Profile_picture_classification.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import cv2\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sqlalchemy import create_engine\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "engine_sqlite = create_engine('sqlite:///app/posts.db', echo=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "query_profiles = \"SELECT * FROM profiles where label IS NOT NULL\"\n",
    "df = pd.read_sql_query(query_profiles, engine_sqlite)\n",
    "df.set_index(\"index\", inplace=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "label\n",
       "facebook           15\n",
       "google_new         20\n",
       "google_old        549\n",
       "miscellaneous     274\n",
       "self             1093\n",
       "dtype: int64"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.groupby('label').size()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def class_to_no(label):\n",
    "    if label == 'facebook':\n",
    "        return 0\n",
    "    elif label == 'google_new':\n",
    "        return 1\n",
    "    elif label == 'google_old':\n",
    "        return 2\n",
    "    elif label == 'miscellaneous':\n",
    "        return 3\n",
    "    elif label == 'self':\n",
    "        return 4\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "df['label_num'] = df['label'].apply(class_to_no)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "import cv2\n",
    "import urllib\n",
    "import urllib.request as ur\n",
    "from PIL import Image \n",
    "import requests\n",
    "from io import BytesIO\n",
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "def convert_picture(url): \n",
    "        response = requests.get(url)\n",
    "        img = Image.open(BytesIO(response.content))\n",
    "        img = img.resize((100,100))\n",
    "        img= img.convert('L')\n",
    "        return np.array(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0, 1, 2, 3, 4}"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label_num = np.array(df['label_num'])\n",
    "set(label_num)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[0. 0. 0. 0. 1.]\n",
      " [0. 0. 1. 0. 0.]\n",
      " [0. 0. 1. 0. 0.]\n",
      " ...\n",
      " [0. 0. 1. 0. 0.]\n",
      " [0. 0. 0. 0. 1.]\n",
      " [0. 0. 0. 1. 0.]]\n"
     ]
    }
   ],
   "source": [
    "from keras.utils import to_categorical\n",
    "label_num_ohe = to_categorical(label_num)\n",
    "print(label_num_ohe)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "images = []\n",
    "\n",
    "for img_url in np.array(df['profile_picture_url']):\n",
    "    image = convert_picture(img_url)\n",
