263/3263 - 37s - loss: 0.2898 - accuracy: 0.9875 - val_loss: 0.9467 - val_accuracy: 0.9681 - 37s/epoch - 11ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss improved from 0.32309 to 0.13956, saving model to /content/best_model.h5\n",
            "3263/3263 - 37s - loss: 0.2539 - accuracy: 0.9888 - val_loss: 0.1396 - val_accuracy: 0.9930 - 37s/epoch - 11ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.13956\n",
            "3263/3263 - 37s - loss: 0.2638 - accuracy: 0.9902 - val_loss: 0.2668 - val_accuracy: 0.9887 - 37s/epoch - 11ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss did not improve from 0.13956\n",
            "3263/3263 - 37s - loss: 0.2105 - accuracy: 0.9921 - val_loss: 0.3990 - val_accuracy: 0.9854 - 37s/epoch - 11ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss did not improve from 0.13956\n",
            "3263/3263 - 37s - loss: 0.1958 - accuracy: 0.9928 - val_loss: 0.5478 - val_accuracy: 0.9828 - 37s/epoch - 11ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss improved from 0.13956 to 0.09843, saving model to /content/best_model.h5\n",
            "3263/3263 - 37s - loss: 0.2430 - accuracy: 0.9917 - val_loss: 0.0984 - val_accuracy: 0.9951 - 37s/epoch - 11ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1737 - accuracy: 0.9940 - val_loss: 0.7547 - val_accuracy: 0.9809 - 37s/epoch - 11ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1464 - accuracy: 0.9940 - val_loss: 0.2338 - val_accuracy: 0.9931 - 37s/epoch - 11ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.09843\n",
            "3263/3263 - 38s - loss: 0.1618 - accuracy: 0.9941 - val_loss: 0.1298 - val_accuracy: 0.9960 - 38s/epoch - 12ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss did not improve from 0.09843\n",
            "3263/3263 - 38s - loss: 0.1557 - accuracy: 0.9947 - val_loss: 0.1642 - val_accuracy: 0.9942 - 38s/epoch - 12ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss did not improve from 0.09843\n",
            "3263/3263 - 38s - loss: 0.1385 - accuracy: 0.9961 - val_loss: 0.1407 - val_accuracy: 0.9948 - 38s/epoch - 12ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1423 - accuracy: 0.9949 - val_loss: 0.2471 - val_accuracy: 0.9944 - 37s/epoch - 11ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1258 - accuracy: 0.9956 - val_loss: 0.1437 - val_accuracy: 0.9955 - 37s/epoch - 11ms/step\n",
            "Epoch 19/30\n",
            "\n",
            "Epoch 19: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1113 - accuracy: 0.9965 - val_loss: 0.2240 - val_accuracy: 0.9932 - 37s/epoch - 11ms/step\n",
            "Epoch 20/30\n",
            "\n",
            "Epoch 20: val_loss did not improve from 0.09843\n",
            "3263/3263 - 37s - loss: 0.1116 - accuracy: 0.9963 - val_loss: 0.1446 - val_accuracy: 0.9956 - 37s/epoch - 11ms/step\n",
            "Epoch 21/30\n",
            "\n",
            "Epoch 21: val_loss did not improve from 0.09843\n",
            "3263/3263 - 38s - loss: 0.1246 - accuracy: 0.9963 - val_loss: 0.1672 - val_accuracy: 0.9960 - 38s/epoch - 12ms/step\n",
            "Epoch 21: early stopping\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7efc743aa590>"
            ]
          },
          "metadata": {},
          "execution_count": 15
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "inception_fe2 = [0.7608, 0.4602, 0.3900, 1.2342, 0.3231, 0.9467, 0.1396, 0.2668, 0.3990, 0.5478, 0.0984, 0.7547, 0.2338, 0.1298, 0.1642, 0.1407, 0.2471, 0.1437, 0.2240, 0.1446, 0.1672]"
      ],
      "metadata": {
        "id": "yqppbbard1Zg"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EppHGc9totAf",
        "outputId": "ea5b0813-97d1-465d-d6a5-0aa8b999b263"
      },
      "execution_count": 16,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 7s 13ms/step - loss: 0.1322 - accuracy: 0.9963\n",
            "Accuracy for test images: 99.632 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "inception = Model(inputs=base_model.input, \n",
        "                  outputs=base_model.get_layer('mixed9_0').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(inception)\n",
        "feature_extraction.add(Flatten())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "A_0gFDA0qEex",
        "outputId": "c4ee04e8-a22e-46a0-f948-cac988d29ddd"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model (Functional)          (None, 1, 1, 768)         12054560  \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 768)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                22301     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 12,076,861\n",
            "Trainable params: 22,301\n",
            "Non-trainable params: 12,054,560\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=16,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "N5Qy4mLTqu7h",
        "outputId": "20ac0742-0370-4822-8ef1-c9666899430d"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 0.38288, saving model to /content/best_model.h5\n",
            "3263/3263 - 50s - loss: 0.6065 - accuracy: 0.8240 - val_loss: 0.3829 - val_accuracy: 0.8794 - 50s/epoch - 15ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss improved from 0.38288 to 0.30682, saving model to /content/best_model.h5\n",
            "3263/3263 - 47s - loss: 0.2873 - accuracy: 0.9100 - val_loss: 0.3068 - val_accuracy: 0.9003 - 47s/epoch - 15ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss improved from 0.30682 to 0.27090, saving model to /content/best_model.h5\n",
            "3263/3263 - 47s - loss: 0.2258 - accuracy: 0.9277 - val_loss: 0.2709 - val_accuracy: 0.9129 - 47s/epoch - 15ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss improved from 0.27090 to 0.26951, saving model to /content/best_model.h5\n",
