CNN_for_image_classification.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "id": "TAjf09VjcJ5w"
   },
   "source": [
    "# IMAGE CLASSIFIER"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "NH8XvZj8XlCh",
    "outputId": "cd8650b9-edad-4934-e7e7-619feb446f05"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "`%tensorflow_version` only switches the major version: 1.x or 2.x.\n",
      "You set: `1.0`. This will be interpreted as: `1.x`.\n",
      "\n",
      "\n",
      "TensorFlow 1.x selected.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "#import all modules and packages\n",
    "%tensorflow_version 1.0\n",
    "import math\n",
    "import random\n",
    "import numpy as np\n",
    "import h5py\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy\n",
    "from PIL import Image\n",
    "from scipy import ndimage, misc\n",
    "import tensorflow as tf\n",
    "from tensorflow.python.framework import ops\n",
    "from keras.datasets import cifar10\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/"
    },
    "id": "0auysqfiXlOr",
    "outputId": "8b9c79d8-5689-4ea9-af1d-3e8942c64fba"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://www.cs.toronto.edu/~kriz/cifar-10-python.tar.gz\n",
      "170500096/170498071 [==============================] - 6s 0us/step\n"
     ]
    }
   ],
   "source": [
    "#load cifar10 dataset\n",
    "(X_train_orig,Y_train_orig),(X_test_orig,Y_test_orig)=cifar10.load_data()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 699
    },
    "id": "OJN1WZqXXmEM",
    "outputId": "845fa6c1-a51d-46c2-8463-9370a466c48e"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAMIAAADACAYAAABWKbw6AAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAaxElEQVR4nO2dWWxc53XH/2fmzkZyyOFOipIla41tWFYSxXCdFM3SFI4fYgco2qRAkQIGkocaSNA8RGiBNi3y4AJZXlqkdVAnLpDGTbMagdvUcR0YaYLEaxxZciRZi0WKokiRI86+nj7MKCX5P6S4aThMzg8gSB7eud937/DMvf97lk9UFY7z205oqyfgOO2AO4LjwB3BcQC4IzgOAHcExwHgjuA4ADboCCJyn4j8SkTOiMixzZqU47QaWW8cQUTCAE4BeD+AcQDPA/iIqp5Y7jWRSERj8fgiW61Wo+1C4DmFhfcXDdiPI4YtCIfJJsI7FDE+F4xxq1Wes3UWw9a4xvmua533V+ftJGRMxqBe5/lZczFfa8xFjJNg2ULGGOEQn1Pr3NeN86LWyTew/oet9+Pi5MyMqg5a+whWNZLN3QDOqOpZABCRJwA8AGBZR4jF4zjytrcvsqXTs7xdiN+Mvigf2i39HWQb7Osk20Cqi2zRcIRsQSxBNoT5FM3OpclWrvL8elM9ZAvVKmQrlUpkKxaLZIsn4mSrgf/p84Us2XpS3WSD8mvLpTLZwuBzZTlWsovPc2cnvx+RCB9HwRhXrQ+mEL8f1pyryk708Gf/+QLvsLnb5f6wCsYAXFzw+3jTtggR+ZiIvCAiL1Qr/E/gOO3ATRfLqvqoqh5V1aNBhD9ZHKcd2IgjTADYteD3nU2b42w7NqIRngdwQERuRcMBPgzgT1Z6QbFYxGsnXltkS8/M0HZ9fAsJ6WfjQC3J2yWGyJarsw7J1gyBJVGy5Yt8/5kv8D19pca6ZsZQ+PGAx61W+bVh4144FosZ88vx/uo8Zyn2ky1k6OeKoVcSAZ/7rHFfPlurkq2jgzWChPjOQAzNBkNo54t8e23dcocDPlcrsW5HUNWqiDwM4AcAwgAeU9XXbvAyx2lLNnJFgKo+BeCpTZqL42wZHll2HLgjOA6ADd4arZUQgESwREAamma3IYz3DHNwamiwj2wJS5wZkcxCiQNWxQoLRTVeG00YgTcjoKZ13l9PHwcBqxV+bTTCYxhBeISjfAJLZT62SpWPo8N4bdDJ48aN7arCIj1kRKWrRnTYyhLo6uTzks3lyVapsjC2Au6Z+WtsXAG/IjgO3BEcB4A7guMAcEdwHAAtFssiirgsjj4mkzyFg2O9ZOtPcBg0UmdRmJ3liGetzv5eyHMUNMSBZXQbmauBIR7T1zK8nXF2+5IsCjPzLDzLRsS4YERVrVTlLiPjs1IukC1U4wlGjOh1zciYDQzFWyrxdtEIn9RQnc99KTtHNhjR/5gRDa/WWaRfy/GDipXwK4LjwB3BcQC4IzgOAHcExwHQYrEciKA3tnjIhCHOeozo5mA3p+nWjNpcI/iKcGAoLCPFt1Q3RKGheAMjglorsRjVMI9x5QqXedYqPOtMnqOq+Ro/COhKGCWYJaNmGTznkBi14TGjjDLHDyU6IjxuYNQOF4009kKFxXLdqDJOZ3ncdJ7fo6zx4KNYWdtnvF8RHAfuCI4DwB3BcQBsUCOIyHkAGTRuzauqenQzJuU4rWYzxPJ7VJULj63BwoLB1GIxloywkI3H2RYKs5hKGOnQFaP5Vt2IvqqyiLN6E9XKLM7qakR4DSGrAUdVM2WOGNdqfLx5owa6atgyOZ7LxCyPETF6RXVn+bxULvNbWbjGwv2Wgf1kGxraSTZJcjp0ae4q2bJZnvO1DIvlmWv8UOL8RR6jZvSjWgm/NXIcbNwRFMB/i8iLIvIxa4OFDb7KRrcGx2kHNnpr9C5VnRCRIQBPi8jrqvrcwg1U9VEAjwJAT4fRt9Fx2oANXRFUdaL5/QqA76DRD9Vxth3rviKISCeAkKpmmj//AYC/W+k1kSCMHYOLU4S7oxwV7OpgkSmGQLV6HosR9S0VWOyFDAHdn+S66M5OjrTOX2NB2dPNkdaMkTZ9YYJfmy2xWI4ad5FjHUaUO2KIx6scvS6pkcZuRJZ7urlp2r2388PA+UmjI3je2N8AZwSU8nwc2Sx/JseMFqG7Rnh+Q0PDZJuaZ6F9/hdvku06G7k1GgbwnWZhfADg31T1vzawP8fZMjbS6e4sgLs2cS6Os2X441PHgTuC4wBodRp2WNCXXBwNDsos7GIRnlZHjGt9SwUWoxWjHjaV4hpoa7mhco0/FyoVIwXZWBnm0jTXyL5xgSOe0xmen5FFjN1GjfaDv3uEbDtHeS7ffPEs2X565jLZrK7ZQYjPSyY9TbZ8lo83mTQ6WteMjuBx3i5qZBN0CG9XNTpu37JrB89llmvI/2cFsexXBMeBO4LjAHBHcBwA7giOA6DVYjkIMNS3eAmjwiyL0ZAYkUejVrVQZuEUiJHSbNQEW58AhQqLx1QvR4zLRuOps+OXyDY7b0RfjdTssFHb3B3n1w4FLADjsyxaD3SPkG2yj8eYSl8hWynP5+DlU6fIFjISKCudRv10D0d9rSVie3r4YUjSWGu6aKTFa3mebHsGucnZSvgVwXHgjuA4ANwRHAeAO4LjAGi5WI6gd2Bwka23i+uOQ8Y6vOl57pZcyWX5tcb6SnWjuZUa0euuLk65roBtJ8+yeMyVuOY2HufmZfEoj5swlk3qDfODgBfPTJGtWub9lXpYLA/28nEIWNxWqvzwIm900s4ZKdflKs9ZjAcQRgY8Isb6T2osBB0xGq5VjbWh1XigsRJ+RXAcuCM4DgB3BMcBsApHEJHHROSKiBxfYOsTkadF5HTzO6d3Os42YjVi+asA/gHAvy6wHQPwjKo+IiLHmr9/+sa7EmCJEBajLtUiZqTudoCjh4Hh2yGj83XFENCxBNcsz1zmaG5+hoX73j4Wo8ZSzogbwvjQvjGyhYwXV8N8DuaNhwhBmNO/k1E+V/29+8i278AtZDv35vNke/3UBNmigSFalR9oVKv8bxcyIu6RKB9v3VgmymrgJrLJ3bCb7Vlml5gfAPB48+fHATy4plEdp81Yr0YYVtXJ5s+X0SjkN1nY4CuTNz4iHacN2LBY1kap17IPbVX1UVU9qqpHkx18++A47cB6HWFKREYBoPmd0xgdZxux3sjykwA+CuCR5vfvreZFdVVaK1gqHLUEOEKZy3GqbdlYHqga4qtONs+Cd96wje3i06FV3m73AIuzfTtY2OWLvN3YQe6AE1W+ZZy7xunGiVQ/2XCVo6+7RkbJls5x5HvvWw6QrbuXxXx37208v2k+L3PXWKRHDJEeUo64V4xlwAxdjJqx7JQRlDZr0ldiNY9Pvw7gpwAOici4iDyEhgO8X0ROA/j95u+Os2254RVBVT+yzJ/et8lzcZwtwyPLjgN3BMcB0OI0bIWiJotFkRoNmyyhk4hzunZXkoXdpWkW3+fGuUFVEOExolNcd1yc4tceGGJh/L53s/B8Y2JpHBJIjg2SbaCf06avTHPKdSplCM+60SzLSF++Ms2R4CDOzdWm05Nkm5jk6HAkwuc+1c3qtlDg86wBf/6KoXjrhoAOiRFFNjIH1piF7VcExwHcERwHgDuC4wBwR3AcAC0Wy+FwCKnU4u7N1YDFcjbLkVY1mnRdy3Ak88KbLDKzWRZ7iTh/Bkye4+j1cJzTg8fGdpMtteNWskUyRmjUSCffeRcvPRe/zOI2UWXhXgOfq1yObaMdLNLLxrrN0sndtXd2Gt2mUyzwM1e54/aVKV5TuWJ0uS6WOYUbRmfuzhhnDpQLhpg3UrhXwq8IjgN3BMcB4I7gOADcERwHQIvFcr1WRSa9WDwFZU7njVj1phwsRRA2Ol9nWUD3JjkimzLWTy7MsVge2sGpz2OHf49sx8e5kdWpM2y7d7SPbOk0bze8j9O1Q+D1osslFtApY63p+SssWhNGZ+nRPmN+NU6bjhzmfg0FIyr9v089SbbxizznsCluOYpsBKpRserUK9a63MvjVwTHgTuC4wBwR3AcAOtv8PUZEZkQkVeaX/ff3Gk6zs1lvQ2+AOCLqvq5tQ4YXqJ/akZUUA2RFDLqmGvGMlFzhkaanzdSgUssUEd7WFS/4z3vIdvOQ/eQ7dtfeYxsI0aUNmx0lp44+wa/du/tZIv37ydbpxoNyGa5l0KizuK2XGDxPZNhW2qQo+b9I3vIVshyd+2QsZpULcqRbysNu2J00pYqZxiIss1qIrYS623w5Ti/UWxEIzwsIq82b52W7X26sMFX1lioznHagfU6wpcA7ANwBMAkgM8vt+HCBl9dHZzA5jjtwLocQVWnVLWmqnUAXwbA6ZOOs41YV2RZREYX9D79EIDjK23/69cBkCW6tWZEAK0aVKPMFVowXmtkPvf1c33tSAeL77cdPUi22+5lYTx3hQV+rMoR7b07d5KtbkxwZIhTpKtFnl/eiEBbyzVVCvy21sDC/Y2JcbL98vgLZLv3Hh63f4Qj7vMZFulGaTMG9vBDibpVd1w2RLDxkOPaNNdelzLGwCtwQ0doNvh6N4ABERkH8DcA3i0iR9DoeXoewMfXNKrjtBnrbfD1LzdhLo6zZXhk2XHgjuA4AFrd4EuB+pLIYKHE4jFqRGSDgNN0wyEWTvtHOKQRT7C/79m9i2x3vYujyKOHDpPtlZ9+hWy37OJxR+64k2zRQV6uKejgJavyRRbkhXmOIk9duki2uSkWwbUKR4wTSU5FHxjg83zx0stkGx7l5a6qeSNLoMC1yJLj5a5qyhF3XfpkBUAiZjQ0GzGW1IoZLbJXwK8IjgN3BMcB4I7gOADcERwHQIvFsoggEl485JyR9lszllxKdHA37LDRAGrIiCJfnOTI47633Ue2nXeyDWARXMnwMkw9SRa8gwePkC0XcE3way/zOsalAo8xP8/HMTPxJtnCNX6IEI/zWz12Kwvewwc51bsa5khwJJxiW5Qj/UGRU67zF7h52dKHKABQNT6ms0adekc/z2/YqDVfCb8iOA7cERwHgDuC4wBwR3AcAK2OLNfrKBUWi6eOGE9B4iyIIiFjiSlj2alEF7/2g3/8QbLd+wFeFLR7YJhsU2dPki1szCVtdOaePv8rsl3KsCj80Xe/S7auhNExusSR25FhFundRkOzc+McgS4bx9G3Yw/ZDt75drLBaPo1m+aItrXW9FyBxxXl/4NigbMOssayYmp0T7+NtfyK+BXBceCO4DgA3BEcB8DqGnztEpFnReSEiLwmIp9o2vtE5GkROd38vmwnC8dpd1YjlqsAPqWqL4lIEsCLIvI0gD8D8IyqPiIixwAcA/DplXakUNR1SdTTWEtXqiySqmrUJxtpuvEYd5Q68nYWe7EIi9ETr3C68dwlbr5VKrE4y8xx66eLZ06QLascIY/UeH9dAYv+7jiL4MFeFsuTU7yEU9WoDc9nWHxfPMeRauA1smSznBIeD/j9qMaGyHa1yu9RIsEp4R1JPleJgEV6Js9dzKt1FuQrsZoGX5Oq+lLz5wyAkwDGADwA4PHmZo8DeHBNIztOG7EmjSAiewC8FcDPAAwv6GRxGQA/e8TiBl+5gjf4ctqTVTuCiHQB+BaAT6rqomuRqioaHS2IhQ2+OhPe4MtpT1blCCISQcMJvqaq326ap0RktPn3UQDc1MZxtgmr6WskaLRvOamqX1jwpycBfBTAI83v37vxcApgsRCuV/l2KTC6QtWMNN2y0SF7uIcfXv3gye+TrW+YBeDQKNcxl/McMY5EWLB1dbIADEIseDsNkT4yxCnDhQzX9SbCPO7V6RmyVYzGWMk4C8+ysf706Ze5wdfk66fIVqpyjTEifLw16xzsZNGPTv4/CMX4IULcEMG94GO77Q7u4A28ZNgarOap0TsB/CmAX4rIK03bX6LhAN8QkYcAXADwR6vYl+O0Jatp8PVjWKu6NeCEHcfZhnhk2XHgjuA4AFqchg0V1OuL77KiRgQ1HhgtrY2lhdSopa0bawfPzHCkNTvNtkSFI5R1Y4Hnvl4Wt6kdRkfrGje3mrjE46rx5DkU4rfG6nwdFhbfnXF+2GAE6xG2jEa0vlbmBwahOr8f83kW+OUYi+rkDj4vuQTXY2fqLKCLOf7s7u/eS7YB4wHESvgVwXHgjuA4ANwRHAeAO4LjAGi1WIYgJIujo/EYRwXViBh3JlgAdiYHyJavcDSyP8k5ToExRvnaFNnqIX5tPsIic3iYI5n1Mou9Q4d5OamfPPsMz0W58VlEWKAWsrxdd5Kj3NGA3+qwsYxV1mjIdW6SRXA6zeevJNyUbPAgf9aOpYwot/J5npvhY4sWjYcDY0ZkPs/R9ZXwK4LjwB3BcQC4IzgOAHcExwHQYrEcEiC6ZMHkfImjjGGjNrdupCDnKxy1DEc4MhqLGnXCER4jaizh1NPN212eZlGdH2MRPLSLO0tPXOG06Tve8U6yZacvke3sKU4dz2U5IhuE+bz09LCAFrBYnpzgcd+8YESWY3xeuof5gcZgnzGuIchllvfXO2d08B7ibuI7U3zuz5zgCP5K+BXBceCO4DgA3BEcB8DGGnx9RkQmROSV5tf9N3+6jnNz2EiDLwD4oqp+btWDBYLhwcW+V7l6lbYr1FjE5ThoCQ1x9DAwIqjd3Rx5jBq1w4Ucp2EnIsYpKrPthZ/8hGx7D7GoHh9nERcyUsw7jPWEw8YDg0SCRWYuy2K5UGBb1agX70rwGPe+9SDZ4kb0uhrmaLO1vnPhIovlUIYbfA11JMn21oN38HYp7iT04uQ5sq3Eako1JwFMNn/OiMj1Bl+O8xvDRhp8AcDDIvKqiDy2XO/ThQ2+5vPe4MtpTzbS4OtLAPYBOILGFePz1usWNvjq7vAGX057su4GX6o6pao1Va0D+DKAu2/eNB3n5rLuBl8iMrqg9+mHABy/0b6iUcEtuxZfFXqERdKZiyywpqY5Ylw2li/q6uJDyhlNump1bm4VNj4XZqdZzGeyLAqLFR4jrGxLdvEd5NRl7qQ9nmNBWVcW1cOD/CBA6ly3PZfmVOpYJ5+/VA8L1GiYz0vJaCKGgAV+rsSvLWeNVOo6b7d/1wjZdozw8V4c54cSV6f5f2glNtLg6yMicgSN9nXnAXx8TSM7ThuxkQZfT23+dBxna/DIsuPAHcFxALQ4DTscCLp7FwulgiFqeoe4qRY6OcV3ZopTuItGnXAQ5SiosRnqFRaAFaNJ17UCC89OIyJbzLPgLRQ5DbtsjFszbKp8XrLzRs1yN6edd3dzinmhwK+ducrH1tXF0WsJ8WeoVPmBRjTgucT4+QiiUT62Pfv3kK2Q5zGee46X6Hr11NpWKfArguPAHcFxALgjOA4AdwTHAdBisSwiCOKLh4x3c/5RXxf7Z1Bg0RpJcLr2vFHnihrvLxHn9X9rRuOuWolrgqMdPEYk4OMIh1ngl5THKFdYuasRRTYaVUPLLMiNZZsRMaK+iLLAT8+xWC4YHcZ7UtZSWXyeQ8Z5yRvN1aZmeN3mOSOCn8lxtP6HP3qd97e2wLJfERwHcEdwHADuCI4DwB3BcQC0WCzX64Ls0hTccBdt19XJai+SYKXYaYQoe3qMDs/zXK+bnefU3azRQblSNNYsjnIqcNyoga4azcuCgD97osbHUSTGkVYR3rDDSDs3Vp1CtcbCM5ow6rtTLPBnZ1nIZgzR393H5yVv1EWfPs+p7a//8iLZho3mYMM7eX4I8VwGjHTyc1eNtaGv72LZvzjObxHuCI4DdwTHAbC6Bl9xEfm5iPyi2eDrb5v2W0XkZyJyRkT+XUS8Mt/ZtqxGLJcAvFdVs80i/h+LyH8C+As0Gnw9ISL/BOAhNDpbLEu5DIxfWLLzNAve5CALu3jCiG6yzkZfHx9SNsdhxnSabXNXjeWLWNchXGchW1djfeKaUddbZ5v1aSRG06+w0bysYETNlU8fIkYdczXPtdI1IzW7ZkSl08aSVVYZ86zxoOL8GT6p6avcwa2c4x2O9HAd8227uc2WMSyeP8sp8Ne54RVBG1yvdI80vxTAewF8s2l/HMCDN9qX47Qrq23nEm4W7l8B8DSANwCkVX/92TOOZbrfLWzwdS1rJME4ThuwKkdo9i86AmAnGv2L3rLaARY2+OrpMkqTHKcNWNNTI1VNA3gWwO8ASInI9ZvWnQAmNnlujtMyVtPgaxBARVXTIpIA8H4Af4+GQ/whgCcAfBTA9260L5UAtcjitZEr0aO0XanOEdlQlYVOvIcFZWqQrzq9IVaPfXmORqZnub42PcPCuJDj01arGg/NlD9n6lUet1jgW8Zo1EjrDngumSLvr2DcgkaUI7zJEEdf6yHuCF6p8PHGOvnhQDxiNAyL8rh7kSLbnXdxXfShw3eRbc9+Xo7r7ntYuI9f4gZueP4s25qs5qnRKIDHRSSMxhXkG6r6fRE5AeAJEfksgJfR6IbnONuS1TT4ehWNDthL7Wfh/U6d3xA8suw4cEdwHACAqBERvWmDiUwDuABgAMDyYb7tgx9He3Gj49itqoPWH1rqCL8eVOQFVeXHRdsMP472YiPH4bdGjgN3BMcBsHWO8OgWjbvZ+HG0F+s+ji3RCI7TbvitkePAHcFxAGyBI4jIfSLyq2aJ57FWj79emouqXxGR4wtsfSLytIicbn43F11vJ0Rkl4g8KyInmqW3n2jat9WxbHYJcUsdoZm4948APgDgdjRW5ry9lXPYAF8FcN8S2zEAz6jqAQDPNH9vd6oAPqWqtwO4B8CfN9+D7XYs10uI70Jj0fv7ROQeNDKjv6iq+wHMoVFCfENafUW4G8AZVT2rqmU0UrgfaPEc1oWqPgdgaZHvA2iUqQLbpFxVVSdV9aXmzxkAJ9GoLtxWx7LZJcStdoQxAAtbmi1b4rlNGF6w6PplAMNbOZm1IiJ70Mgs/hm24bFspIR4KS6WNwltPIfeNs+iRaQLwLcAfFJVF1XjbJdj2UgJ8VJa7QgTAHYt+H27l3hOicgoADS/r20pxy2i2ZbnWwC+pqrfbpq35bEAm1NC3GpHeB7AgaayjwL4MIAnWzyHzeRJNMpUgVWWq241IiJoVBOeVNUvLPjTtjoWERkUkVTz5+slxCfx/yXEwFqOQ1Vb+gXgfgCn0Lif+6tWj7+BeX8dwCSAChr3ng8B6EfjCctpAD8E0LfV81zFcbwLjdueVwG80vy6f7sdC4DDaJQIvwrgOIC/btr3Avg5gDMA/gNAbDX78xQLx4GLZccB4I7gOADcERwHgDuC4wBwR3AcAO4IjgPAHcFxAAD/B4Sm4G0ijQMzAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1440x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGEAAABiCAYAAABAkr0NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAcsElEQVR4nO2d269kyZXWfysi9s7LuVV1V3W7xx67GcYz2BoYPLJA4iKNhJDQvMw8jQAJDRLSPCGBxAMW/AMDD0i8WgLJSEiABBLzMBIaRvCAkJDxXIwvuO2223Z3V5er6lzytveO2+IhYufJU9duV9Wpg12rlJWZ++TeETtWrLW+9cWKTFFVXsqLFfOiO/BSXirhSshLJVwBeamEKyAvlXAF5KUSroA8lRJE5G+IyLdE5Dsi8oVn1amfNpEfN08QEQu8Bfx14F3gy8DfUtVvPLvu/XTI01jCXwC+o6rfVVUP/Dvg159Nt366xD3FuR8Hfrjz/l3gLz7uhBs3buibb775FE0+hShkzaBKSgkAYy0CqCoCiBFAEBGQ8TTZXkIeuOhHk6985St3VfXm/cefRgkfSkTkt4HfBvjkJz/Jl7/8ZS6NKtFtH8g5M3QdIQQWZ6eoZmbzGUYgeo8xhv2DA6xz2LZFxKBSlJLr8BueThHW2u8/7PjTKOE94Gd33n+iHrsgqvpF4IsAn//85xXKoFyKXGhGib7D9x2r49sE77nje1JKdN1AO5nw85/5LPO9PeykRYwgW299roTnIU+jhC8DnxaRP0UZ/L8J/O1n0qvnIKqZGDxh6OnWS3zfM3SbooQhMJ3NScGT0hTus9TnPWV+bCWoahSRvw/8F8AC/1pVv/7MevYMZHR7IkpOgbPju6zOTvjBW9+k32yIQw8I0s7YOzikX69onEP398GY6s0ElfJ8/v+zlaeKCar6e8DvPaO+PFPRMSCgkDM5JfzQMXQbwjAQ/UCOARDEOFLwRO9JIRRLqKersH39vEziuQfmFyWCoJLRnMnR44eO0+O7rM/OaK2hmU1p9maoKsvOo2FgfXqCEbh28ybWOpASBbQO/lWMCT+WPO+g/CDyUlKKxBDou56+L8GYnMHY8okUSTHQdxuayYQYItYlcBaQYgn3dftZ3sdPrCUUhK9kzazXaxaLM+7cucvq7Ay/PENyZtY6UKXvNhjruPX9d1guFlx//Q32DiPzo+uI3SYMKMLzmEM/wUoooprxwTMMA5uuZ9N1DOsONJNjg0FJIWBypttssE1Lt15jXctk7xArZmsxz0t+QpWg9ZFJMXB6fMK9u3f50Z27nJ6ccPzBbTQlDmdTWme5ttfgnCPpXfq+5wdvv83hK6/SzPaYzvdoZ7MLWfSzlheghCdly/fd6YhSnnCe3HdeCQ2KqhJjIITAEALD4FlvenKKkJWps8wayKqI9Rjbs16cYa1j6DqsdTTT6Rijt+FhDD1jq+e90wvHP4y8IEt4GsynOyMwnn//dQquVE1kTaSciCnRD4F1P3DvbEX0nlNZ0TaWlA+ZT1uuq5BS5tb3vsfi3jFHr7zC4fVXaOcTGjOtaKkEarn/Nu7Lzj+KvGB3tDt/Hm4BNSLWV+MLHcd5m0KpPPrGrbVYZ0EEBUJM+BCJQFLLZvAgwn6IGDEMmzXOWjaLM5xzRD9grMW4tg6+bPtVeJj6tt6C6EdTwxWyhIfgwJ2P6vaOFaWwoSKmnvXAVKzjIjjreOXGDbLCbG8fd7ZkCJGuLwNvQyLdWzCftLS2Yd4GrI/I0PPeW99k8eqrHN64xvzoGnvXb2Jsg8mlL6LnbaqMXf3/xh09RB6YOjvTqbqfkgNUJdTPSJ3dj7ppMYa2bZlMp8z39pjN5xjrwBhSymTN9D4iCP3gsapEYwjWsFkusY1js1hibMP04BVEbJ3yYyotD531D8awR6vlBSnhYR16hDvSHQvQjKqScgTAOQcYRB60LEVADGIc8/kBOQu/8Iuf4eDgGu98913MvWPu3b1HSpGcEsFFPrCn7E9a2puGoJA/uM16tWb/6P9yeOMmk70jZnv7SDPBjFS36APIKd2ngCdZxeUrYetEq8iFp/sy3tHAFc1KiIGcEzEFRMAYgzEUt6QVH1244zJjjbE413J07Rp913N0dA3vAyfHp8SYSEmJZPohYBD6EIurGjzG9KxOF1jbMqw3WOtwrkGNbGe7jMG6uqScU+lznTzo49VwddwRXAzGOwezZkL03P7Rbbz3hDDgnONjb7xB205oGotIBswFqJrV1GzB0kxm/Omf/zQ3Xr3Jndv3eO/d9zm+e0JOS3KIhJi5t9iwaj3NpGFv0vJ6MmQPH3zre6yunXJ042Mc3niVNz79czg7wdc5b1UwCBbQrKz8piKyiGYl5/zYSH3pSrgwzrsMJZz7/h2sNOL8wQ+cnZ3RDz1+6GmahoPDQ1LOGGuxxtaZqBfaGa8kIsxmM2JI3Lh5Ez8EDvb3iT6yCWuyJkJSiInN4BEgTCIOS7/ucLZleXKKcY6h78kGvFFUwORyK45iBd2wJqaED56smRzzY/OcF2IJW5i/RZwF+2vOgCKm6kiEmCInp8ecnp3yR3/yRyyWS7puw6RtOVmcce3aNf7ML36G2WxG44r70XrLuV475eISJpMp9prjlz/35/mZj/8M77/3Hh+8f4tvfPXrdH2kS5lBE++fnjFvWvZlQmgSrRfyEHn7q19j/8Z1zOGU2dEBHM5RgWHTQc44zaQcOdncw0fPpu9IKRJCeOyS7guFqKPLTDkVs02h+FZTMVBWQvCcLs44OTvl3skJy+WCzXpNO5lw4/geCvjgadsW62yhsBl9c3EFMRX00zqHdZaDw31CGHjt9ZvEEJjOp4QcCHEgZaUPEYPQ+YBTwZuIGM/q7Ixs4ez4GJ8jjcuoCP16Q84JkxIpB5b9GT4Gur6s3F1RJYwQU8ia2WzWxBjo1ityzogRUk6cLZesNxu+/d3vcHJ2yre+92026w2b1ZrGOUIMvPbaa3zqzU+RSRyaA6y1iLGoKl3viTGyWW8wxtC+ch3XWK5dP2A6dfzVX/1L3Hr/Fnfv3eKD27f57rs/xPtAGAwhJD7glFUzYXJkaH1gkQaak7sMLcyuHXLzF97EtI511xFTJPiOlCOeJUkTYXRH+fHZ2wvijgpyEKlrv9HjvWe5WpFSJGkixsTJ4ox1t+buyTGL5YJ+GBiCL742F4p6tVqxWq9oJy2ucThnsdaSk7LZ9IQY2XQbrDF4PydbWxhUTcz3Zxwcztk7mjNfTxFbMu+claiZzgesCl2IZMD2kSyZxfExIUf2zq5jpy198MSc6P2GnCNRhnJfqVhAzo9nvi5ZCQqEghZq+E050PUrlsslb33nLVarNffOTvDBc7Ze4UPgdLUgpEi0ipk2NGGKAMvNGjkxfOvb3+Lo6JCbr92gbRpc05BS5vjeghiKT26cww+Fjhj64j40R7QJfOxTr8Ik8Patd4jZE3sl5szJpqezkYlrmDWO67nBxYHwve/T7s8wjZTVuaM52cAmDmQyUuGAZodqgdaPq/K5VCWoZrzv68woJhpiYPA9g+/ZbNas1itOz04ZvGfte2JO+BzJKM20xaaM+gxZCSHQdT3HJycldzBK0zY0rijh5GRJDIkUA85ZGgfWGny/AZSmMfR9RzM1TPZaprOW4APZR4jFGjyJTfAomf0IkNEOEKU7W5QBtplshXUu7sfqFnGQVQneX52Y4P3AD959B2MK/EkpEWNkuVqy2XQsNyuW3Yq7J8XcJ/tzGuPY35sgxrC3fwBZWdw+Ydj0fPD+eyxWS/haZjJt2d+bVrLOoArDkLbBHRQnmZLcZpy1XH/lAOMM2gT2rk/45Kfe4OxkxQ/e+gCfIl4zMSsfrBbMnGGSp0ydZS/NUB/40XfeQSeOzdGEaIWlJDKCi025v5jIOeO9r9b/cLlUJaSUOFucYihJlKZUzFQVYwyuabDWEVMipszMWIyzYMA4y2w+g6QMbUfyhZ5OMbJYLmh6RxgarDWFMUVI2RSImgKQESKC4qzQOEfvLY06mj3HRB1HR/uQYTJtyEnxvtAkQ8oIwhAMRh3T2IDAsO7IwdKJJ1ihsyUvsWkCGYIPpJTphx7NV8QSur7nG9/8OqH3ALSuZW++xy/92V/GWkfShrvHx3znnXfJeWA2mWMbhyfgnOP64SFkIZ56sk/0PtD3HcOwwhrYby3OGiazFusa2vk+1hpcm7EWZnPBOsN8OsM5x95+eW5nEzTDzM5YnK1ZL9acHi957707+BDJWUhZuLvxzF1TXFtK2JRIRuhWmWiFNG1Q45AGUlJOjs/wIdBtigt+lFyuJeTEYrlk2HSICvPZHGscjXO4ZoK1DmMsMZZgmlPG2EzOxcw1JFBBKuTLWiiNMfm2Uh7OGqwVrIARaJzgnDCdWpwzTCYGay2NNUVJxiDGwP4czXB0dEBKmebOcWk7ZpIKPoFF8DFR1jAENZAlo1YwTYMIOGNAlRQj0QfC4K+OEmII3Lp9i83ZCmssH7vxMSbNFGdbjHGcna24d++UH92+Sz/0zKYTmsYR0oAxBtl4DIZu4Rk2HaqKaxpeu3nEfNLy+uGMSeOY7c9QhNVqAJS9faGdCNdenWCM4IdS9NUARqFVgzGO+X7LvJ3xS3/usxwfn7JcblicLVmcLEkxsYwZn5VJ1zFxjkNmiCnO1YrhYDLHTqdMjq7jfeDe7dsMyRNrzeuj5FKVkFXxw8AwDDjr6iwu/7JmhmFg6AdSSuSUiD4gqqQYUCP0qzWiQr+ODH2P5ow1hmtHR+zPJlzbn+CsQZwhpkwOCdVMDoYsxZLUCKRCkyQSGCFKxFrFtA1WDEeHB2iGa9eOQGG92JBSJmUIWRliscjQZgyQsiLJ1IUesEYw1hSmV8AYHsukXi5EzZl+s2HwA9qAbQqN0A09WeHu8R1Ozo6ZtA1GgJDIKUMIpJw5O16RYuL0pKyOpdRxeO2Az//K57h+dIDxG/q+462332K5XHPnB8doTBzsOZpGOHnf0LaOo8PrGOtKwBZBnMFax97BEW075Y0bb3Dj2mtEb7lz5x7Lxf9kdbYk9IGU4WQz0NoI1mGdIUSFkOiXS5oQYG9OzBnbGJppw4HMr05gLkglknIi5UhMER88y+WClJXFYsFqtQLNxQJCQIHY9wVJ+UAMkX61wecILpVYYAxGDFm10BRdx2azoVv3aEzYbHGNoGrwraO1c4xJhEJVYRrBWodmSzuJTKYdWQVnG5pmgnMNxjmiyeSUCbkUlg0pY4GgWqBviGQb8SGgVMTnLNBenTxBVYnBM4SBmCK379xmuV6RsuJ94Gtf/Sqr1YbgS1HuYtOTY2R59x45Rmwuizs+FbdijqakfuDtt77NbNoSugV93/HOD39ItxnojgMaMytbYOneYUE2ZyclsHabAQHm8xZjhCwWxGKnb4OxiJvSD4HJbM48C6u0JIeEj5mEoH1EjKANSDJMNgNNUsziDGMtk3lLO2sx8vgq1ktVwrglaeT9ffCYznBycowfAsuzMzabHkEKzZ2VHGNBFyFS8p3CjIqzOFMeMQS8gO8G/IhEtGD2jCHESFaljZU67GMh+DYliVNNGBGigmLQjQfjcJM5IWWyltU7pOQsKpCoFqCgGSQrLoPVuqIm0DRNuW/z+D0+l6oEYw3zgzlRCs08+IHgPf1yw9APHN+6Rd97BINBmFiLKJgsGHGQI6pl/cway/Wja+wdHXA4P6BxlvXgEbW8duMN+iFyL6/xQ2CzXpEFUrNXArxPpJQZhoSmzGq5KQR4XTcOYkkKfcwkhaRCypDFko2gzhR3RFm9yFkxGaauQZopk/keTeswde35SdvDnqgEEflZ4N8Ar1Nc6BdV9V+KyCvAvwfeBN4BflNVT55wLZqmoZ0UbocImhTf9/hugJiQlCq1YMia6ypxvRkZ35VyF6EgkhBS4WmSQgbBYkTLRg9jyGLIQMgFkvpUPhtVSh1BLIQiUkjFQRMxKxsfUQTTTNCxqqIsbm9RXVk80m1VXuGIBTAYa6sS6vV/XCUAEfhHqvqHInIAfEVEfh/4u8AfqOrv1I3kXwD+8WOVYA2zw33MbEpOibTypD6wOjlGO8/1dkIyDYgr3E+IZX3Z5LKQ3zaICk4EcZbVakPvA5oybeM4aFqyZlbrNb2PrPsBHxJdKsPlV/19JYy25B9OkZy3dMli05FUiUgpmbFtQVFiwGit1NadR4lRgy9oq9t4cob5flOsgcfq4MlKUNVbwK36eiki36Rsn/114Ffrx74E/PcnKgHBOIPFls7ZWHF7RnJmYh0ZBXGFfYyJrCPQBurMLxNLCCGRFdarDcE63LRwPf16wKdcWjTjIk8mjsfE1LlaZrFKqc5LWtxPTJmUlSym7tisFmjGqor6Xuq811yq++pKXogREywpZVTNljH+sZVwYRBF3gQ+B/wv4PWqIIAPKO7qsaICagWSgEihC0RwWZEM8+leYR/FElJdcDeCaVrUGIxrSunLokOTEjceBDbLDgssKIO98gPSOK594uNgDcasKofTAdBOWgCiL149GQeihJAImvFJSbkwrojB2aa6lhLgc10cN9UiJOlY/UpMieVyQ1dzCmsNxj2+9uhDK0FE9oH/CPxDVV3s7lRRVRV5eDHo7j7mo+sHJC03qDljKoywxiI2Y2OZMDHnmqEmkipYg1iLOAepLJtkzcRUmkwpYRRMjqhmhuixZZV/WwIjyNYNGVMsIRaHDbawrVnKcoyK4bwcZLdGaryagpzXIxtTYpYxBhFD1gI8UiolMaX+4CnRkYg0VQH/VlX/Uz18W0TeUNVbIvIG8KOHnbu7j/ljn3hN16sNYShYcRLAqGH/8Ig8RDb3lgy95/bxCX0InIYBtZb5wUHZ4N20aEx4Sm6RkEJLDBHJGR8V0VLIZXJkcbZCrKXrelJMaFCMLT5eRAjGoyrgyo6dbANZwU0m5JzRrIgxxJQQVUaSJdYKQLEOMYK1DcYI03aKdY7JZIaxlpQEVcWIeewOnyfuhZMy5f8V8E1V/Rc7f/pd4Lfq698C/vOTrqVaKN5U0UnOgmLANmAdEcEr9CHSh4iPiVD9c6r+NmWtsLGsWuVcLCfkTEgJnzOx/t37gPeeGCIxpe2ie67njjWlKuVBZVONc/VhMdZuwy91RovU+FBfl0rA8lljTM2Fdkf9wYLlXfkwlvCXgb8D/B8R+eN67J8AvwP8BxH5e8D3gd984pUU8A7KGgsZV/C7gWDgRCwdwnFWhqR0UUETrDqM81jnyFnpYyKroql8X0VIuVa5JQTBWotBCasViBBTrrVNYEzCrNeIKS4IodY7Uaho55i5prinkfmsA2rs6NrKd2IUkk6wzlWU1YAdwUOu3wpQvhng/k0sH0kJqvo/eLQa/9oTB/7CxQTJgmjpaKaYq89KUMWr4im+OlKyUlGIMZUSQ4WskGphsFaXrmacaQWXi7Ol6roWNiklx5BKa2Y9RzQyFrBKyWxFFVxJsErFt9Zkq1KilKfz7L+iLTFQB71kG+XePsz3eFw6bWHFEbXM4CFEUkysug7vPYvgGWIkGCFbi6aawYZASqnOXoi5BDyDAStM3BREMKN7sLYMoLmYsZqKxmz9uwjVndT+lRqcrZUYWwYxa7EUlR1ljW5pVK6Mrq3Uo2ZJtaSnBOjHbbm99LqjnJVc3UeIRQkhRkIsSEgpSZ2owaS6a3Kb75wPpgLG2AuDOg7yNlMdzxtZTjHbz5Xr1Azg/siY83YWj25sNyaY+5QAbK9rarwYh3z374+Sy13UyZlu09F1PTFmht7Xb1kpz6hirDCbz0gp0bTN1i2Ms01EaI2rvrjZuXmDdTtKqDAWHs3djFB1LIrNdfBTSttzVAv2325NkYsWNcpIyRhjaGqsQPOF44+SS6ayIcZAjJEYMzFGUkp1JS1XUxdcDXSjPx65o6KIsvGjQEO3RSpbizBl/bi4IXO+R4BzZYwZwJZirkoQuejHx9dWiwJkzA/MRYsq5xZLM6OlQOnrFj1dESWUhf4VMYz1OJGUMrFWqI2UhmuaUkTlS03pFmVUSvicGpZa+FvriaxgjdBO3Nb1FNEa1C/W/mxdxQ7u2K0TGi3DWlsmRA3MUgf0fhcz0hnFVwqNK/nDFbOEsrJWqqXHGXfuKowdoZxg1JRq7VqTRJ1pdSrW68mF88UUBCPGFH9fOaKRuxE9J+/g4Uq4GLjloYF5dIu790Vtp1T5605bcuEzD5MXoIRUs/7xZg3OASI0bbvd/6VZGXHpqATdLtSUN0WRFKVIIQeNMRhbXICrsYHq1nJ++IBodfbjwM7ncwBijLWWNFZaQx+gH8p1c80r6maQLIXKAFKSq4WOxn1mZYZVzG1GD103AkqdUaJYa7ZrtVTuZ3TxZY9arqymqe5qzGS3l0TkvAJ8nNlbdll3n8/jwi6i0Z3n2vBjeKCLFpFzxhhT3eUVUYI1lsPDAwRXkrU8oo9ck67yPoRQFnTq5jwZ3U8euRsla+GdiqcpwXvMYEsJTSLlSl0/skejhVQ4WkOGsaW9ESUVykNJWjLo+2PChTwEMOhW4Smlx9ahwqVbQglSohVCViWIKeu4ibzdJGJUUHvxiwAz46BmTM2eR9c2Xr+My4huxtcPqkE5Rz8XHqWh0l4Nznk3bxAq+1uvs/NaxsDNbrx5fI4Al73GbAyz2RzNoz8oLibEMms73yNCdUNSyTOtC/cKUtyTFVvearWg8ZnybGT8/roxGF+ciVnHvOA8EdN6vGxUjBfyhVwHv4aX863sW8UpRgRrCgpy5jyRE7iQTzxMLjdjrpi5TND7d8KXgR5JtW1uoILKuD1bL34etgPzAGkJ5Xr392E8pmN7sLNXtP6vFwa4dqXGh/P2d61gq3TOLfKCRTzGGC7/69fqHSnn/jJEv80XMuc3YaQU1mrOZAoKyaqElMeL1WfZueFzZKKJC4oYB3J8r2P8qJamUmkTIwUFW7v98G4Vo25dXFnAz7rrBnO1gPp35ZyNfYS8oC20Y0DMO4/C9xeaf0zO8g4UPX+cz8qtFsr/D5tulZQ7fyvnx0tBUXFrFfaO+RYq54v0jMrb4ZJ2chxTkde5kZz/7WI/Hy6Xq4QK20KIjJvEcy4bB1PKxBDKpr2Kt7fmPnqemjk7V+mKGrhzqnvgduiGekJ5knPCrljKefa6RWM5k1NpzIjdckTjZ8ZYkes2rfPvXK1gY6QnoHyhlbJldV3jrg5EhfGG8nZ2550ZPsLwC/6Y4sJEpLgnkS13MzrasQpitw3Y9cdl8B+gn6EsvoyxSs+r53blfib0YesE29SkxrHyxSM80N7D5PJrUWO8gD5G+Adl/RdDKXHZGchz6rnOYFOCeqr429bj47XKW8HISD+YCwO5azEixbK2izh6PtPHTHdEN6OixgRsxP8pJdSYsjnFGNrGbZXyJAXAC7MEvTibtrNqrK7bmeXcN5tG2PEYuZg3PDgTt5kzF332roJ2+3vh2g+hLXZbHj9jdvr/sPMuXPPDLL89KxGRO8AauHtpjX50ucHz69+n9CG/n3CpSgAQkf+tqp+/1EY/gryI/r38dakrIC+VcAXkRSjhiy+gzY8il96/S48JL+VBeemOroBcmhLkiv06oYj8rIj8NxH5hoh8XUT+QT3+ioj8voh8uz5ff+6deejCxjN+UFaL3wZ+DmiBPwE+exltP6ZPbwC/Ul8fUH498bPAPwe+UI9/Afhnz7svl2UJV+7XCVX1lqr+YX29BHZ3IH2pfuxLwG88775clhIe9uuEH7+ktp8oIvImT7ED6Wnlpz4wy307kHb/piOz+JzlspTwoX6d8LJFHrMDqf79kTuQnqVclhK2v04oIi3l1wl/95LafqhIoTWfyQ6kp5ZLRCO/RkEgbwP/9EUio9qfv0JxNV8F/rg+fg14FfgD4NvAfwVeed59eZkxXwH5qQ/MV0FeKuEKyEslXAF5qYQrIC+VcAXkpRKugLxUwhWQl0q4AvL/ACVKVw1yhAFDAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAGEAAABiCAYAAABAkr0NAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO2d368kSXbXPyciMrOq7q++3dMzs7szZte7Y+yVLFgL8AM8ICEk5BcMDxYgIZBAfkICiQdW8AdgeADxuhKWjIQESCDhB8uWWWPAL8hg2d5fZnfHO7s7Mz3Tv+6PulWVmRFxDg8RWbe6t7tnd2bnzpW2j1SVVZlZmZHxjfMzzokSM+M5fbTkPuoGPKfnIFwLeg7CNaDnIFwDeg7CNaDnIFwD+kAgiMhfEZH/JyLfFJHP/7Aa9aNG8n79BBHxwNeBvwy8Cfwu8DfN7Ks/vOb9aNAH4YQ/B3zTzP7YzEbgPwB/9YfTrB8tCh/gt58Avrvz/U3gZ5/1g8OjG3b7xZe330UEkLIVKG/CxJtS38V2v+9yrmBl585e256//YU9fs720OXBp9DTjtqjb088//Fzvv36V++b2e3Hr/VBQPi+SER+EfhFgBduv8Q//9e/jGEIQhsC4hw+NIhzmG8AR8YhCG7aKohBMEqnmWGASXlp3W6fN2vthQAIpgVYlXqhx2gSyVvRXAFSLffRx+CfzlMt98k5P3q9+tKpnVbO+/s//9PfflIffRAQ3gJe3fn+St33aIPMvgB8AeDTr/2UmROcK52bMMiZPiZEHK4TxHmcdyC1ZwGrnaBimAmWMqpKjBEA3wQQUAw1ZRgHzAyRBieerp0jTjC0AGhWBoLsdK6B1W6W2os6dfZjHDV1vtVrTd+n62m9kj4G1tPog4Dwu8BrIvIpSuf/DeBvPesHxtTgOjJTRtUYx4SJIyRDfMCHDhGHxyMGMg1NM0wVHROmyjiMgBHaBhFBUdSUzdhjakDA+4Di8N7jG6H0Ux3523ZNrXu8tTufJ5G23T6Fo0S+d9970PsGwcySiPwD4DcAD/yymX3lWb9RVVb9hpwzOWculheMQ+Th6Tk5G76d43ygm+/hnSdIU1WFYGYkzZha5YTMsOnBjDY0Va8YaplN2pDVyFkIoeHF2y8zn825/eIt2ibQNOV8U63i4nvluokUGHYOFUloT+1Y2QJQhs1WX70HEB9IJ5jZrwG/9v2er6qsVitSSqSUubhYMY6R5cWKnA1pEs55ZlFxztNIlelSOiBpKuyfCxhxGMCg8w0iIM5QlD71qCkxgvcNs/kFMWf21nNy1wIUkbgzak2EbY9vN9OHrRb/ATvokb566mkfumLepb7v+eoffY1+KOLCVFCFfshkNaI5EEfTLIs4smJBa31Fy0XuV3neOo8TRyseL0LTFCsrE1FT+iGD9Sw3G9rQcHr2gPl8xou3b9O1Hfv7+/jgadq2AKKTQi26Q61wguzIrWd15qVil63e2dUZT6MrBSGrsl5v6KvMFkIZ4RnUIKuBKEhCENQcIOSiUhkpIGRTEEF9gxfBRHEimHhwYKJo7URTI6aRnDIXlQtnsxldN2ICTdMwM8M5h5NH3SZ7/ItdovE4FmY7Vu/EQ9eRE1SN1TBiSjVSDRBc8NUY9cUkdR4RR7DSvIiSMTICpmzGVBR60mLqisM5YWYNzgnOF3vK+RZzRu4TY87cfXiCc8LdkxO89+wv9mjblls3b9J2LQf7+4QQWCwWW1AEEJt0w6MdqVA4pmgAqs10aUVtracPUSf8oFRM+EddsV2aRLSIIRhua/tXm1F2TpxMSDMihjPBZ1e21QEM7vI+BsSUQSCq4p3HDNox0jQt3Vh0Rdu0iPOE4GmbtgwOe7S1u5bS5BOA1NEuW07YYrc99mS6WhDMGJPhvdt6y0XOFmenjHVBUBxu+4Qpx8IJHlSEWdsVXdKPqCqbFBGBqEWshODxzjELJS4jLiBmxGyogVOHMyNe9DgGLlYbgvfM5x1N03B8fJPZbMbtF27TNg0zH8p1vUcA1YxSfMIyXOqgmHhhC1LZn99Dn185J6haGenfywjT+Kn29uUYMyvuT1bDRHAuMEU6pP7OzMiaC3c4qVynGIKb7s2uNwtSTdRxNLJLmBhNSoT2gpgTe4s9UttiocE7T9d1xbQ1q6O8AjCFT0wmhx4Atal17pnRkSvnhCGOuFwUoEgRG0218w22PVts9cvwRFZlGAYMaFopOsMVnslZigKOEZyQpXCEAU6EUBWuTWhMusi5bchKMfqxZxhhtVkRQuD8/JwmNMzbBW1ouXl8TNO2LBYLvPP4EGBHmatmzArABmSkfr9G1hG1QaaX8bpiiuoWBDEpJihWwhQUMBADm5yrDGYl1CHgRVAnxRS0wjlqVFtq62/XUEW5sUgBYtIYlwrUMFOSZsJmTfCBNCpNaPGhoWtbDCGEQNtJsahcBVkLh+TKJZlLrnsWK1yxODKSGUw6wIq9b3LpbYqUuFEWV7xWAZVczFAPmJFTLOC5