            "3263/3263 - 47s - loss: 0.1980 - accuracy: 0.9355 - val_loss: 0.2695 - val_accuracy: 0.9151 - 47s/epoch - 14ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss did not improve from 0.26951\n",
            "3263/3263 - 47s - loss: 0.1721 - accuracy: 0.9442 - val_loss: 0.2784 - val_accuracy: 0.9132 - 47s/epoch - 14ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss improved from 0.26951 to 0.25739, saving model to /content/best_model.h5\n",
            "3263/3263 - 47s - loss: 0.1580 - accuracy: 0.9483 - val_loss: 0.2574 - val_accuracy: 0.9213 - 47s/epoch - 15ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss did not improve from 0.25739\n",
            "3263/3263 - 47s - loss: 0.1443 - accuracy: 0.9525 - val_loss: 0.2718 - val_accuracy: 0.9210 - 47s/epoch - 14ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss improved from 0.25739 to 0.25632, saving model to /content/best_model.h5\n",
            "3263/3263 - 47s - loss: 0.1352 - accuracy: 0.9557 - val_loss: 0.2563 - val_accuracy: 0.9236 - 47s/epoch - 15ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.1255 - accuracy: 0.9584 - val_loss: 0.2693 - val_accuracy: 0.9239 - 47s/epoch - 14ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.1218 - accuracy: 0.9593 - val_loss: 0.2658 - val_accuracy: 0.9252 - 47s/epoch - 14ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.1137 - accuracy: 0.9617 - val_loss: 0.2711 - val_accuracy: 0.9263 - 47s/epoch - 14ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.1072 - accuracy: 0.9645 - val_loss: 0.2814 - val_accuracy: 0.9229 - 47s/epoch - 14ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.1054 - accuracy: 0.9654 - val_loss: 0.2793 - val_accuracy: 0.9248 - 47s/epoch - 14ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.0982 - accuracy: 0.9663 - val_loss: 0.2918 - val_accuracy: 0.9244 - 47s/epoch - 14ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.0957 - accuracy: 0.9670 - val_loss: 0.2742 - val_accuracy: 0.9286 - 47s/epoch - 14ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.0913 - accuracy: 0.9689 - val_loss: 0.3047 - val_accuracy: 0.9237 - 47s/epoch - 14ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.0892 - accuracy: 0.9700 - val_loss: 0.3080 - val_accuracy: 0.9232 - 47s/epoch - 14ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 0.25632\n",
            "3263/3263 - 47s - loss: 0.0848 - accuracy: 0.9717 - val_loss: 0.2909 - val_accuracy: 0.9301 - 47s/epoch - 14ms/step\n",
            "Epoch 18: early stopping\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fb840163fd0>"
            ]
          },
          "metadata": {},
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "inception_fe3 = [0.3676, 0.2993, 0.2748, 0.2683, 0.2715, 0.2471, 0.2809, 0.2545, 0.2586, 0.2856, 0.2849, 0.2792, 0.2801, 0.2796, 0.2967, 0.2948]"
      ],
      "metadata": {
        "id": "I9Ygz3fziieR"
      },
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RCGJx4SvqxQ8",
        "outputId": "21ad581a-17e3-4a41-bb17-56c5244e65b7"
      },
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 8s 15ms/step - loss: 0.2876 - accuracy: 0.9274\n",
            "Accuracy for test images: 92.741 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## VGG16"
      ],
      "metadata": {
        "id": "DkjvanJfW8jj"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "base_model = tf.keras.applications.VGG16(weights='imagenet', \n",
        "                                    include_top=False, \n",
        "                                    input_shape=(76, 76,3))\n",
        "base_model.trainable = False\n",
        "\n",
        "inputs = keras.Input(shape=(76,76,3))\n",
        "x = data_augmentation(inputs)\n",
        "\n",
        "x = base_model(x, training=False)\n",
        "x = keras.layers.GlobalAveragePooling2D()(x)\n",
        "# x = keras.layers.Dropout(0.2)(x)  # Regularize with dropout\n",
        "outputs = keras.layers.Dense(nclass, activation='softmax')(x)\n",
        "model = keras.Model(inputs, outputs)\n",
        "model.compile(loss='categorical_crossentropy', optimizer=keras.optimizers.RMSprop(), metrics=['accuracy'])\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "EgkqOPLCAeSP",
        "outputId": "a51a0c6c-9ae5-430d-9f2c-24368d6f6ee8"
      },
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/vgg16/vgg16_weights_tf_dim_ordering_tf_kernels_notop.h5\n",
            "58892288/58889256 [==============================] - 1s 0us/step\n",
            "58900480/58889256 [==============================] - 1s 0us/step\n",
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_3 (InputLayer)        [(None, 76, 76, 3)]       0         \n",
            "                                                                 \n",
            " sequential (Sequential)     (None, 76, 76, 3)         0         \n",
            "                                                                 \n",
            " vgg16 (Functional)          (None, 2, 2, 512)         14714688  \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 512)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 29)                14877     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,729,565\n",
            "Trainable params: 14,877\n",
            "Non-trainable params: 14,714,688\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "file_path=\"/content/best_model.h5\"\n",
        "\n",
        "# checkpoint = ModelCheckpoint(file_path, monitor='acc', verbose=1, save_best_only=True, mode='max')\n",
        "mc = ModelCheckpoint(file_path, monitor='val_loss', verbose=1, mode='min', save_best_only=True)\n",
        "\n",
        "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=20)\n",
        "\n",
        "callbacks_list = [es, mc] #early\n"
      ],
      "metadata": {
        "id": "jGelabxFlhSI"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.fit(x=X_train,\n",
        "          y=y_cat_train,\n",
        "          validation_data=(X_val, y_cat_val),\n",
        "          batch_size=32,\n",
        "          epochs=30, \n",
        "          shuffle=True, \n",
        "          verbose=2,\n",
        "          callbacks=callbacks_list)"
      ],
      "metadata": {
        "id": "rOjfgg1DBG5h"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "vgg_loss = [1.7572, 1.3727, 1.1815, 1.0704, 0.9933, 0.9286, 0.8818, 0.8435, 0.8132, 0.7917, 0.7734, 0.7515, 0.7269, 0.7259, 0.7073, 0.6938, 0.6851, 0.6713, 0.6692, 0.6626, 0.6525, 0.6395, 0.6318, 0.6301, 0.6086, 0.6120, 0.6147, 0.6100, 0.6044, 0.5925]\n",
        "vgg_acc = [0.5866, 0.6645, 0.7047, 0.7250, 0.7417, 0.7557, 0.7658, 0.7731, 0.7801, 0.7809, 0.7838, 0.7888, 0.7936, 0.7957, 0.7934, 0.8004, 0.7997, 0.8062, 0.8044, 0.8034, 0.8086, 0.8146, 0.8137, 0.8127, 0.8197, 0.8179, 0.8126, 0.8170, 0.8199, 0.8223]"
      ],
      "metadata": {
        "id": "nGp3eJAG717n"
      },
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = model.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "aWw37qrsGgt4",
        "outputId": "595941f1-77b4-4820-85af-47dec455d37d"
      },
      "execution_count": 19,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 13s 24ms/step - loss: 0.5856 - accuracy: 0.8257\n",
            "Accuracy for test images: 82.575 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## VGG Fine Tuning"
      ],
      "metadata": {
        "id": "g2DZVqeQ7EJZ"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "model.layers[2].trainable = True\n",
        "model.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "pKeWnUki4cy0",
        "outputId": "b424f130-2319-4904-e7dd-7d19de48ebe7"
      },
      "execution_count": 27,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " input_3 (InputLayer)        [(None, 76, 76, 3)]       0         \n",
            "                                                                 \n",
            " sequential (Sequential)     (None, 76, 76, 3)         0         \n",
            "                                                                 \n",
            " vgg16 (Functional)          (None, 2, 2, 512)         14714688  \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 512)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dense_1 (Dense)             (None, 29)                14877     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 14,729,565\n",
            "Trainable params: 14,729,565\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "model.compile(\n",
        "    optimizer=keras.optimizers.Adam(1e-5),  # Low learning rate\n",
        "    loss='categorical_crossentropy',\n",
        "    metrics=['accuracy'],\n",
        ")"
      ],
      "metadata": {
        "id": "TUNH54ji7CPy"
      },
      "execution_count": 28,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = model.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Pretrain make sure: Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "4rnSJDUw7J5-",
        "outputId": "4740b75b-c36c-4a33-89df-3ed4933df94b"
      },
      "execution_count": 29,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 13s 24ms/step - loss: 0.5856 - accuracy: 0.8257\n",
            "Pretrain make sure: Accuracy for test images: 82.575 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "file_path=\"/content/best_model.h5\"\n",
        "\n",
        "# checkpoint = ModelCheckpoint(file_path, monitor='acc', verbose=1, save_best_only=True, mode='max')\n",
        "mc = ModelCheckpoint(file_path, monitor='val_loss', verbose=1, mode='min', save_best_only=True)\n",
        "\n",
        "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=10)\n",
        "\n",
        "callbacks_list = [es, mc] #early"
      ],
      "metadata": {
        "id": "aRjXcR2_lkVZ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "model.fit(x=X_train,\n",
        "          y=y_cat_train,\n",
        "          validation_data=(X_val, y_cat_val),\n",
        "          batch_size=32,\n",
        "          epochs=10, \n",
        "          shuffle=True, \n",
        "          verbose=2,\n",
        "          callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3Y_PDo7i7IeU",
        "outputId": "61d339a3-5402-4d8c-c898-552dbc6402ea"
      },
      "execution_count": 30,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/10\n",
            "\n",
            "Epoch 1: val_loss improved from 0.59253 to 0.06947, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.2436 - accuracy: 0.9212 - val_loss: 0.0695 - val_accuracy: 0.9772 - 128s/epoch - 78ms/step\n",
            "Epoch 2/10\n",
            "\n",
            "Epoch 2: val_loss improved from 0.06947 to 0.03214, saving model to /content/best_model.h5\n",
            "1632/1632 - 127s - loss: 0.0699 - accuracy: 0.9778 - val_loss: 0.0321 - val_accuracy: 0.9885 - 127s/epoch - 78ms/step\n",
            "Epoch 3/10\n",
            "\n",
            "Epoch 3: val_loss improved from 0.03214 to 0.02396, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0431 - accuracy: 0.9857 - val_loss: 0.0240 - val_accuracy: 0.9922 - 128s/epoch - 78ms/step\n",
            "Epoch 4/10\n",
            "\n",
            "Epoch 4: val_loss improved from 0.02396 to 0.01751, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0307 - accuracy: 0.9903 - val_loss: 0.0175 - val_accuracy: 0.9939 - 128s/epoch - 78ms/step\n",
            "Epoch 5/10\n",
            "\n",