BifQtR5Vh+QilVW03quAHbV0dxCHE8Ft3TCro7Z0V7bLGFA243x1XtlwiSDcuXuPEAKHh4d0XcuNG7domoaum+G8p/Gh+Ai17bmKJ61i72l0tZxAsWgm39RMEROSFk4QimWUzQOQ68hXFBPd6oFLaVzGuZMiFbzKTkS8jup6/hSvMqug205Ar4bcJsUqk/WlWuR8Khw2poR3HvHCEFtcaGhCwyxGnK/WlBNwHqvmxWVY5NpwAuRtKMDqCDHGPoEI3jmcc6iUsIKXdKkfMJwHRwnsYdXPqDLexJDgyz2qmDa3c8/t/aQq1yI+ilWTC8e5SaNum0hB1AOCZoMM69MlThx3H76Lc46m6fDO0XVdCbc0TWl/CIXzqlX1NProOGHyaCle7WQuKYbLGVFFnWOyZHemeKDaJFv/ocpbqVtXBjCarcSdNG/1sSBoBaHIIquctjNAKk02/hS2yPX6ORfOSqmE4Js44pwnpqGC0hSDoJri3vtH4lSP05VbRynFbYeb2iW7mqKaABjj8EjjnUzOl6+hhUmpum08x6pCnuS8AmO15xNaLaXqlws4BF91hkkFQbRYZJUDxBWvha2+qs+xfR5XxGksB9ZD4SyxDKZIHqoBIdeHE4qtr5eTKduBXD4UEIolJCI41yI1oCFWwTAQp5RurNaS5mIR5eIRO1eiqo5LgAFwl/fderXbllWOsEsQHK7oj6nbZeLAcnx7bnXKTEvnkwdME9qvCxg7nPwk+gj8hIjfesuuPoBimhmHcyATvBGCZza7SXCOtlqoTssDOV8cPhMtHvN4QU6Jsd/gvefWCy8jvqHPnphL0FANGt8BFLCq6iycUyHZ1QViJAoAAUXEcC4XHZInT3+SlSXWbmmAHNH1Q3TccHFyF82pOihPpysGgTIzJsLlHHIxS02sRESLw1teZnhTvDocEKTIWHGumoKQAV9tqCCGd9AEhwuhRFkVciwjPIQWBJIr8xJSR75aqqN60h1VrEyoTBZ/PU/U1ZBEifJuY0M5YSmiw5o8bkj9ipwiplMQ7Ml0xYrZtpPiDscsdHjnmbctYIydLza7JBzQjJlAps1CcIHDwzkuBGLbokIJUyB0TYv5gHQdLgQWe3tIM2O/vYEiHB/Gqn+KCBq1GKWjGSlnzpZnpBTJaQ1meMslgJi1bC0VnZU3pe2uBTxKQwlZ+CKGxh4dN2xO75H7NePyhBwjw5CeGUm9co8Zq7a9E7wTgpMSw6keLDgMh6qS4lgineYhCF4cIo6sxaJPWcsUpRpihq8JBDEmxCLiFcRXTlNyjKiVMKGKoD5gOGg6BIfVBAFXucSTEVNER0wzmvrCId4AV7x/ceADmOEs4yxDSmiOWE6YJnJK1yeKCobTzLyd0XjPfucxNc7OH5JUGbwvnWsZSwl3dp/G4PbeAYv5nCMvmIN3zs4Ycy7hDzWafiAA8y6gRO7d/y7qAt3RUEIN3pFT5P67d4gpo90CaWfMX3wF17Y0Ry/hcqTPCRt7fFqDRmbWI5aJ/YqUI2N/jpkS/BzDsYkeE0+7OMB7x34wzGU2klByCY1IsbKeFT66cj9BMBrvaJwrCrnGaZIpWTyKkM2V2IuW6RJpPNL4Eh8SGE0ZbYqQlnkHMUhqZDM2/UCWhM16fFYIBYRx6Ik5E80hCsRE4xuCK+EG386KvTVWRaoJciSPPTmNpH5dIrpeMXOMo8OcB+dpQsCcA82guXDoTlLAtfETMCOYctR1OBEuNmUCPnVKFoebz3CuzLJZiuR8QeMdtz75CrOuY2g8oxpxryVpCeCJgcQWy8pm6Ikxs4kjGYcNa3waiE4RU7rO4TOsx4FkRr+8oEmK39ujaWYcvfwqNm7YvD3CZsm4vIuNay7O7pHiwLpflrkE9agKqxGQQLt/RNc2zG8d4CyR+zV52DAOI1lrCPEZjsKVgiAC3jlaXzoaVbCMeHBeEF8ikooDlKbrmAfPfLGg61rWKaJiiHeIs2JdGQQJRa9oKJM43tUHL2Zs0ogzxTcenENSRgRyTkhKJNWiB9oWAUK3h+WEmqC56JIURzRFVJWcM6qQo2AkZGhxlkhDgydjOVafoXrz1Rp8Gl0pCM45DhZzDmdzBLi4qKq48eA9IZQRoznjvOPg9kssupaP37qNOMdbD+8hqsyDp62pL94JhwdzggikOTEmei9llm3RYGasLy5wBrf2bwCOISRGc6yIWN6wWp0zpg5pb9G2DTc+/uPY6oyz+98l6ppxjOQ0bgOFpiW87iaPfH2ODI4lGzxK3Gyq4yjb859FVwuCCPOu5XB/Dwes13uMOZK6wgG+aYr1kxQvwo22Y951LNriZHU+oGrMRcmmJDOCE+ZtW6ysoIze0XUNLhvSFAvGu+I9e1+sr7bxmArOEpaFHPsy9xwjLgTEd0gzx7WL8vIB1QAaS1S1ZnFAiQJjxSHTNG7nPbAp5XIKbV0T6yh4zydevMmf/VOfpQsNd995iSEO9GlETUmpmJs5ZoIPvHjjJk1oCN6TVLH9I4acWOUiQoahx3vHCweHeO9LHmqMbIbImDLWtqgpM9sDNRpKVxzOW4as9MsL4gBDHpDQkqPStnO6wxu0zDj62GvkoxewtGJYn7E8u0tKI0Pfl4hsVsTBrHEErwQZS4jDlfhVHnQ7D/3MfrmS3q/kBOat5+aNfRbdDJdHxjiyGXuyKikWpybHkg1x8+iQ4DyqJQXyKC8Yc6ZNiaSZ3pfQ9/5sjveOlDNeHPtdx+AzFgJqSuhmWFbIDtViAgczupqsMsShePJDX/RAypgIYX6IF9i78QKhbUk5EYeenJbF9pcRESP46u/UOY5tTGr68GwMrhgEjMNO+cyP3eT46JD4Y7fIWdkMI2pGUikjKBfTTlJxxmJMZFVeirfIqvRRyVnph6E4c22R/at+wxgTDQ0xJWIu2du2f0RKidOzM/phpD95iMvGx+b7JIOHF2vSKIh5fDuHdg7djO74VVqvvPyxl7C4YXn/Dv3qgjtvvEG/uWB59g6WRzq3Lgn9aagDRojqyjPYlHIJ2+DURwkCGGIZ7zKtN7pFixl0XVs6X0ouUNYS5k59T86ZoYLUaYeqFRBU6fuuPKiTEhYwo/WFm1LKjLFYM6aZmBKp73FqtCIED/uLeZn9UogqJO9wXmgwAiWP1QdhNjsGXWAKzWzFapXoNhdIcGjqadIp6EjeLEkpo8RtWB3hMiT+FLraDLysrM/OeHj3TXxecXz0AqFpWSwOER9w7QzEoeLRnNksl6QYWa9WACUmJK7mrirDEMlZGceRlDKbfkHWkhGnagxDT85KjCNxHDlqA6v1hhATOM+nPv0avu0430SGpDw4H0gmaJiDC0hKePPQ7uObA2688jIGHH3qc2hKxOVD8rBic+d1htUZ99/6FpvVivP+HTIjEgbE6ty4GAzDE/vlyj1mUyUOG8Z+RtobcCLQFa83BI84Dy6g2ZPbFidCiiMIzLu2VPZ4JavRhKaA0TtSyiWGb4YLLZgxdg2aM+NYQEj9hlnT0K/WyKRzmpZulhliBtkQE4zWkPEktBg7yaHmkFBsVNc2SKM4hNwt0PUKCQsWFz3SLJlvIm7oYXOGaqIExa+JdeREmPvAuF6yDoo3pWnnHGShaWfMvSeElnbWgPOwmJFTwElCMBazMrGzN5vV+oEivlKKjMPI22+XkX9wUHJKm6boiqHvySkzvHSLFBPnr326pGP6BkXoM6RsfPx2Jibl7CIxxszp+YYxDizv3ydqZk3JAhnbGfhAMzvCu0NmH/9p5pZ49ZWfIo9rbr3zBuuLU9785h/Sr5csz++VeYWn0HuCICKvAv8OeIkC5xfM7N+IyE3gPwKfBN4AfsHMTt4LhOBd8UZTYhx6VA13scSFgfWY8aFhvrdfwhciVbYW+Wo5Ys5VGetqukzJObUgtKHEl7o20ISGtiscEcTQlJlJqS/r2paUlaGmRbYZYjaMyBAzY59xJuw1QoMQveJM6WMqsS3NmAska3G+QdsWL4FZuwDnmB3eBO/YOzzGOaHvl6Q0vn8QgAT8YzP7PRE5AP6viPwm8HeBL5rZL9VC8s8D/+RZF/LecWMxx6uh48gqLsm65PRbd9j0I2++c49scHR8i739fT7z2mvsLWYc7c0IDiKxeNi+BRy4pk6DFkvk+GiGIHRdg/eeJpRMPW06UMUWTVH4uYizzZjJZgxZGGJC372Py5GLvMSZcfOFPYw5myPPECN3z85YDyN3Ht5nFZU3xzsMeMZmgXjPwaxh5oWX9g7oFjM+6TP96gz5Boz9hpP7b78/EMzsDnCnfl6KyNco5bN/FfiL9bRfAX77vUDAjBgHNus15AQuErPy4MEFF+sNb735NlmNISb212uObhzR7+/RyBFtcAQbcYDzJcthAmF6OR8KB1lG1LApE7dm5gnTHHR9+CCgIDmjeWR5dsLFas3ZyQmYEISSKj9vaVvPEHsaZ5w5QyXT2khWxwaH5ED04HFkLXUT3WIPQZkv9vDu6VOcP5BOEJFPAp8D/jfwUgUI4B2KuHomDUPP61//OnZ+l3nXMd8/YEyZb735NsuLNd956118CPwEcL444+79u+ztzfmZz/4E+4uORVCCdzTNHHEeHzrMIKYRJ46D/QOCb7BmVmJQKT8ymSI1UThZEWO4hpgy7z444d79B/z6r/8GDx485Ox0SRMaPv2ZP8mNm7f46Z/5WY7mC46Ojxg2a7o8su4HbiZhneGtjRDNwGdEE6uH54weXjreo2kbXvz4K6Rx4Cu/9wFBEJF94D8D/8jMzh+p9zIzkScUCJffbeuY513Ler3mbBkYhpFRYUyZs7NzVpseM8V5z8HBPu1szvnqAixzcvKQOHRoV3RK05Y8n9AUEMY4ltR1gRAaaFNJZawgbFMsXcme2ILgE0OMPHx4j3v373H37h0ePjxltVwRmpbDezfIZoxxpJ3NaJuA5JZ5E7Cc2XfgMiyiMhpoSe8oWSM73DmbzcnBP7Vvvy8QRKSpAPx7M/svdfe7IvIxM7sjIh8D7j7pt7t1zAd7Czu56PFBaILDny1JKfHugweI8/z4T3yGl156mZ//638N5z1f/K3f5vzslK98+Us0Ynz8eI+28YQQcN7RzRalszWXCO3eIW1oONw/wvtQ8ozclL9UKm8U6Ov0aPaO0+WSL/7W/+Kdd+/xzde/zmbT19wm4cHpQ24c3+KFVz7J7dsv8onbLyJAExq6oCysJDoezyAaqBdUHYPW1Myc8AK3btxkyr99XyBIGUb/Fviamf2rnUO/Cvwd4Jfq9r++17WKQPZEsxL8skhKiahK4z2L/T0Ojg45vnmMiGM+m7Faeh6enuEssZBI2wRCKHXJ3XwoilcV7xwaE01o0aQEH/De16KRUEEoeRmDGVmMUeDs7Jx79+/x8OQhSAmBzOdzVI2T0yXr9YrzsxNmXUs6PsZDKclyss11ynEkG6iV1HixUnOHZnDU4pIP5jH/eeBvA18Skd+v+/5p7fz/JCJ/D/g28AvvdaHQNNx44SXEDyXZVyPZPL7raGczbty8ydHxDcSXUXt845h+veb333ybYbNic7JP23i6tsF5R9uVovMCr9CGYhXNZwuC87RtW3JDQ+GKJrRl0igEMsYy9pycL3nj228wDJHXXvsM+/sHvPpjrzIMI//jf/4OMWa+9fpXOD95l5eP95nPZrjGkGxcnFxwvtnwnXffZcgZ17UE79lvZ3QhoK1i3uHCs1dy+X6so9/h6XHAv/Rev98lESn5QMQy62UOnBWrxhcRI66M7JL5VvJ11puefrVm2ZW5gLFtcN4T+rSTUi80tayp70b8FgRHE0qCbhtKPZo1gYRx2q84W17Q9z2G