            "Epoch 5: val_loss improved from 0.01751 to 0.00622, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0219 - accuracy: 0.9929 - val_loss: 0.0062 - val_accuracy: 0.9983 - 128s/epoch - 78ms/step\n",
            "Epoch 6/10\n",
            "\n",
            "Epoch 6: val_loss did not improve from 0.00622\n",
            "1632/1632 - 127s - loss: 0.0212 - accuracy: 0.9933 - val_loss: 0.0127 - val_accuracy: 0.9956 - 127s/epoch - 78ms/step\n",
            "Epoch 7/10\n",
            "\n",
            "Epoch 7: val_loss improved from 0.00622 to 0.00543, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0160 - accuracy: 0.9947 - val_loss: 0.0054 - val_accuracy: 0.9984 - 128s/epoch - 78ms/step\n",
            "Epoch 8/10\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.00543\n",
            "1632/1632 - 128s - loss: 0.0159 - accuracy: 0.9951 - val_loss: 0.0059 - val_accuracy: 0.9979 - 128s/epoch - 78ms/step\n",
            "Epoch 9/10\n",
            "\n",
            "Epoch 9: val_loss improved from 0.00543 to 0.00517, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0096 - accuracy: 0.9972 - val_loss: 0.0052 - val_accuracy: 0.9981 - 128s/epoch - 78ms/step\n",
            "Epoch 10/10\n",
            "\n",
            "Epoch 10: val_loss improved from 0.00517 to 0.00251, saving model to /content/best_model.h5\n",
            "1632/1632 - 128s - loss: 0.0130 - accuracy: 0.9960 - val_loss: 0.0025 - val_accuracy: 0.9993 - 128s/epoch - 78ms/step\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f32e2b22690>"
            ]
          },
          "metadata": {},
          "execution_count": 30
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg_ft = [0.0695, 0.0321, 0.0240, 0.0175, 0.0062, 0.0127, 0.0054, 0.0059, 0.0052, 0.0025]"
      ],
      "metadata": {
        "id": "aBbFDoqdcq5V"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = model.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Pretrain make sure: Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "BLLOgm2h7Kk8",
        "outputId": "9c2b2055-3567-400d-c18e-43c26d7ef8e9"
      },
      "execution_count": 31,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 13s 24ms/step - loss: 0.0029 - accuracy: 0.9990\n",
            "Pretrain make sure: Accuracy for test images: 99.902 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## VGG Feature Extraction"
      ],
      "metadata": {
        "id": "CJMUA1Q27MVl"
      }
    },
    {
      "cell_type": "code",
      "source": [
        "base_model = tf.keras.applications.VGG16(weights='imagenet', \n",
        "                                    include_top=False, \n",
        "                                    input_shape=(76, 76,3))\n",
        "base_model.trainable = False\n",
        "nclass = 29"
      ],
      "metadata": {
        "id": "lUyZ8ewjGpYv"
      },
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "vgg = Model(inputs=base_model.input, \n",
        "            outputs=base_model.get_layer('block2_pool').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(vgg)\n",
        "feature_extraction.add(GlobalAveragePooling2D())\n",
        "feature_extraction.add(Flatten())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "js4dERN4Hz4l",
        "outputId": "605d983a-0b5f-4348-99f6-0b7d9bfa6d61"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model (Functional)          (None, 19, 19, 128)       260160    \n",
            "                                                                 \n",
            " global_average_pooling2d_1   (None, 128)              0         \n",
            " (GlobalAveragePooling2D)                                        \n",
            "                                                                 \n",
            " flatten (Flatten)           (None, 128)               0         \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                3741      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 263,901\n",
            "Trainable params: 3,741\n",
            "Non-trainable params: 260,160\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "file_path=\"/content/best_model.h5\"\n",
        "\n",
        "# checkpoint = ModelCheckpoint(file_path, monitor='acc', verbose=1, save_best_only=True, mode='max')\n",
        "mc = ModelCheckpoint(file_path, monitor='val_loss', verbose=1, mode='min', save_best_only=True)\n",
        "\n",
        "es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=10)\n",
        "\n",
        "callbacks_list = [es, mc] #early"
      ],
      "metadata": {
        "id": "X5vWdWfhV_ls"
      },
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=8,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "VNqUw9YCNWLa",
        "outputId": "8b673db2-b480-4eaa-ba72-1d5bec3360a6"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 2.91959, saving model to /content/best_model.h5\n",
            "6525/6525 - 41s - loss: 3.5101 - accuracy: 0.3461 - val_loss: 2.9196 - val_accuracy: 0.4176 - 41s/epoch - 6ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss improved from 2.91959 to 2.04898, saving model to /content/best_model.h5\n",
            "6525/6525 - 31s - loss: 2.7789 - accuracy: 0.4905 - val_loss: 2.0490 - val_accuracy: 0.5480 - 31s/epoch - 5ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss did not improve from 2.04898\n",
            "6525/6525 - 32s - loss: 2.4511 - accuracy: 0.5494 - val_loss: 2.1687 - val_accuracy: 0.5862 - 32s/epoch - 5ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss did not improve from 2.04898\n",
            "6525/6525 - 32s - loss: 2.3275 - accuracy: 0.5829 - val_loss: 2.5096 - val_accuracy: 0.6157 - 32s/epoch - 5ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss did not improve from 2.04898\n",
            "6525/6525 - 32s - loss: 2.2117 - accuracy: 0.6110 - val_loss: 4.8075 - val_accuracy: 0.4945 - 32s/epoch - 5ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss improved from 2.04898 to 1.89344, saving model to /content/best_model.h5\n",
            "6525/6525 - 33s - loss: 2.1473 - accuracy: 0.6247 - val_loss: 1.8934 - val_accuracy: 0.6561 - 33s/epoch - 5ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss improved from 1.89344 to 1.50382, saving model to /content/best_model.h5\n",
            "6525/6525 - 32s - loss: 2.0104 - accuracy: 0.6517 - val_loss: 1.5038 - val_accuracy: 0.6971 - 32s/epoch - 5ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss improved from 1.50382 to 1.39179, saving model to /content/best_model.h5\n",
            "6525/6525 - 33s - loss: 1.9987 - accuracy: 0.6613 - val_loss: 1.3918 - val_accuracy: 0.6943 - 33s/epoch - 5ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss did not improve from 1.39179\n",
            "6525/6525 - 33s - loss: 1.9548 - accuracy: 0.6737 - val_loss: 1.4602 - val_accuracy: 0.6903 - 33s/epoch - 5ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss did not improve from 1.39179\n",
            "6525/6525 - 33s - loss: 1.8837 - accuracy: 0.6830 - val_loss: 1.4357 - val_accuracy: 0.7039 - 33s/epoch - 5ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss improved from 1.39179 to 1.19526, saving model to /content/best_model.h5\n",
            "6525/6525 - 33s - loss: 1.8113 - accuracy: 0.6974 - val_loss: 1.1953 - val_accuracy: 0.7382 - 33s/epoch - 5ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.8247 - accuracy: 0.6986 - val_loss: 1.2655 - val_accuracy: 0.7491 - 33s/epoch - 5ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss did not improve from 1.19526\n",
            "6525/6525 - 34s - loss: 1.7413 - accuracy: 0.7110 - val_loss: 1.5201 - val_accuracy: 0.7363 - 34s/epoch - 5ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.7508 - accuracy: 0.7163 - val_loss: 2.0422 - val_accuracy: 0.7121 - 33s/epoch - 5ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.7432 - accuracy: 0.7190 - val_loss: 2.7007 - val_accuracy: 0.6681 - 33s/epoch - 5ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.7072 - accuracy: 0.7263 - val_loss: 1.4836 - val_accuracy: 0.7361 - 33s/epoch - 5ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.6564 - accuracy: 0.7298 - val_loss: 1.4557 - val_accuracy: 0.7482 - 33s/epoch - 5ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.6666 - accuracy: 0.7374 - val_loss: 1.2607 - val_accuracy: 0.7763 - 33s/epoch - 5ms/step\n",
            "Epoch 19/30\n",
            "\n",
            "Epoch 19: val_loss did not improve from 1.19526\n",
            "6525/6525 - 32s - loss: 1.6444 - accuracy: 0.7417 - val_loss: 3.5298 - val_accuracy: 0.6167 - 32s/epoch - 5ms/step\n",
            "Epoch 20/30\n",
            "\n",
            "Epoch 20: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.5994 - accuracy: 0.7482 - val_loss: 1.3522 - val_accuracy: 0.7494 - 33s/epoch - 5ms/step\n",
            "Epoch 21/30\n",
            "\n",
            "Epoch 21: val_loss did not improve from 1.19526\n",
            "6525/6525 - 33s - loss: 1.5811 - accuracy: 0.7505 - val_loss: 1.8049 - val_accuracy: 0.7241 - 33s/epoch - 5ms/step\n",
            "Epoch 21: early stopping\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fe4103f1e90>"
            ]
          },
          "metadata": {},
          "execution_count": 14
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg_fe1 = [2.9196, 2.0490, 2.1687, 2.5096, 4.8075, 1.8934, 1.5038, 1.3918, 1.4602, 1.4357, 1.1953, 1.2655, 1.5201, 2.0422, 2.7007, 1.4836, 1.4557, 1.2607, 3.5298, 1.3522, 1.8049]"
      ],
      "metadata": {
        "id": "QPz-eU5Ko2_S"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "o_hr0dmSOmHu",
        "outputId": "d69f3f7a-b847-4fd0-f859-3d7d03e65268"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 8s 13ms/step - loss: 1.8354 - accuracy: 0.7186\n",
            "Accuracy for test images: 71.856 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg = Model(inputs=base_model.input, \n",
        "            outputs=base_model.get_layer('block3_pool').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(vgg)\n",
        "feature_extraction.add(GlobalAveragePooling2D())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "rVsXSMdUO0mf",
        "outputId": "9616ec7b-6d89-4d26-d763-215f2d02571e"
      },
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model (Functional)          (None, 9, 9, 256)         1735488   \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 256)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                7453      \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 1,742,941\n",
            "Trainable params: 7,453\n",
            "Non-trainable params: 1,735,488\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=8,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "G9ugBretO22S",
        "outputId": "019fe2de-5e82-4e5b-c743-9bddb6acdd2d"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss did not improve from 1.19526\n",
            "6525/6525 - 56s - loss: 5.1679 - accuracy: 0.6162 - val_loss: 2.9293 - val_accuracy: 0.7527 - 56s/epoch - 9ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss did not improve from 1.19526\n",
            "6525/6525 - 53s - loss: 3.5373 - accuracy: 0.7791 - val_loss: 4.6177 - val_accuracy: 0.7864 - 53s/epoch - 8ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss did not improve from 1.19526\n",
            "6525/6525 - 53s - loss: 2.9825 - accuracy: 0.8301 - val_loss: 4.0248 - val_accuracy: 0.7916 - 53s/epoch - 8ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss did not improve from 1.19526\n",