4+bNmxwf3+SVV16h73tu3Dji4mLNerWkDZ71eomQiXEooZDYM449w2ZFnxKNdZj3qINMII6+gKBuSv57fyD8MGlKNcw17cWHFueVbg5t29KPidV6w+nZOSLC2XLJ2fkFD06XrJfnWM40taCv6IKSpGVMIHi8E2ZNqFnRJZe1qxzRNoUTzAspZ+4tz9j0I8MQOTy6wZ/+3Of4xCc+wU/+5E+yXm+4d+8+b799hz/88lc4O7vPH7/xIvuLPdIwEmPi/OGKdd8znt8na8brHA2BUXuSc/QX93FO6EK4RiAYJM21uLrk7kydqGb0w8C679lsekSE9WbDui+T9zErY9KavFhCwnFM2zJVgCaUovKh8TUlveQptW2xntqmKxFXV0A4XS4ZY8LU8M5zsH/A4cEhe3v7iDgODg5Y7J0yxoFh7Dk9fcg4bNAxEWNieb5mPYyMmxVRtawo4D0+xxK+9qW012piw9PoiovJM+cXS5KOJdPNOYQSUtgMEf3um5wtV3zyU28hCH/8xre5++5d+pRJOJJvwAVi9uScOT1dF8WeymovTU1FD062I09ECE2pEwhhZzUYVVZDD+KYLfYJoWV/f5/FYsE4DMRx5PDwkP2DA/p+zWq14stf+gNC8MR+IMXM+fmGMSurXFI7Q1eWDAqUmoS9wwOatuXw6AjvP6CJ+kMjKzNkOddE3DoHYDkjolys1jgfuHfvPiCcnJ5yvlwScyYbjKmWNJmRkzKkTEqZGFNxwnKtYXOXtoiI4JOWjD9fik5KtX+ZChXnaGe7q7hYWXsjl3OFEm+KMbJcLvHOFRBSZrUaSGpEcZgr+bRltRnBh0BOCed9ifBel1zUKYtcXJl2HMdYQs19cdTMzrn/8JTlxQbLypvfeYt+07NcD6gqw8myFhTUSZ+USrqkFpHkcqlInOpiJiic5Jq3q4+0xtSKqTuLDMPAxcUFy+WSpmno+w3r1ZrNpszW9f3IvfsPS6rOtMBJtpJO2bTb5ytpPSWA2DUzgm+Io5Hk6SkXV5x3VOS01fzT7KpH6xylILMswXN2fo5mZRjGUg3jS3GITaWwaPV4a5Y2tq3SKTrmsg5BgDTVQdQ6tq10rqkQKZcp1NVqxWq1Yn9vj3GM9H1P3w9oLlybYq5Y6k6pbK36qbrJ6pI0plY4XhSxxG6p7eN0tfMJ3nFweAC4sjbFpkc1E7sZ2yUJzOg3PVmNvf19VI1Fnh56quxJmCk5l8WntK6DlOt5VmpYi6lrRky5cloBYdqKONRgvdlwen7GN775DZYXS5q2YRhG3nzzLd6+807hWIU0JTMzrT5TCgNzTX/RlHF1ibcYMzGVSIBr5tcHhBJ2LkV4IoYPAdFpJJX5BlUlpRKG8OYplfq2M/IMtTKHqymyXf+ojugJBNNaKKiGi/V8dZSKzRpVrfJDRMg5c3JyQgiBBw8eMI6R07MzLpbLcn+KrJ8E3WTV2U7FjlMt8w1SlopTG0ug0fz1AWHyE3Tq9OBx5mmaMg8whXv394+qwq7rDmnhEnGF1XOuqeqa6vFa7FFrAIqUMXJdK68fSsJASgk13aaqT2n1KSdwwpe+/GXm8znf/s53SDnz1a/9ERcXK2LMiPia0LKzxl3lqmneKdesuxIlqisGOEdjubT9KXT1c8zThykLwWrFjritGbeNx+bLLDYoUVAwSt8rpn7b+Y9W5JfzNZdyKu9LhneMhXOKGCvgFc4rU48xRgx4eHJSzh/LAiahaXC+ijGbcoqmeuydtfJ2fQGRbQDRuctI7pPoikGoo0OKkybTWjoqj5aaToqOmqvqallSfQ6rlf1TuVOu4mhbDTOlo0/LHdh+SaXJ6RGwdFKw1YGMY0mRuVitUFWOj29yqMa6TyWro4LIxAk2bfWyzKreV7ZbwflSUnVx+uReuXJOmGq8tmtMwKMjiJrLL4DtPIxMvzWsoEApoZ0UOkhdTmHa51wt6hNKJY1zVVTVCv4p7GFlFUkRX1eKAecM15WVycSX5LMwjjUHdRJHk46yrXXEtr2yE9d6do9cvTjaFttd2vHUYN30EFO4Aee3x0QqOGbgCwhTKMDXzkzbkf6YTV4VukhZ9yhrKQScZhylpq5389IW7/yEG2owxLKET6qTRKX43YqyFZiKQbbt33Jg2Tet3/rGN7/+xD658po1K8uiFqqO11Sutp0Fm4aOuwTJdrinvEtdjKScXmrNypbHHKMpHVGkdpbIVr7DJedNozeEEv9XK/VrWYpJLc4XQ2A74mV7/e19auIBsBWTUyLC0+jqA3ixoCBS+liEUhBYG6mq5SFqagtcKvMMl5yBIFJ0iG98jQdmDKO5XImk3viys4xLHbK1kqq2zTWwqKk4hJPVNeZprlomabptK4DVAvlpEmkCYbrn7vcn0UcgjnY+14e3usjco8smTyuAPWkyYxecuvqWwHYhjJ3Y0eUFime96xROpuykR2x77yoqBUo6xuU1dm685cDdOY3d7fT5WZYRfCQgCJNo31ozkqoYCFvHCQHV8vC+igersszV5ZvVinc6rVExiYkJnEu9PHWiq7EdT1k1JmzjPWbT0jq2DY/nnFEU0fyI3J+sKldvZjsLlj9pBeFrBwJcxnTqmKyWzQ4n2PbI5W+23x7dx3Y0P3Z9ma4/bbex7bKBRzhmu3hI1QOl0l9xVurjKkrbk3ekEoh8T8c/Ts8C4n3/k8j7IRG5B6yA+1d20x+cXuDDa9+fsCf8f8KVggAgIv/HzP7Mld70B6CPon3P/13qGtBzEG78+S4AAAE3SURBVK4BfRQgfOEjuOcPQlfevivXCc/pe+m5OLoGdGUgXLd/JxSRV0Xkv4vIV0XkKyLyD+v+myLymyLyjbo9/tAbs+sJflgvysKirwM/DrTAHwCfvYp7P6NNHwN+pn4+oPx74meBfwl8vu7/PPAvPuy2XBUnXLt/JzSzO2b2e/XzEtitQPqVetqvAD//YbflqkB40r8TfuKK7v2e9EErkD4o/cgr5scrkHaPWZFJH7r5eFUgfF//TnjV9KwKpHr8qRVIP0y6KhC2/04oIi3l3wl/9Yru/UT6PiqQ4PuuQPqAdIXWyM9RLJDXgX/2UVpGtT1/gSJq/hD4/fr6OeAW8EXgG8B/A25+2G157jFfA/qRV8zXgZ6DcA3oOQjXgJ6DcA3oOQjXgJ6DcA3oOQjXgJ6DcA3o/wM74YeIAVi/ggAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plot some images from train set\n",
    "n=6\n",
    "plt.figure(figsize=(20,10))\n",
    "for i in range(n):\n",
    "    plt.subplot(330+1+i)\n",
    "    plt.imshow(X_train_orig[i])\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "id": "vLEDjiAxXmHh"
   },
   "outputs": [],
   "source": [
    "#convert pixel values to float and normalize input\n",
    "X_train=X_train_orig.astype('float32')\n",
    "X_test=X_test_orig.astype('float32')\n",
    "X_train=X_train/255.0\n",
    "X_test=X_test/255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "id": "lXcEibTlZlOa"
   },
   "outputs": [],
   "source": [
    "def one_hot_encode(x):\n",
    "    encoded = np.zeros((len(x), 10))\n",
    "    \n",
    "    for idx, val in enumerate(x):\n",
    "        encoded[idx][val] = 1\n",
    "    \n",
    "    return encoded"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "id": "sX-e3BbBZliU"
   },
   "outputs": [],
   "source": [
    "Y_train = one_hot_encode(Y_train_orig)\n",
    "Y_test = one_hot_encode(Y_test_orig)\n",
    "conv_layers = {}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "id": "8HFb2UxoZrTu"
   },
   "outputs": [],
   "source": [
    "#Creates the placeholders for the tensorflow session.\n",
    "def create_placeholders(n_H0, n_W0, n_C0, n_y):\n",
    "    \"\"\"\n",
    "    Arguments:\n",
    "    n_H0 -- scalar, height of an input image\n",
    "    n_W0 -- scalar, width of an input image\n",
    "    n_C0 -- scalar, number of channels of the input\n",
    "    n_y -- scalar, number of classes \"\"\"\n",
    "\n",
    "    X = tf.placeholder(tf.float32, [None, n_H0, n_W0, n_C0], name = 'X')\n",
    "    Y = tf.placeholder(tf.float32, [None, n_y], name = 'Y')\n",
    "    \n",
    "    return X, Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "id": "BORBztIBZ0KP"
   },
   "outputs": [],
   "source": [
    "# initialize_parameters: Use xavier_initializer\n",
    "\n",
    "def initialize_parameters():\n",
    "  \n",
    "    W1 = tf.get_variable(\"W1\", [3, 3, 3, 32], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W2 = tf.get_variable(\"W2\", [3, 3, 32, 32], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W3 = tf.get_variable(\"W3\", [3, 3, 32, 64], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W4 = tf.get_variable(\"W4\", [3, 3, 64, 64], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W5 = tf.get_variable(\"W5\", [3, 3, 64, 128], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W6 = tf.get_variable(\"W6\", [3, 3, 128, 128], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W7 = tf.get_variable(\"W7\", [3, 3, 128, 128], initializer = tf.contrib.layers.xavier_initializer())\n",