            "6525/6525 - 53s - loss: 2.4976 - accuracy: 0.8621 - val_loss: 3.0464 - val_accuracy: 0.9087 - 53s/epoch - 8ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss did not improve from 1.19526\n",
            "6525/6525 - 53s - loss: 2.2774 - accuracy: 0.8777 - val_loss: 3.0521 - val_accuracy: 0.8707 - 53s/epoch - 8ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss improved from 1.19526 to 0.76607, saving model to /content/best_model.h5\n",
            "6525/6525 - 53s - loss: 2.1540 - accuracy: 0.8919 - val_loss: 0.7661 - val_accuracy: 0.9384 - 53s/epoch - 8ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss did not improve from 0.76607\n",
            "6525/6525 - 53s - loss: 1.8617 - accuracy: 0.9050 - val_loss: 1.9763 - val_accuracy: 0.8922 - 53s/epoch - 8ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.76607\n",
            "6525/6525 - 54s - loss: 1.6972 - accuracy: 0.9126 - val_loss: 1.5442 - val_accuracy: 0.9095 - 54s/epoch - 8ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss did not improve from 0.76607\n",
            "6525/6525 - 53s - loss: 1.6785 - accuracy: 0.9191 - val_loss: 1.7313 - val_accuracy: 0.9184 - 53s/epoch - 8ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss did not improve from 0.76607\n",
            "6525/6525 - 53s - loss: 1.5906 - accuracy: 0.9232 - val_loss: 1.0591 - val_accuracy: 0.9391 - 53s/epoch - 8ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss did not improve from 0.76607\n",
            "6525/6525 - 53s - loss: 1.5921 - accuracy: 0.9277 - val_loss: 1.3125 - val_accuracy: 0.9299 - 53s/epoch - 8ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss improved from 0.76607 to 0.68529, saving model to /content/best_model.h5\n",
            "6525/6525 - 53s - loss: 1.4213 - accuracy: 0.9346 - val_loss: 0.6853 - val_accuracy: 0.9566 - 53s/epoch - 8ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss did not improve from 0.68529\n",
            "6525/6525 - 53s - loss: 1.4315 - accuracy: 0.9356 - val_loss: 1.6308 - val_accuracy: 0.9310 - 53s/epoch - 8ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.68529\n",
            "6525/6525 - 53s - loss: 1.4043 - accuracy: 0.9393 - val_loss: 2.9953 - val_accuracy: 0.8978 - 53s/epoch - 8ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss did not improve from 0.68529\n",
            "6525/6525 - 53s - loss: 1.3807 - accuracy: 0.9411 - val_loss: 0.7137 - val_accuracy: 0.9579 - 53s/epoch - 8ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss improved from 0.68529 to 0.39439, saving model to /content/best_model.h5\n",
            "6525/6525 - 53s - loss: 1.2586 - accuracy: 0.9463 - val_loss: 0.3944 - val_accuracy: 0.9776 - 53s/epoch - 8ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.39439\n",
            "6525/6525 - 53s - loss: 1.2555 - accuracy: 0.9466 - val_loss: 1.5541 - val_accuracy: 0.9330 - 53s/epoch - 8ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 0.39439\n",
            "6525/6525 - 53s - loss: 1.1834 - accuracy: 0.9488 - val_loss: 1.9430 - val_accuracy: 0.9243 - 53s/epoch - 8ms/step\n",
            "Epoch 19/30\n",
            "\n",
            "Epoch 19: val_loss did not improve from 0.39439\n",
            "6525/6525 - 54s - loss: 1.1329 - accuracy: 0.9511 - val_loss: 0.7042 - val_accuracy: 0.9629 - 54s/epoch - 8ms/step\n",
            "Epoch 20/30\n",
            "\n",
            "Epoch 20: val_loss did not improve from 0.39439\n",
            "6525/6525 - 53s - loss: 1.1658 - accuracy: 0.9528 - val_loss: 1.0657 - val_accuracy: 0.9516 - 53s/epoch - 8ms/step\n",
            "Epoch 21/30\n",
            "\n",
            "Epoch 21: val_loss did not improve from 0.39439\n",
            "6525/6525 - 53s - loss: 1.0984 - accuracy: 0.9545 - val_loss: 0.4286 - val_accuracy: 0.9771 - 53s/epoch - 8ms/step\n",
            "Epoch 22/30\n",
            "\n",
            "Epoch 22: val_loss did not improve from 0.39439\n",
            "6525/6525 - 60s - loss: 1.0741 - accuracy: 0.9572 - val_loss: 1.3303 - val_accuracy: 0.9488 - 60s/epoch - 9ms/step\n",
            "Epoch 23/30\n",
            "\n",
            "Epoch 23: val_loss improved from 0.39439 to 0.20586, saving model to /content/best_model.h5\n",
            "6525/6525 - 53s - loss: 1.1217 - accuracy: 0.9561 - val_loss: 0.2059 - val_accuracy: 0.9863 - 53s/epoch - 8ms/step\n",
            "Epoch 24/30\n",
            "\n",
            "Epoch 24: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 0.9864 - accuracy: 0.9600 - val_loss: 1.4783 - val_accuracy: 0.9617 - 53s/epoch - 8ms/step\n",
            "Epoch 25/30\n",
            "\n",
            "Epoch 25: val_loss did not improve from 0.20586\n",
            "6525/6525 - 54s - loss: 1.0129 - accuracy: 0.9582 - val_loss: 0.6255 - val_accuracy: 0.9714 - 54s/epoch - 8ms/step\n",
            "Epoch 26/30\n",
            "\n",
            "Epoch 26: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 1.0480 - accuracy: 0.9609 - val_loss: 2.7225 - val_accuracy: 0.9335 - 53s/epoch - 8ms/step\n",
            "Epoch 27/30\n",
            "\n",
            "Epoch 27: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 0.9803 - accuracy: 0.9619 - val_loss: 0.5283 - val_accuracy: 0.9747 - 53s/epoch - 8ms/step\n",
            "Epoch 28/30\n",
            "\n",
            "Epoch 28: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 0.9453 - accuracy: 0.9630 - val_loss: 0.6670 - val_accuracy: 0.9716 - 53s/epoch - 8ms/step\n",
            "Epoch 29/30\n",
            "\n",
            "Epoch 29: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 0.9306 - accuracy: 0.9645 - val_loss: 3.9524 - val_accuracy: 0.8941 - 53s/epoch - 8ms/step\n",
            "Epoch 30/30\n",
            "\n",
            "Epoch 30: val_loss did not improve from 0.20586\n",
            "6525/6525 - 53s - loss: 0.8818 - accuracy: 0.9664 - val_loss: 1.1458 - val_accuracy: 0.9530 - 53s/epoch - 8ms/step\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7fe33aba13d0>"
            ]
          },
          "metadata": {},
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg_fe2 = [2.9293, 4.6177, 4.0248, 3.0464, 3.0521, 0.7661, 1.9763, 1.5442, 1.7313, 1.0591, 1.3125, 0.6853, 1.6308, 2.9953, 0.7137, 0.3944, 1.5541, 1.9430, 0.7042, 1.0657, 0.4286, 1.3303, 0.2059, 1.4783, 0.6255, 2.7225, 0.5283, 0.6670, 3.9524, 1.1458]"
      ],
      "metadata": {
        "id": "KemxEB70qWGt"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "QcV4HPx8SYLp",
        "outputId": "7c4a2320-1a45-4ef1-ff77-608e958cc679"
      },
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 6s 10ms/step - loss: 1.3281 - accuracy: 0.9398\n",
            "Accuracy for test images: 93.983 %\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg = Model(inputs=base_model.input, \n",
        "            outputs=base_model.get_layer('block4_pool').output)\n",
        "\n",
        "feature_extraction = Sequential()\n",
        "feature_extraction.add(vgg)\n",
        "feature_extraction.add(GlobalAveragePooling2D())\n",
        "feature_extraction.add(Dense(nclass, activation='softmax'))\n",
        "\n",
        "feature_extraction.compile(loss='categorical_crossentropy', \n",
        "                           optimizer=keras.optimizers.Adam(learning_rate=0.01), \n",
        "                           metrics=['accuracy'])\n",
        "\n",
        "feature_extraction.summary()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Uj5wkvflYelH",
        "outputId": "510c8f97-a1ae-4a6c-a991-333032ab8df8"
      },
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " model (Functional)          (None, 4, 4, 512)         7635264   \n",
            "                                                                 \n",
            " global_average_pooling2d (G  (None, 512)              0         \n",
            " lobalAveragePooling2D)                                          \n",
            "                                                                 \n",
            " dense (Dense)               (None, 29)                14877     \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 7,650,141\n",
            "Trainable params: 14,877\n",
            "Non-trainable params: 7,635,264\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "feature_extraction.fit(x=X_train,\n",
        "                      y=y_cat_train,\n",
        "                      validation_data=(X_val, y_cat_val),\n",
        "                      batch_size=8,\n",
        "                      epochs=30, \n",
        "                      shuffle=True,\n",
        "                      verbose=2,\n",
        "                      callbacks=callbacks_list)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "0UM610PCYieE",
        "outputId": "d3942493-33cb-4f09-dc03-b546a81d5fea"
      },
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/30\n",
            "\n",
            "Epoch 1: val_loss improved from inf to 0.43401, saving model to /content/best_model.h5\n",
            "6525/6525 - 91s - loss: 0.7466 - accuracy: 0.8582 - val_loss: 0.4340 - val_accuracy: 0.9296 - 91s/epoch - 14ms/step\n",
            "Epoch 2/30\n",
            "\n",
            "Epoch 2: val_loss improved from 0.43401 to 0.30335, saving model to /content/best_model.h5\n",
            "6525/6525 - 87s - loss: 0.4147 - accuracy: 0.9395 - val_loss: 0.3034 - val_accuracy: 0.9526 - 87s/epoch - 13ms/step\n",
            "Epoch 3/30\n",
            "\n",
            "Epoch 3: val_loss did not improve from 0.30335\n",
            "6525/6525 - 88s - loss: 0.3322 - accuracy: 0.9583 - val_loss: 0.4125 - val_accuracy: 0.9502 - 88s/epoch - 13ms/step\n",
            "Epoch 4/30\n",
            "\n",
            "Epoch 4: val_loss did not improve from 0.30335\n",
            "6525/6525 - 88s - loss: 0.2634 - accuracy: 0.9681 - val_loss: 0.5565 - val_accuracy: 0.9434 - 88s/epoch - 13ms/step\n",
            "Epoch 5/30\n",
            "\n",
            "Epoch 5: val_loss improved from 0.30335 to 0.15415, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.2245 - accuracy: 0.9739 - val_loss: 0.1541 - val_accuracy: 0.9787 - 88s/epoch - 13ms/step\n",
            "Epoch 6/30\n",
            "\n",
            "Epoch 6: val_loss did not improve from 0.15415\n",
            "6525/6525 - 88s - loss: 0.2210 - accuracy: 0.9767 - val_loss: 0.2354 - val_accuracy: 0.9722 - 88s/epoch - 13ms/step\n",
            "Epoch 7/30\n",
            "\n",
            "Epoch 7: val_loss improved from 0.15415 to 0.14905, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.1818 - accuracy: 0.9804 - val_loss: 0.1491 - val_accuracy: 0.9799 - 88s/epoch - 14ms/step\n",
            "Epoch 8/30\n",
            "\n",
            "Epoch 8: val_loss did not improve from 0.14905\n",
            "6525/6525 - 88s - loss: 0.1706 - accuracy: 0.9825 - val_loss: 0.1841 - val_accuracy: 0.9813 - 88s/epoch - 14ms/step\n",
            "Epoch 9/30\n",
            "\n",
            "Epoch 9: val_loss improved from 0.14905 to 0.11025, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.1605 - accuracy: 0.9836 - val_loss: 0.1103 - val_accuracy: 0.9872 - 88s/epoch - 14ms/step\n",
            "Epoch 10/30\n",
            "\n",
            "Epoch 10: val_loss did not improve from 0.11025\n",
            "6525/6525 - 88s - loss: 0.1382 - accuracy: 0.9861 - val_loss: 0.2234 - val_accuracy: 0.9776 - 88s/epoch - 14ms/step\n",
            "Epoch 11/30\n",
            "\n",