    "    W8 = tf.get_variable(\"W8\", [3, 3, 128, 256], initializer = tf.contrib.layers.xavier_initializer())\n",
    "\n",
    "    parameters = {\"W1\": W1,\n",
    "                  \"W2\": W2,\n",
    "                  \"W3\": W3,\n",
    "                  \"W4\": W4,\n",
    "                  \"W5\": W5,\n",
    "                  \"W6\": W6,\n",
    "                  \"W7\": W7,\n",
    "                  \"W8\": W8}\n",
    "    \n",
    "    return parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "id": "LCPgAAX6Z7wR"
   },
   "outputs": [],
   "source": [
    "# forward_propagation\n",
    "\n",
    "def forward_propagation(X, parameters):\n",
    "    \"\"\"\n",
    "    Implements the forward propagation for the model:\n",
    "    CONV2D -> CONV2D -> MAXPOOL -> CONV2D ->CONV2D-> Maxpool->CONV2D ->CONV2D-> MAXPOOL ->CONV2D ->CONV2D-> MAXPOOL -> FLATTEN -> FULLYCONNECTED->RELU->FC -> softmax\n",
    "    \"\"\"\n",
    "    \n",
    "    # Retrieve the parameters from the dictionary \"parameters\" \n",
    "    W1 = parameters['W1']\n",
    "    W2 = parameters['W2']\n",
    "    W3 = parameters['W3']\n",
    "    W4 = parameters['W4']\n",
    "    W5 = parameters['W5']\n",
    "    W6 = parameters['W6']\n",
    "    W7 = parameters['W7']\n",
    "    W8 = parameters['W8']\n",
    "    \n",
    "    # CONV2D: stride of 1, padding 'SAME'\n",
    "    Z1 = tf.nn.conv2d(X, W1, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A1 = tf.nn.relu(Z1)\n",
    "\n",
    "    A1_norm = tf.compat.v1.layers.batch_normalization(A1, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # CONV2D: stride of 1, padding 'SAME'\n",
    "    Z2 = tf.nn.conv2d(A1_norm, W2, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A2 = tf.nn.relu(Z2)\n",
    "    A2_norm = tf.compat.v1.layers.batch_normalization(A2, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # MAXPOOL: window 2x2, stride 2, padding 'SAME'\n",
    "    P1 = tf.nn.max_pool(A2_norm, ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'SAME')\n",
    "\n",
    "    drop_P1 = tf.nn.dropout(P1, rate = 0.2, noise_shape=None, seed=None, name=None)\n",
    "\n",
    "    # CONV2D:stride 1, padding 'SAME'\n",
    "    Z3 = tf.nn.conv2d(drop_P1,W3, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A3 = tf.nn.relu(Z3)\n",
    "    # batch normalisation\n",
    "    A3_norm = tf.compat.v1.layers.batch_normalization(A3, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # CONV2D: stride 1, padding 'SAME'\n",
    "    Z4 = tf.nn.conv2d(A3_norm, W4, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A4 = tf.nn.relu(Z4)\n",
    "    # batch normalisation\n",
    "    A4_norm = tf.compat.v1.layers.batch_normalization(A4, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # MAXPOOL: window 2x2, stride 2, padding 'SAME'\n",
    "    P2 = tf.nn.max_pool(A4_norm, ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'SAME')\n",
    "    \n",
    "    # dropout\n",
    "    drop_P2 = tf.nn.dropout(P2, rate = 0.3, noise_shape=None, seed=None, name=None)\n",
    "\n",
    "    # CONV2D: stride 1, padding 'SAME'\n",
    "    Z5 = tf.nn.conv2d(drop_P2, W5, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A5 = tf.nn.relu(Z5)\n",
    "    # batch normalisation\n",
    "    A5_norm = tf.compat.v1.layers.batch_normalization(A5, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # CONV2D: stride 1, padding 'SAME'\n",
    "    Z6 = tf.nn.conv2d(A5_norm, W6, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A6 = tf.nn.relu(Z6)\n",
    "    # batch normalisation\n",
    "    A6_norm = tf.compat.v1.layers.batch_normalization(A6, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # MAXPOOL: window 2x2, stride 2, padding 'SAME'\n",
    "    P3 = tf.nn.max_pool(A6_norm, ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'SAME')\n",
    "\n",
    "    # dropout\n",
    "    drop_P3 = tf.nn.dropout(P3, rate = 0.4, noise_shape=None, seed=None, name=None)\n",
    "    \n",
    "    # CONV2D: stride 1, padding 'SAME'\n",
    "    Z7 = tf.nn.conv2d(drop_P3, W7, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A7 = tf.nn.relu(Z7)\n",
    "    # batch normalisation\n",
    "    A7_norm = tf.compat.v1.layers.batch_normalization(A7, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # CONV2D: stride 1, padding 'SAME'\n",
    "    Z8 = tf.nn.conv2d(A7_norm, W8, strides = [1,1,1,1], padding = 'SAME')\n",
    "    # RELU\n",
    "    A8 = tf.nn.relu(Z8)\n",
    "    # batch normalisation\n",
    "    A8_norm = tf.compat.v1.layers.batch_normalization(A8, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "\n",
    "    # MAXPOOL: window 2x2, stride 2, padding 'SAME'\n",
    "    P4 = tf.nn.max_pool(A8_norm, ksize = [1,2,2,1], strides = [1,2,2,1], padding = 'SAME')\n",
    "\n",
    "    # dropout\n",
    "    drop_P4 = tf.nn.dropout(P4, rate = 0.5, noise_shape=None, seed=None, name=None)\n",
    "\n",
    "    # FLATTEN\n",
    "    F = tf.contrib.layers.flatten(drop_P4)\n",
    "    # FULLY-CONNECTED with relu activation function \n",
    "    A9 = tf.contrib.layers.fully_connected(F, 128, activation_fn=tf.nn.relu)\n",
    "    \n",
    "    # batch normalisation\n",
    "    A9_norm = tf.compat.v1.layers.batch_normalization(A9, axis=-1, momentum=0.99, epsilon=0.001)\n",
    "    \n",
    "    # dropout\n",
    "    drop_A9 = tf.nn.dropout(A9_norm, rate = 0.5, noise_shape=None, seed=None, name=None)\n",
    "    \n",
    "    # FC\n",
    "    Z10 = tf.contrib.layers.fully_connected(drop_A9, 10, activation_fn = None)\n",
    "\n",
    "    return Z10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "id": "JuMGq9MKaBk5"
   },
   "outputs": [],
   "source": [
    "# compute_cost \n",
    "\n",