            "Epoch 11: val_loss improved from 0.11025 to 0.06603, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.1371 - accuracy: 0.9873 - val_loss: 0.0660 - val_accuracy: 0.9921 - 88s/epoch - 14ms/step\n",
            "Epoch 12/30\n",
            "\n",
            "Epoch 12: val_loss did not improve from 0.06603\n",
            "6525/6525 - 88s - loss: 0.1293 - accuracy: 0.9882 - val_loss: 0.1042 - val_accuracy: 0.9902 - 88s/epoch - 14ms/step\n",
            "Epoch 13/30\n",
            "\n",
            "Epoch 13: val_loss improved from 0.06603 to 0.02037, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.1058 - accuracy: 0.9899 - val_loss: 0.0204 - val_accuracy: 0.9967 - 88s/epoch - 14ms/step\n",
            "Epoch 14/30\n",
            "\n",
            "Epoch 14: val_loss did not improve from 0.02037\n",
            "6525/6525 - 89s - loss: 0.1242 - accuracy: 0.9893 - val_loss: 0.0494 - val_accuracy: 0.9932 - 89s/epoch - 14ms/step\n",
            "Epoch 15/30\n",
            "\n",
            "Epoch 15: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.1217 - accuracy: 0.9893 - val_loss: 0.0722 - val_accuracy: 0.9913 - 88s/epoch - 14ms/step\n",
            "Epoch 16/30\n",
            "\n",
            "Epoch 16: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.0856 - accuracy: 0.9922 - val_loss: 0.1363 - val_accuracy: 0.9879 - 88s/epoch - 14ms/step\n",
            "Epoch 17/30\n",
            "\n",
            "Epoch 17: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.1080 - accuracy: 0.9912 - val_loss: 0.0525 - val_accuracy: 0.9937 - 88s/epoch - 13ms/step\n",
            "Epoch 18/30\n",
            "\n",
            "Epoch 18: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.0928 - accuracy: 0.9916 - val_loss: 0.1675 - val_accuracy: 0.9860 - 88s/epoch - 13ms/step\n",
            "Epoch 19/30\n",
            "\n",
            "Epoch 19: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.1242 - accuracy: 0.9905 - val_loss: 0.0752 - val_accuracy: 0.9926 - 88s/epoch - 13ms/step\n",
            "Epoch 20/30\n",
            "\n",
            "Epoch 20: val_loss did not improve from 0.02037\n",
            "6525/6525 - 88s - loss: 0.0957 - accuracy: 0.9924 - val_loss: 0.0529 - val_accuracy: 0.9943 - 88s/epoch - 13ms/step\n",
            "Epoch 21/30\n",
            "\n",
            "Epoch 21: val_loss improved from 0.02037 to 0.01361, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.0734 - accuracy: 0.9941 - val_loss: 0.0136 - val_accuracy: 0.9980 - 88s/epoch - 13ms/step\n",
            "Epoch 22/30\n",
            "\n",
            "Epoch 22: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0886 - accuracy: 0.9929 - val_loss: 0.0874 - val_accuracy: 0.9921 - 88s/epoch - 13ms/step\n",
            "Epoch 23/30\n",
            "\n",
            "Epoch 23: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0828 - accuracy: 0.9924 - val_loss: 0.1932 - val_accuracy: 0.9883 - 88s/epoch - 13ms/step\n",
            "Epoch 24/30\n",
            "\n",
            "Epoch 24: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0750 - accuracy: 0.9939 - val_loss: 0.0842 - val_accuracy: 0.9932 - 88s/epoch - 13ms/step\n",
            "Epoch 25/30\n",
            "\n",
            "Epoch 25: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0752 - accuracy: 0.9937 - val_loss: 0.0949 - val_accuracy: 0.9929 - 88s/epoch - 13ms/step\n",
            "Epoch 26/30\n",
            "\n",
            "Epoch 26: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0806 - accuracy: 0.9935 - val_loss: 0.2263 - val_accuracy: 0.9827 - 88s/epoch - 13ms/step\n",
            "Epoch 27/30\n",
            "\n",
            "Epoch 27: val_loss did not improve from 0.01361\n",
            "6525/6525 - 88s - loss: 0.0805 - accuracy: 0.9935 - val_loss: 0.0866 - val_accuracy: 0.9916 - 88s/epoch - 14ms/step\n",
            "Epoch 28/30\n",
            "\n",
            "Epoch 28: val_loss improved from 0.01361 to 0.01032, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.0704 - accuracy: 0.9944 - val_loss: 0.0103 - val_accuracy: 0.9987 - 88s/epoch - 14ms/step\n",
            "Epoch 29/30\n",
            "\n",
            "Epoch 29: val_loss improved from 0.01032 to 0.00570, saving model to /content/best_model.h5\n",
            "6525/6525 - 88s - loss: 0.0674 - accuracy: 0.9948 - val_loss: 0.0057 - val_accuracy: 0.9990 - 88s/epoch - 14ms/step\n",
            "Epoch 30/30\n",
            "\n",
            "Epoch 30: val_loss did not improve from 0.00570\n",
            "6525/6525 - 88s - loss: 0.0826 - accuracy: 0.9944 - val_loss: 0.0377 - val_accuracy: 0.9963 - 88s/epoch - 14ms/step\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<keras.callbacks.History at 0x7f41e01ede90>"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "vgg_fe3 = [0.4340, 0.3034, 0.4125, 0.5565, 0.1541, 0.2354, 0.1491, 0.1841, 0.1103, 0.2234, 0.0660, 0.1042, 0.0204, 0.0494, 0.0722, 0.1363, 0.0525, 0.1675, 0.0752, 0.0529, 0.0136, 0.0874, 0.1932, 0.0842, 0.0949, 0.2263, 0.0866, 0.0103, 0.0057, 0.0377]"
      ],
      "metadata": {
        "id": "7ObuGW8Y3VqQ"
      },
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "score = feature_extraction.evaluate(x = X_test, y = y_cat_test, verbose = 1)\n",
        "print('Accuracy for test images:', round(score[1]*100, 3), '%')"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "AwVj1bMOZszX",
        "outputId": "823b46ef-65bb-4a66-93b1-f8447594825d"
      },
      "execution_count": 14,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "544/544 [==============================] - 14s 25ms/step - loss: 0.0282 - accuracy: 0.9964\n",
            "Accuracy for test images: 99.644 %\n"
          ]
        }
      ]
    }
  ]
}