    "def compute_cost(Z, Y):\n",
    "    \n",
    "    cost = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits = Z, labels = Y))\n",
    "    \n",
    "    return cost"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "id": "lpxK2n-4aFMW"
   },
   "outputs": [],
   "source": [
    "# GRADED FUNCTION: model\n",
    "\n",
    "def model(X_train, Y_train, X_test, Y_test, learning_rate = 0.001,\n",
    "          num_epochs = 100, minibatch_size = 32, print_cost = True):\n",
    "    \n",
    "    \n",
    "    ops.reset_default_graph()                         # to be able to rerun the model without overwriting tf variables\n",
    "    tf.set_random_seed(1)                             # to keep results consistent (tensorflow seed)\n",
    "    #seed = 3                                          # to keep results consistent (numpy seed)\n",
    "    (m, n_H0, n_W0, n_C0) = X_train.shape             \n",
    "    n_y = Y_train.shape[1]                            \n",
    "    costs = []                                        # To keep track of the cost\n",
    "    \n",
    "    # Create Placeholders of the correct shape\n",
    "    X, Y = create_placeholders(n_H0, n_W0, n_C0, n_y)\n",
    "\n",
    "    # Initialize parameters\n",
    "    parameters = initialize_parameters()\n",
    "    \n",
    "    # Forward propagation: Build the forward propagation in the tensorflow graph\n",
    "    Z = forward_propagation(X, parameters)\n",
    "    \n",
    "    # Cost function: Add cost function to tensorflow graph\n",
    "    cost = compute_cost(Z, Y)\n",
    "    \n",
    "    # Backpropagation: Define the tensorflow optimizer. Use an AdamOptimizer that minimizes the cost.\n",
    "    optimizer = tf.train.AdamOptimizer(learning_rate = learning_rate).minimize(cost)\n",
    "    \n",
    "    # Initialize all the variables globally\n",
    "    init = tf.global_variables_initializer()\n",
    "     \n",
    "    # Start the session to compute the tensorflow graph\n",
    "    with tf.Session() as sess:\n",
    "        \n",
    "        # Run the initialization\n",
    "        sess.run(init)\n",
    "        \n",
    "        # Do the training loop\n",
    "        for epoch in range(num_epochs):\n",
    "\n",
    "            minibatch_cost = 0.\n",
    "            num_minibatches = int(m / minibatch_size) # number of minibatches of size minibatch_size in the train set\n",
    "           \n",
    "            for i in range(num_minibatches):\n",
    "                # Select a minibatch\n",
    "                a = random.randrange(num_minibatches)\n",
    "  \n",
    "                minibatch_X, minibatch_Y = (X_train[minibatch_size*a:minibatch_size*(a+1)], Y_train[minibatch_size*a:minibatch_size*(a + 1)])\n",
    "               \n",
    "                _ , temp_cost = sess.run([optimizer, cost], feed_dict={X: minibatch_X, Y: minibatch_Y})\n",
    "                \n",
    "                minibatch_cost += temp_cost / num_minibatches\n",
    "                \n",
    "\n",
    "            # Print the cost every 5 epochs\n",
    "            if print_cost == True and epoch % 5 == 0:\n",
    "                print (\"Cost after epoch %i: %f\" % (epoch, minibatch_cost))\n",
    "            if print_cost == True and epoch % 1 == 0:\n",
    "                costs.append(minibatch_cost)\n",
    "        \n",
    "        \n",
    "        # plot the cost\n",
    "        plt.plot(np.squeeze(costs))\n",
    "        plt.ylabel('cost')\n",
    "        plt.xlabel('iterations (per tens)')\n",
    "        plt.title(\"Learning rate =\" + str(learning_rate))\n",
    "        plt.show()\n",
    "\n",
    "        # Calculate the correct predictions\n",
    "        predict_op = tf.argmax(Z, 1)\n",
    "        given_op = tf.argmax(Y, 1)\n",
    "        correct_prediction = tf.equal(given_op, predict_op)\n",
    "\n",
    "        # Calculate accuracy on the test set\n",
    "        accuracy = tf.reduce_mean(tf.cast(correct_prediction, \"float\"))\n",
    "        train_accuracy = accuracy.eval({X: X_train[:10000], Y: Y_train[:10000]})\n",
    "        test_accuracy = accuracy.eval({X: X_test[:10000], Y: Y_test[:10000]})\n",
    "        print(\"Train Accuracy: \", train_accuracy)\n",
    "        print(\"Test Accuracy:\", test_accuracy)\n",
    "                \n",
    "        return train_accuracy, test_accuracy, parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 665
    },
    "id": "uFFoWEiLaU8N",
    "outputId": "16296cbc-75f5-4836-b4c6-2c003f872f98"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Cost after epoch 0: 1.863191\n",
      "Cost after epoch 5: 0.907195\n",
      "Cost after epoch 10: 0.674389\n",
      "Cost after epoch 15: 0.560958\n",
      "Cost after epoch 20: 0.477994\n",
      "Cost after epoch 25: 0.419867\n",
      "Cost after epoch 30: 0.371566\n",
      "Cost after epoch 35: 0.334123\n",
      "Cost after epoch 40: 0.302258\n",
      "Cost after epoch 45: 0.281272\n",
      "Cost after epoch 50: 0.263376\n",
      "Cost after epoch 55: 0.242304\n",
      "Cost after epoch 60: 0.234722\n",
      "Cost after epoch 65: 0.214463\n",
      "Cost after epoch 70: 0.198589\n",
      "Cost after epoch 75: 0.190348\n",
      "Cost after epoch 80: 0.183613\n",
      "Cost after epoch 85: 0.178441\n",
      "Cost after epoch 90: 0.170148\n",
      "Cost after epoch 95: 0.160700\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train Accuracy:  0.9501\n",
      "Test Accuracy: 0.8252\n"
     ]
    }
   ],
   "source": [
    "_, _, parameters = model(X_train, Y_train, X_test, Y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "id": "KYkuNjc_P2Hc"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Model-1/ Copy of TensorFlow with GPU",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}