    "    images.append(image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP4AAAD8CAYAAABXXhlaAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAIABJREFUeJztnXmspNV55p/Tfft2N5duVtPgBtwQMPtmMG7weAeMPU6cWM7ETmShyJYVKTNxMpESO/NHMlJGSqQoyx8jSyieyDOKxsk41tjyWCSIGIexDabZ1wYbHMA0ixvT0EDv3/xx76++p546dfuyVTdT55Wu6lbV953vbHWed39L13Vq1KjRdNGyA92BRo0aTZ7aD79Roymk9sNv1GgKqf3wGzWaQmo//EaNppDaD79Roymk9sNv1GgK6VX98EspV5ZSNpdSflhK+fxr1alGjRq9vlReqQNPKWW5pAckXS7pMUk3S/pk13X3vnbda9So0etBM6/i3osl/bDruockqZTyFUkflTT2hz87O9utXr1a+/btkyT5ocP/y5bNMyHLly8fupfv9+7dO/gs2+G1lDL06sRnPCfb9Xv4f/Xq1UNt7Nq1S5K0e/fukXFku9mW959rVqxYIUmanZ2VJK1atWroXsYpSXv27Bl6PfTQQ4faeumllwbXMocrV66UJO3cuXOoL7ThY56ZmRn6LufHx0V/c769v3kP7dM3+st72vT2uIf5pm98zqskvfDCC5Kk7du3D/WBtnKP+LNph/WAWNO8z9/TBmP3e/LaHJfviUMOOaT6bPr94osvjvTDx//000/rueeeG934Qa/mh79e0qP2/jFJ78iLSimflfRZaX5Db9y4cdD52oZm069du1ZSP3AWfdu2bYN7aIcfIm1wj/+AeFb+uLjn2WefHfpe6if0vPPOk9Qv0KOPzg97y5YtQ21I/Q/RP/O2nnnmmcFnjGn9+vWSpBNOOEGSdPrppw/du2PHjsH/TzzxhCRp69atkqR3vetdQ3274447BtcedthhkqSTTjpJkvTjH/9YUr9JacMP2Te96U1D3zG3/JB8XMcff7yk/ofCGOkv1/rGPvLIIyVJhx9+uCTp7rvvHnpPm94ufWK+f/rTn0qSjjrqKEnSMcccM7jnBz/4gSTpu9/9rqRRQOG9j5ln85xjjz1WThyc3qd8z37lcE2wkPr5YY/xPN/T7DV+2OxT9sptt9021IYkHX300YOxfeELXxh5bo1ezQ+/dqqMyA1d110t6WpJWrNmTbdr167BD9ZPRSaFjcYPkYmdm5sbepX6H/Nzzz031B6T4j8YPvMfnj/3zDPPlNQvhn/HJuFHzYY79dRTGePgHq6pHVRSv0iSdMQRR0ga5UJ4zzz5mPkx/9zP/dzQd0899ZQk6ZxzzhlcS7/5oTz//POS+o3NOEBJ7zc/JjYy8+Lo4v9L/Q/9LW95i6R+LR3FnnzyyaGx5Y/6zW9+8+Ba5oE55MfEHPJ8R3cOuVtvvVVSv4+YixrnyD5hfvihs5Y+Tu7L9phD9itjl/q9l4cQAOJj/uEPfzg0L7RHn4477jgl8TtYu3ZtldOo0atR7j0m6QR7f7ykx19Fe40aNZoQvRrEv1nSqaWUkyT9RNInJP3qYjesXLlSGzZs0OOPz58Pjr6cWqAGiACrs2bNGknD6Ac6gSjc60gPcWpffPHFQ59zgoMijh4gDHJonr4pi/t3IBunPM+HVZd6JF63bt1Qe3ACPNfZ69Q7PP3005J6dhUW3Z8J6iUqPfbYYyPj4Fk/+tGPquOCpZV6hNywYcPQtaAWiOnrAWLx7CTu8X6xTxgjfWFcPj/MCxzcv/zLvwxdU5PBmafkGHkP11PrZ+pLWO9an1KHwDXMl9Tvb/ZPipupH5D6ddi1a9eIbmkcveIfftd1e0op/17SP0paLum/dV13zyttr1GjRpOjV4P46rruW5K+9Rr1pVGjRhOiV/XDf7k0OzurDRs2DFhZVyrBtsG2/OxnP5PUK0lgkZ1Fg/2H/UTB5coOCE15sp+wYTy/ZnpC0YVCBxEjWUOpZ99g31GgwdY72wgri6iCeMMc0Dfe+z2whPSb8TjbniY+NOb0n/l3thG2mvlHyXTiiSdK6pVwUi8msTYoDdNi4sSacC0KrjS3SfX5rV3rClSeTftpfktTo3/GHKYFA4Wq1O+5NDHyCrn2nz3MOGifvrp4STspTv7rv/6rpN4q4nub/bN169bq+GrUXHYbNZpCmiji7927V1u3bh0ghCMNihts2px4nIY/+clPJA1zCSAkJzJ2ahR4bhNOhw6QF8SEC+FVGlUapgMJbfgpy5g4qTn5MdlgbvL+JgcB6p588smSegWe1M8TKJfmKjdTMWbMjnAQ9913n6Qe1V2h9sgjjwz1n7FyrXMUtMezUdjR3ubNmyVJ55577uAe1p57vb/+uc8HeyIRmXl3tAWRN23aNNQeewXy5/Icxsq6s2aYTn2sKFHhxlIB6JwGKJ0OU6yhc7GMheek2fChhx5S0q/8yq9ImufGGP/+qCF+o0ZTSBNH/G3bto3IylKPcumyC8LV5Ea4AOREEBL5zrkDTmTQAvMdiFCT1TCpQMjVnKogkJ/YjAO0gANAznYkYEzpUcc99N9NmCAVKMK96BCYA28P7gnUS72Am9tS7mR+4KrcGee0006T1CP7PffMG3XQp9BHn8dx3njoQnysrBljxsQI90G/0adI0s033zx0DZ6NzAHjcV0Ca8K6guKsGfvM+8n+Yb4YB/Pm68z4kwtk3ziXkya/NP2l05Qkfetb8/r1ubm5qim7Rg3xGzWaQpoo4kOcXo7IEKdunnA1bSWnL2jEiQpCeMAK7SHDczKDgqAjaOj3QLSHHAzSuPyIBhcnJVxo4QRc24vsDpdB/9Pn3eU2kAz5MFHbET+DcngPFwVKud867dBvdzGWesuJj5Exn3XWWZJG59rRFQsJ/advONw4sT+YuzvvvFNSvxdwanFnILg/5oz1hAthb7ilhOdwL9ewHjWnGO4BtZknuE2X+X1PSf0+qjmAQfQhdTm07+vA/nnppZeqlpQaNcRv1GgKaaKIv2/fPr344ouD07B24qX2HfSrBUzwPzISaAWS+umX7p6cpKAVKOLoByeRrroZleYEahA4lOGsfg/95hQHpUB15gnORupl4NSGM3ZHJ+RovgNpQAjm1BEJlGZu0R1kcIrUzw/X8Bw+Z1x+D2NmbHAJ6ETgpnwst9xyi6SeO2A+0toi9cjOusKV0TesCj5/3M81+d6pFnEojcrvbvuHG2D+4RYy9NzbTysB7TMuv5d2ly1btmSX3Yb4jRpNIU0U8UspWrFixeBUc/RLxMpglBrip5wL0a5rS/F0SnsuJytteCw2z6x50Ek9Mrvtn3sSBWvaZDgIxjQu8UeNGHvGl7teIy0kmQgCrbujLDJ4xoODpMS5S6NWB+aOsFzWwbk4f5bUozcckmvoH3zwQUn9PIOccG+8evsgLfND/+HEaMv3XqJ4JjvxPZfWDuY/uSpfB/YHnnu5Hk7Zh+QK2K9u/fBnNhm/UaNGY6n98Bs1mkKaKKu/bNkyzc3NVZUb6TaJogt2DhbI2XrugZV6+OGHJfXsai14BhYPtp3Pk43071B+4aZJX3CeqeWhQ6TAeQP2F5a21j7zgnKJtFSeiuuMM86Q1LN8xLejtELE8PZxE0YMgU2ERfTUVfT7/vvvHxoPClPSg0m9Qg6RgbXBXZax+zrTf8YEu4vJzkWBdKBijdgbsPUeOISzTbLgPJdgowyqkXqWPrPq1JR8sPy81tyfIRTHiB3pQOVm7WTtuYc2WDu/DpFxbm6uKj7UqCF+o0ZTSBNF/K7rtGfPnsFpxSkpjTrdeAJBqT9Z/fQl80sqYTgl4QCk/qT0cEapP/lBencG4TvMYunkQ5vuxpoKQPoLpwGXIPXBM4w5s+qAgou5vKaTTM1xhAwvoDXmSO51l1S4KJCfuYWTcNMfiR9T0XXKKadIkt73vvdJGk5WSTvXXHPN0Hf0mzmRpAsvvFBSv64e4CT1ASuOmMnRwRExHtbXkTERPRNcevv0l/3CHmC/olR0hzM+Y5/QPu3WFHLpSs5za1wIHNdiDkFJDfEbNZpCmriM73n1HfFB9DR9JIr7qZYOFyAZCOlmngzEwPQEgmVyBKmXITMxBoic7qxSzxXgaIHMR/uOyKA2YbIgA6iYsqHUy/JwCcisIB2nv8/LvffOlzoALeB60Du4XoP2MwSWPrk+A/dg0IhXdCHXX3+9pGETKu3m3BHw4/3PsOvsUy0kFh0K+4h5QV/CfvLnsyeYL+7JnI7+zAy8YT/RV+fS6G/qFWopy6F0P4dowz9nz83NzU0ky26jRo3eoDRRxN+9e7e2bNky4morjcpmmWCilkGXUw/uIItyOFJmggmQi9MeFPd7kN9Aj7QMcMqTFknqUTU5FhxfPKlDBtxkhtZaYBL9A+UI1qhxH8jaIBbFJhhPOhf5mBh75q53RKbfONrAhbB2oJPnmKd99A1wPfTRNfTI4/Q307NBjqTpeMS68zlcoXMhWRgl05gtJjunVp+5dDfrzMCbtSJc35AyfGrpM/WaX7t3797mstuoUaPxdEAScYBWjgSZYICTDRdOkjC4PIQ8k3IQp6Gfuh66KPVoBTJzrZ+wifBpcSDxhGt9M0wzT3AP3kC2RyZG1ubUBvFBVEk6//zzJfUyd1oEnCNiTrG3X3HFFZJ6FEz/AalHMOT1dF/1xJaJYFnphjl1SwrryZhByhtuuEHSMJIlN5CJOUB+5wC8EpL3CaRHB+DpzDLYarF6hBnMle7WNTdlUofRf9pnL/ieZk0yGIcx12R8xv/4449X/RNq1BC/UaMppPbDb9RoCmni0XkrV66sliSChYLFyfhpWEBXeGW+ddhfXFA9cw1sdGZbgeiTu0LCTsHO4TCE0onPnX2n3xk59ta3vlXScNkwnHlQdNEObGjNNMP9sL/0N82hUq/AyjyDjCMrEfuYmG+uqcX7Mz88k7WD3WWtamwp42B9UUR6ZmTchnNO05nF1zkr9/IKi48y1Flx5olrEU1qZbgzB2HOBeTse0b0MQc1BV7m8kNxzBqx52u5IJpyr1GjRovSxDPwbN++vRr0wGeYizIuP0sMS6OulSBMzcEmHWlw5UzliTuzZPYcXFvTcchNQ5iruAY3U65FIehjQeGVnBDI426yDzzwwNDYQNvkTqRRMxLokQUwHSUSabJ2gKMrz6rl3Pcx+3pnDD8ZeOAAvv/97w+uZa0uvfRSSb0LNu2iNHOOhTGz3nANvHeX6bwHhyRXCmf/uRZOJV2oWVOfC+YQxE+lsQdWZZnyzL9Iu65wdIVfch7jqCF+o0ZTSBMP0iHvnlTPEY6ciByE7Md7R3yQC8TiPQ4ZflKDDimXZ44/dyAB5eBCOFnz5HYTFBwD7XMP5it35Ux3WzgK0ITQ1ZpLJzI+7YNS7gbN+JELmUNQgTZcb0K/4YiYL+7x+YITQV/C2tAec+GhqiAj63zttddK6mVwz4BETnyINczc/x6MBYcIZ8GzmVP65OHRZ599tqR+PRgP+8p1SYyRNeMa1sH3AsTaJPIzf47emYGX/rN2Nddpz2HZMvA0atRoLE1cqz8zMzPiziqNZrvlFAOVOB1rwQkgfQbIeLgpJ2aGRnItJ7nL2ZyqoEUG+riGHkI2S214LfNv6h047ekjY3UdAtwHn2V9P3ewYT4ymy/X8jznjEAUEIw1SldeSSPZkpkfrAbc4y7BzCnOS/QRTsBDb8eFZGd1W+9/um9DjCOr3UqjSV4YO3Ph/c8KTOlSDjlnhB4m8zvWON5cRzgL9spi69zCchs1arQoTTws95BDDhmgrJ94nFZUVOEUTlRZLGUSpz0cgOsDQO0MheQkRcPtSSOgzN+OZp10Ti6Xpnsv8iK2fw/xBAlIlAEHAddQQ2R/ljSaVsvdb5PbyOAQ5g153u/hlbHjMu1rRv/QwyA/w5XAZdUCb1gH7Pa1evb0lznLOnK4Hju60j/kau5hH2FB8XEwd1Tqyaq5nm4s/U0S+dm3zpkyl1yTegcfM//n2sFFMQ7XFfl+b1r9Ro0ajaX9In4p5QRJ/13SsZL2Sbq667q/KqUcKenvJG2Q9GNJ/67rup+Na0eaP0mfeeaZwWnp2lJO0syfzwnI564BTXmO96n1deI5meiyFoKZXlCgxyWXXCKpRyvvE88EedEz8NwMIpFG66QzZrzMfJ7gTEBt+ggyuOdh1uLLmoW12oWZUoox1+zHyLtZfwD9AzZzD4iBq0kNPWP2gJtEygwRTj2E9yW9B3PsNc86ODCex/y7pYT5YazME+3CUbhtHosOHER6nNY08cnxwinBTfk9rMlRRx1V1RvUaCmIv0fS73Zdd4akjZJ+s5RypqTPS7qu67pTJV238L5Ro0ZvANrvD7/rui1d19268P/zku6TtF7SRyV9eeGyL0v6xderk40aNXpt6WUp90opGyRdIOkmSeu6rtsizR8OpZRjFrlV0jwrOzs7O2D1PR4fFg2WLxWAsI/Ofme8NOwdbKSzW7BzPIcYda5J1l/q2SpYepQ8sHuwgq6coZ8o7HgubJ5n4IEV5rtxWXadfePaVMLheONZahkTDi70JZ1x3LWZ0tdQ5jx001aap7JwJM/xdU6CtUfR5aw+SscskgnVxA9EKdYB8x3sO+vtrsepMGYPpOgl9eMflwsCt1839TJ+RLoMEKsFhqU5OJ3HUKhK/Toee+yxVeV0jZas3CulHCrpHyT9dtd1z+3vervvs6WUTaWUTUtNEtCoUaPXl5aE+KWUFZr/0f9t13VfW/j4yVLKcQtof5ykp2r3dl13taSrJWnt2rXd6tWrB6d7Kjmk/sTOLDG1gI/MEpNZac8777zBtSAhaIqJKTOs+ulOthuu4cTOPG6uEIQLyCxDH/vYx4a+9/vgKOAACJBhPOeee+7gHhRQjAOlEkjp/QdJaB/zYzqouGkoFaMZJu1IBiqlQjZz1zkXlRmJsn13782sOXA7PKfmPMNakVmYPvJ5LfsTHA/zwHdwUY7IcHugNmOj39/73veG2vJ24EyYd7hO5xr4P/dy5op0bpZsydu3b3/tzHllfid9SdJ9Xdf9uX31DUlXLfx/laSvL+mJjRo1OuC0FMR/p6RPSbqrlHL7wmd/IOlPJP19KeXTkh6R9Mv7a4ice5yOHiiRTh8p83PqugmK/5EpQUHQw91LOfGR/5G1QY+UV6XeDJO1zriWU9/vyQCkK6+8UlJvjvGAGOQ0as2hFwAZQHecmqTetASS8ZwMwPF5wfmG+UjXTpfx6VNmgAXBXJ7OkN2sich7R1c4CeaU+aLfnp8PeZYxJ4cHd+jVj3g2Y4c7IxCnlpsw9T/MAevv88WeRW+SefVZQ9DcKcNx4WC8NHiWcc9MyzVEX2oufaf9/vC7rvu/ksZV4vvAy35io0aNDjhN3GV3dnZ25FSTes0tCAZacNrjmuqppVLTCSJw6lJbTRp1GAElQI1ahRU+Q/ZG482pzzhqJy4JJriWvp5zzjmDa8gse8sttwy1AxK4jgLKgBvQFL2A5/gH1XD35DVTQbnSFcSkDxn67BwXCJaWiwz39TkF0ZFReeU57iwDR4VczfPSocf3BJwEFhgScYD8tOVImpmE77rrrqFx+bWMDQ4SzjHTgLn+ij2Q+4Rrao5sjCP1GjViH+7YsaOl3mrUqNF4mnhe/e3btw9OPj8B0eDmKYuMCep6aiy0pZn66YILLpA0bBMGUbAooA+4+eabh+51+y6nOJp4EAE5i2tdLsUSQKANiMy1oInUh6YiK9NH7Ne0VTvtkT+zoq8H8WQ/QQaeV3MVRc5NXwue5zJmhjiD3vkcDyFN9Etttfcpg1hS35OhtlKPlMxDJttkjl1uZ+4YI+0hizuXAyfhdnQfYy18mb3L3oZ7Zcyub2C+s3ZehrL7nmCeN2/eXE1rV6OG+I0aTSFNPNnmiy++OEBIl/VBGlAjg13clgpxqua1GdYq9Sdm2nVBiEzUKY3amjNcE/upoyy6hAwQAr3caytTVnO6p97BOaP8DO1yei1K/ZymxSJTcvvcgnqZ7LQWFp3ps8fVAHStfoa18uzsmxPryJpl6jC/J+Vo9gLz7yHCEBwW88ReYf6cy8zALHRF7D30Gc4ZwW2wXxlHJp2R+jlL60Ouh4/ZK/80Gb9Ro0Zjqf3wGzWaQipLZQ1eC1q1alV3wgknDFj8D37wg4PvYKFgdWD5MfPB+jjbBbsFG5eKDXeiQLwYl9MMBZ671MI25rNRECEmuPIKlg/TXwaweBBMOo4Q4PFLv/RLknp2sZY5F1Yv8wDWAj6SvaUN2HpXRCVLmQ5CrlTimbwyRlhb5sLdV8leBJvOtTWTV2bpxckrS0X7nkhXZpRwrFEqar3/uVZQTZGWgVWIZ4hyPueYUTOPPutaq9STFXVSKel99KpE3/nOd/Tss8+O87sZUEP8Ro2mkCaK+LOzs926desGp+VnPvOZwXecxFlemlMSBHJFCCd9lk3O6jtSf0JzTZaXhrPw4AdQAhThOZy+t99++1BfpdHqNRl04YiMEunnf/7nJfWBRKBfZmGRRkNIE528phpIkpl4aS9Dbv0auKh0tPFr6V+WJcehJrPg+PhvvfVWSdIdd9wx1L6jH/+zzuyNdD32UGHWJusb4mgDOjoXxV7AzJYmM1cSZ/075oA9x3PdaQkFI9+x7vTFg5jgeOBWqbkIR8He8D3h+/Paa6/VM8880xC/UaNGozRRc97q1at1zjnnDBxT3JyHY0Sa8zJnuMtmnJScoJzGnNQevEHuOxAhuQVOUnefxEEIF+DMjc/z3VwF0mfOwFotgY985COSeqQBATjNs0KqNCqvM44alwN5qK40mm/f5zRz7GEeTHfoWrtQOg45l4Mc/Y53vENSj7zMpfdlXM5AqKZ3gBvAxIrMvXnzZkn9mi6W6z8dzHx9k3vKcWXyEKlfz8wrWHNdhzviO9zOmUPmyc3O9HP58uXNnNeoUaPxNFHEX7Nmjd73vvcN0NzlRRA55bfMwuoBGZx6nKggTLp4evto9/Na0NZDJDk9QSeCaR588MGhtmqOSJnrH6S47LLLBtcix919992SRuV35DqXR7OWHRrzdGrxsdEuyJXZXtE6S6N1BVJP4O6xaSEByegv13qfQEbugeuhLefS0nEKZMuKsl4BN+slEBQFUhKi7HoH5vntb3+7pF5nw3N8zOg+Mo9+pkBzB568B50Bfa1VwMm6BgQb8fuopUBbs2ZN1QGqRg3xGzWaQpp47bxly5YNTlsPIc3TD9mVkzprukn9qQiycU9WdJF67SjID0qADJz6LiNxeoI06CayuozbnvkOpMStd+PGjZKGtbG33Xbb0LUZKFRLXcWz0upRS42V+hH6ku6gmWBTGpXfMwmGXwPnAopnFV33E0DbTd/Qrbz3ve+VJF1//fWDa1mjTK0Gxwh34IhJn+gvXAh9IKmJrwOU1ZbgXJw7OO2004bapW/oFNiftOXzAsGp1Or4jat2mynRavfs27evyfiNGjUaT+2H36jRFNLE4/G3bds2xLpCmD9gPzFjwJLBSqE8kcaX1kYscFdO/kc5SB9gmWpsNaJCRu6RXQd2zyPxeA6sPeYdFHhOiBnplonSEBa8Vi4p2UfG7vPD/7CWmAKTXfTMtumUQ98QLdyMxFi5hrHCwjI/PqeIJtxLv3GwceUnzj2IVIgxzAfz5BGJzBnKsHFZgZzSXJiOQ+4KTjQkOR8YOyIkjmh+D5mU0gyc5b/9mhQzTzzxREmjZcukYRfg2vhq1BC/UaMppIkj/vPPPz+Sz00aNc1xonmJYmkY/UAATmYQjuATd3aBYwDRPIbZn+uKqIzhT7roooskDStysooMqAU6ecAKfcjAG67JfH3SeKSnjzUuh/4xd1kmeykE0rhpEeQC4TMTEmvn6MSaZDAQ8wSSeru0861vfWuofUyvnuOA9pmPVO5l1iDvH+uQee+8/5deeunQc1BApvnt8ssvH9yTmYUZD2ZgN82x1jhKsQfgGNmTbnbGHDszM/OaFs1s1KjR/2c08Sy7q1atGqCWBz+Mc//klEynCmnUeSIDZDzgBtmeExFuI81i3n4GaYDInNToJRw94DaydDdjrrmKZjALKMV7nyf6l/Io5shaEA39drT2vrgJKDPt8B3o6/0H9bg2nX8ym4zU6024h7HVSpvT7+RQcp6Yc2l8nT7GkWWzpdHy6vSX917vMPUMvNIH5Pma00/WEnTHJojfAWHqcH+sHfe6OZLvVq9eveQc+w3xGzWaQpoo4u/evVtPPvnkQC5yV1H+zxxmnOq1gJJ08wTV/bSFQNzM0U67WcPNKZMtpFbWZUCcZDipGUfWyfM+5MmfXI6f7owNy0IGHzkxlgxFhmoZZ/nfx+R98vyCmfQi3ZSRT13uBD2Ryz2k2fvq/UuOLh1snEvIBCJwDakbqeUBTNfZD33oQ5KGXYKZh3w2gVY4ihHwI/XzzrVZQ8+tEsnFppMPc06dAKmfsxUrVrQsu40aNRpPE0X8nTt36uGHHx4ggsuWoDZIiRY88677PZnNldBLTl1sn1J/KmYuc05U5EhHhgyq4JSHa6CvjhS0D/LTLjKg29mxT4NsGeyCHOwBH3BGuDunDbrmUgvHwlgzR76POTXQIH9Nn4G8DnIxtgxNrsmjcD58h33crThwA2iw6QttoGupuVlndtpM0+VcSPqV4EvAuJyDZN7hLqlkzFzgR+Bt8j/ryhwSQOTry5pnGHomOXF9je/DZsdv1KjRWJoo4i9fvlxr166t1nLn5Od0zyAIkMBPX+ScTF7ACUgqK2lUG8qzOdXxbnMk4BoQEjTEQw1ylOV+NLnoBdDcOxJw0megTXptvfvd7x7cAwpmMEbNyyzz6SfagdDuW8C8o7FPz0ZHZNYKTXdaLrIOohMICWeE7Op9Sc9JXtP3whOVZKqzrInIOnv4KuiKjR7UZcweTPaxj31MUu8ZCKcC4bnnegHa4dmMmb3neg24mrTaMC+ZAFbq9S6PP/54Q/xGjRqNp/bDb9RoCmmirH7XddqxY8dAmeWKLlgUWExYYlgaWCfPIpsOHpn5tKYSDnJnAAAgAElEQVTo4h7MSbB5KE3cLEZ7KOFQGkJZtNOfQ/u8v/DCCyUNOxVlv2EJMYPBSrvpKV1E6QNiiLsPZ554FKjMD33xOYV95h6eTdYhJ0xziDPMHaxyxs1LvUIrsw2l+CSN5mhA4Zt5DN1phXZhkVEi4gqMuOPsNWvDPDCHiC5kYJKk6667TlIvbtB+llB3cyjiEetxzz33DPXfM/Jm4BmvmQ/Bg6XowwknnDCUE3ExaojfqNEU0sQRv+u6gZLDc9VhrgAlOLmzEKYrtTLoAUULJ6gr0jhtQSfKVaMsxGTnyj36QD9TCYfyyc0xyZlQySXDgXMs3i7PPeOMMyQNB6Fw4nMtip0s0ukEiqfSEOcfV47hTJQotWnTppG+nH766ZJ6riMVaihFnSPKEtrpIuxzyTqm8xVcIO074tMH5vvss8+W1DvYwGU5Z5SZlukTnAAoL/XIztqwb7IiEAFcUs/xwDWxZrw6StN/nHoYK4gPt+O/Hf5fu3Ztc9lt1KjReFoy4pdSlkvaJOknXdd9pJRykqSvSDpS0q2SPtV1XT1Com9Dy5cvH6C7m3lAP05BTmROvJoZBgRI11BOe5ezeBYIBiKjU0BOcsQHyUCADCihz575F7MLyE67tfxtoHeau3DSyQpBUo+M9J/XrDnon6Vugvp1cCw1N+gMfc38gH6t6wh8PHAUzo1k8g84DK7xuWSt+Yz5ANVqMj59Yd4x6bLumNLom9TPKX0BVeEK3T02E5WwB0H1rNcg9Vwg68C4MAnWXKZ5zXoMWRFKGg4iez3MeZ+TdJ+9/1NJf9F13amSfibp0y+jrUaNGh1AWhLil1KOl/RvJf0XSf+xzB+J75f0qwuXfFnSH0n64mLtrFq1SmeccUY1wQEyHzJ4OjJwKoPu0qjLLHJXyvx+PyclWlHQAjmSirXSKHfAKc4rbbozBZpnUAKkp33X0CObZRVYOA2udUSgL5zyIAKci88p/2flXtCDzLbf+MY3BveA8KwRfQHJ3IEqnXvggFgXuA9HIeYq3W251ucyszET+MKcMr5aqDbzkmnMmCcPjKIvtAf3ASLX+k/7jJ19xP7FEiT1Fgvm9qabbhq6x9v3/S2NOpYh63vCGPo9Ozv7mufV/0tJvyeJHh4l6dmu6+ARH5O0vnZjKeWzpZRNpZRNGR3WqFGjA0P7RfxSykckPdV13S2llPfyceXSakLvruuulnS1JB1zzDHd7t27ByjiwRuZ6z2TRKSrrf/v8pR/7m6TJEjgJM5kCOSGdw13Jv/gZOWkpq8uo4POmf+ez2sutXkv/Yfb8foAtZrzUj9fyLBSz2XQDnIv8ug111wjaRhxQFOCT7JvHkqNNtqfKUl33nmnpJ67qiUABYmZ26ywK/VzxXPgOuhvJhaRevmZ9jNkuKaXgPtj7bEWwAnUknakOzFIXAsfT38P9hGvrqGHKyB/P9xOuv267Z/7d+/eveS8+kth9d8p6RdKKR+WtErSWs1zAIeXUmYWUP94SaNVGRo1anRQ0n5Z/a7rvtB13fFd122Q9AlJ/9x13a9J+rakjy9cdpWkr79uvWzUqNFrSq/Ggef3JX2llPLHkm6T9KX93dB1nXbt2jVgcdyJAhYH1g8WCmUHrI4rx/gMZQxOD15GCoL1gxXKeOdUOnm7iBJZ+jqzpUi90o120jHIr80MOVDm/3OFTeady6zBrhxDHMhCm6nk86w6KPUQfdI85Fl8cUhBqQrLnHnvaoU8EYXoE5/78+gnCjnWjPbYG7USZsyZi25Svw9qJtI0+Wa5bP+f52TGH8hdv9kDaY6smd4yIxT3+rxLwyZSL9S6VAeel/XD77rueknXL/z/kKSLX879jRo1Ojho4i67u3fvHnHKkXqkyUy5nHgopNx0g9mOqjUoZXCxfM973jO4FgSGy+BkJNAEpyLP2Io5MLPb0AdMRI4eoF4qq1Ds1LKt0CeUQKAHc+KolU5KWYLZlUo8O5WGWSDUOSTmAcUj845JymsVZLFP+gISw4F5rnzWPgOpsuy0NBqQlKZe+u1oi1IMV2nMqiB9jZhLEJ+14r2jN/uIeaEvXFOrb0A7memHufC8EawJTlbJkWbJc++TK1H3R81lt1GjKaSJZ+A57LDDRuR4qUdKzEWcpJzGBFk4gWiYe2699VZJvWukmzZuueUWSf3pmOYpMqe4jM8zQbnMnwdKuTmPE5rTNx1K3BkHvQbmL7gbzG7Ist4+KArCkGcw8wBKPXpzT8qLcCo+tzz7O9/5jpyYJ/oqSVdccYWkHpmR+VlXkB4UdsowbMjf33vvvZL6OQPxQe8MlPHvUm7PWg7OJeCUxDylc5HrKGg/nZPYG3BK3ifayQxOcGCE6Uqj+gXGSN/Y6673cWeepVJD/EaNppAmivgQJ50HZICinGiEU3IawxG4DIsLJ7I4KItzgyeA4ATmJP34x+ctkekg5PI6SIjsmlpf+gS3II0GWXAa81qr1EO/6SOyIAjn4+CkR7/Bd/TJkYZ2QIusf0ef3BGGOU0tMtyUy+tcQwAP6AdK8eroxNqzZrfffruk3rLgzyXXIPPA3LFWIKjvCTiGDIACtdlzzg2yx/iMvYccTeXmfJbU6zMYY62mA/Oc9RNrmZHpJ67j6CoYK32Cm5P6vbZ169YRZ7Zx1BC/UaMppLJUF7/XgtavX9/9xm/8RrVyLAT6gAygKidZVjuRelkfmRn5y9NFcWJeeeWVkvrTF6REQ1w7sQnWyMQQIIEHVnBig+agB0hTCyjJ4BD6z71uaaCfID7a3n/6p3+SNIyYaSnJzK01xMna8LQP8rg8yfrdcMMNQ+P45Cc/KWm0NqDUrydIzzjQLdRSRyUXlX4OPj+Zlx5dSCb+qNULZMzoY7JSk9TvKfrNvZk73y0Bme2Y73ieu98y32nlgBNjjj3Umrl8/PHHde+99+qFF16oF6I0aojfqNEU0kQR/9hjj+0+9alPVb3YQESQHlmN07aWWooTExkWv4Cbb755pH24gfPPP19Sj6p5GnvSR053rqUNkI7vHTHhJJDF6CPyLvKqNKp5zvfI/o4emZzDA3ik4SSMrC1zmGjOfDmKM5fI68wLbYAuUq/9BrGoNYf8iQ3d/QRASMaRqdbc8zCrCGe+/hynNKpV557kplwWzrRWzClr6VxgellCpCSrJVzJ57BvmAsfD/9nxV44AObaOSP239FHH60bbrhBzz77bEP8Ro0ajVL74TdqNIU0UVb/yCOP7C677LKq6SbL+8LypCLHCYVKlpXObChS70QCCw4bmvn2nXBWob8ZVAN75w4esHiwX8nq1/KrJXEvSkQCZqSevYadhrWHPXW3UMZKTDfiE+OAbXS2NPO553g8ww/9x2UaJWeaJd2BBzGJ9USRxvw7K50lxnIPcE8tRwNEXzKjk+835jnZ6jTJSr0omBmRWQ9MuzhW+f2IUcniuxjLHLqJ2OegVn4OEWLfvn268cYbtW3btsbqN2rUaJQOiAMPp74rcrJEdKIr72vohGIwM8O6gxCZd1DuYD5KpK+5T4KyIDv3JIJKGgmL5H1WmfH+whXAfTEXPMeLNuI6iwkL1KYPzh3wLFCKzLI4z8At1DgPUCqz4rqikeASUJW+kZ0Wc2QtI2ya21Ae+npk9qIsiAnVsgQz7yjoeC5I6YgJ18HeSHObj5l+Zt571gMu07Mds08YD5wYfXPOlz3gvw2n2uf05ZBDDhlSvi5GDfEbNZpCmijiz87OasOGDYPT0MNNM/stKMvpmK6SUu9EgQyeoam1aimcrnADyFc8z9Epk0UkBwDVcuVzL7Ia5hc3vyXSIseBMLU8fZmcI3UjHtADqoHWIENWk3E9T4ai0ieucb1JBodgtuN5tOXoN86cB9UqGfHsce6oziWkyY/+855rPd9hJlqBO2AOaqa/LAme9Q0cxdnTyO9cw9jd3Jn6jEwGkiZZaTgJS6uk06hRo7E08bDcQw89dHDiuTYcWTVdLDl9QdXaiZahnbTl8lAmuQCJOc1rzjicqmjvU/ZOGd3H5NVXpB7xPQQWtM46A3AwtO+a9HTy4bVmleCzTGtOf0FvdzmGC+CZIBDz588ByVkT7oWrIVGJo3q6AoOGjMsdkNKVOZNcoM9ITb406m7LPZmUxJ8DZ5Fcpu8JOMLkQpgfOElPinHmmWdK6sOM2XvsFefa4Cazcg77n7l1XQ7zvHv37hEOcBw1xG/UaApp4lr9mZmZwQlbC2d1TbzUn6y15B1wByAC12JTddmVEz6RErTNBJRSj2how2kD9MjEi35PhvBSF97bBz1oB6TB6oF1wsfMPenayVhr3EfarjOlmLukEvwBkoFANf0G6AxHATeDhpu59fbRZGcqsqyl5/dnUhYIDsPnJ6vugpDsEdr0dcj2eZ998zHDSaSMn3UEpX4/4i5O0FctdRh6L16ZD65lnT0wiX0yMzMzklx0HDXEb9RoCqn98Bs1mkKaKKu/Y8cO3X333UN5wCHYqozhzkgsJ1jvjOxKc5s06kYKq5QsrGcvhS1Np48soeXPy6KHsF4oe9wcBruLeMCYYddrSlBYVxSNjCedf5xQyNE+40BB5E5FaZJLt9JayWjKk5F7j2tZD2eVk9VmPvgcRZXUiwiIdIg5KH5zXH4t1yAOZP57V1Jm0c/MTuzKScQB1pW155XP3VzI3uM5ODrxHGLt/RqejRhDmTJMsi5euqtxM+c1atRoLE08r/7evXuHTDYQJyROIZxcmbPOT3dOuiyxnbHXUo9qfAYScMKCSo74oDcogVkGc2EteIT2Uf6ABJgE3WmnlpNO6pGUvtQUaulYk9VynEA7UDWzu7gjDvOSpiwUnNQhkPqsPHzHOtA3zJKOTsxzmjJzrp1SOUm/QVJfs6zik5+nI4zUc3ZwU6nodY4FQmGXyrdaUc7MEQi3CRfIPPo19AEOA2c15qvG2e3YsaPK7daoIX6jRlNIE0X8ZcuWaW5ubpD5xd1LcaDJMFMcXzBZ1HKleZZRb8NNHiAKJz8nKic2spk7u/D/hz/8YUn9yZ+hpS5XERDDGJE50yHG+5mZfqFEc6lHaxAmkczRiftqVXz8ew/lRcZOxKddL5PNfcwTKJQZhd3ExLzTp9RVOHfAGNPRhX4TJOQcF+2gX0i323QJ9+fwGXsPTsmdYtKMCmpnNuiaWQ1uMKsHgeZSP3f8NtBrsM701R2E2HOzs7MtSKdRo0bj6YBU0gE9PL8dWssMmCCDK265ruFGpudUv/HGGyX1jkHu2JGVbJA/Cc9FI+3yNtr1rFKTGmLnEvguuZG0Kkg9UqXsSp84/X0ctIdGPnP8e2BSZpzlFbSCw/Ac7RAZhBl7cg9+fwY60e8M1vHxw42xrhnAIvVrRjvIwsmpeNisW02knmuiDcbjXAJcB30BVdmffm0Gi/EdfcpKPtKoDidz/rP3vJ/okZDx2Xusu3MJrPkDDzww4p49jhriN2o0hTRRxJ+ZmdGb3vSmAUI4eqSrIycXrq7f//73JQ3byUlQwWeuxfc2pB6VOH3hCjhtkZFd605tOPqZfQRtXTbPAB7QBBmtlnoLhIdboK/I287lZOhx5sp3l104iAxIyvrsbqdGNwHaMrcgmGuTeRbzkGjImB3xeRYcSmY79vk58cQTJfWWhNRoMxcejAX61SxHUi+/O5eQVYRBfl5rdRtT+46Mn7X1nLiGeUd+9zVjDhkrui3WDE71jjvuGNzDPl+/fn3VZbhGDfEbNZpCmiji79u3T9u3b68mVgAdOJH5DhknZTdpVFOf3mau4c4gh5SzuAfEk/oTmudwb1ZcdZTKBJG8R2Zz9E5tfqZmAmX9FAdhsJ2nXbqWRos5THm9Jg+CZFwDOjl3BqFZpp9pw2YcrqNgrOgDuBfUdfSmv8khpqWhFjabVWfpA2vp60A7GQadwV8+LxDrm8FBNa1+zmkt7De9COkDawiX4xwN871z585qPv8aNcRv1GgKaUk//FLK4aWUr5ZS7i+l3FdKuaSUcmQp5dpSyoMLr0fsv6VGjRodDLRUVv+vJF3Tdd3HSymzkg6R9AeSruu67k9KKZ+X9HlJv79YI6WUIdNGLUgHFjbZLXKzY/aTepMQ8eWwObDrnqWE/8m2i8MQLCfKlNNPP31wz7hMLOkq7CxzKnBQrKGUcTNPOrEkMT9uokvnD/pYy0IDZRluWEPEj1oJpzTR1VxRWaN0JqJviC4u0qH0Yu34jntc1Eq2Nk2lmNtcDEklarqwZrlpqVf0ZeBQ5saTepaba7MuBX110ZR5TyXxYm7WiDnME/d4Ac/s/5YtW6r1J2q0X8QvpayV9G5JX5Kkrut2dV33rKSPSvrywmVflvSLS3pio0aNDjgtBfFPlvS0pL8ppZwn6RZJn5O0ruu6LZLUdd2WUsoxi7QhaR4Fd+zYMZI5VOpPc8IOs5hlzWSUyh3QhFO+ljsdlMNphXYxnblTESc0/UxlXK0MNOjAKQ9CoEzy0uBwBRnswytI7aahnI8MGHKOCiRBYQaKwGGAJjU3aDgXzElwLDj2SKNZhphj3uPI4y6vjJl+8orpzhWyGToNN5B1FLx9PoMLYczMP/PkgUmMlX3Ed5jmaoEvGTLMnmCvuPINZGc/sve4p1YVh36mWzTjcs6Xtdm+fftras6bkfQ2SV/suu4CSS9onq1fEpVSPltK2VRK2bTUyKFGjRq9vrQUxH9M0mNd19208P6rmv/hP1lKOW4B7Y+T9FTt5q7rrpZ0tSStX7++O+ywwwZIVMttDhohI3FSg4IuT4PIXAtqY7JBnvdncTrSLp8j8/mJmXnn0kEF9HLEzDBfELMWhILpKZOQgCbkZnNEA2XpZ5rfXLZMhx36BOJkjT5pFGUziMnNbdk+12Ym25oOAeIexux5GJl3kJf+Z2ivO3Vlnj7Mw8jGcEiey5459fz0Us8tOHrzLLiONOfVdAuZYIV+u2MTlDqbdDhjrl3vwP+PPvro2LDkpP0iftd1T0h6tJRC5cMPSLpX0jckXbXw2VWSvr6kJzZq1OiA01K1+v9B0t8uaPQfkvTrmj80/r6U8mlJj0j65f01smfPnqGTtkaZJCIdSBwxODlBcVIYca/LxpktFl0CcjSo4idmpnpK7Wutyg/95OQHKUGNWp03XkEN9AC1wJXUxNM+nIXLrukIlA5C3OPVbJHhWadxmnSpn1PQmtdMb+VORbgns648J2sX+FjgClILzqu732bmY/ZAVqj1ceCam9VxmSdfd/ZJyvaMp6ahZ48xtkzPRpveB8aMRYc+Mcce7AUX/Nxzz42tNpS0pB9+13W3S7qo8tUHlvSURo0aHVQ0UZfdPXv26Omnn67mIAcdONlAgqxM6zIs/2P7B/mR7UE0qUdGTtDU1CN/eQ74cYkLU0fhuopxLq+gresD/KSWRhN+Zu11qdd+Z4AHSMFc+Fhol8QV6fdQS/TB2EH8WmXdtHJkAo5arQLQ7eSTT5bUrzOcUs29l3Yy4CnDf/1ZqaNg3mq2+ayIxPrWqhTBcaXfRAZPufzOOjBG1pt+16xbyQ2yB/gdeI0B5vC5555rLruNGjUaT+2H36jRFNIBicdHMeJKK1gclDyw7YgDsDZu2oI1g83iHth5ZzFx/oDNIg6fa9/1rndJqmdOSfaRfmekmdSLFJk7rVbWOrMBZ2w98+R57ugfDhzJGvq16VjDc1CwYeLCbOjPhLVM5yJXKmH6g1XOdU1XWJ8XxDDWcNOmTZKGHZzGzTtUU+6xBzIbECw/okathBYsN2Zh2nUTMu1k6TWeV4siZZ5gyVOE9P6nuzNjztz8tbJes7Ozi7puOzXEb9RoCmnilXTuu+++kUKD0uiJzCkJOoGuKLWkXmmCwgXlT+ajk3qlEd+BdnyeASfefp6+ecrXsrCOQ29XKqW5LmPr0x1X6lGOZ9J/+lIL4jj33HMl9dyNI4yPU+o5o4xjzwovUm+mAnGy3eyzNFr9CIR897vfLann+KQe/UFXlJLsEQp8uqKLOeMeOKBUFvs4UBYyDuY/naWkHnnT+So5L+cc2ct8l3n8XeHLb4N7klurVaHyPbVUaojfqNEU0sRl/KOPPnpgcnLZiVMbGQ305mRGJvSTGjmdCjecqJmTX+rRlRM1yxCn6Usazdab2VJ9XBDICPLSl5qOAlTNdkELuJOajJxIw/M8swxyOnI07++66y5JPdr68xkz14Li9MnRhb44F+bEejii0Q6oB6ryHOrwST1q85olyHn1PcH6sX82b94sqUftzH4j9XOJUw+cEZ/j7OXjz34zFzW9Bs9Ml+Y0nfozmVP2Kf1Nt1+p37svvfRSq53XqFGj8TTxvPpr1qwZ5AR3eY4TLBEhZRtHj3HVU0F1d2ZIR4h0Ca7lQ083zEToWp23dNzhHpDAg1wy2ARaynNSq1zT9vJZJqcYV3ev1pd0CKmF/WZSikQd1zRzDc9hDWsJJNA3gGigHeudlXf9WXCO3MO1ILWvA2tDnzKkupYlOCskZWIU5ygyH2Kup89X5i/M3H2Mz38HbnlpWv1GjRqNpYki/q5du/Too48OTmGvvIpsiU2ZfPpoN7kWra3UcwPoDEAIbNy4N/q1nPwZbsqp7mmcMmMuxMmN/O6oyDiQNUEIUMoRGYsFfeJ5KQvWXGozDRgci4e1Mrb0c0AbT59cq89YQWL6xrWeAz6DZtC5oDsAlVzrnjUKQbvMAOzPZn7TxRjUdUTOcFaIuc5MvVJfn4Ex03+fSwhOgX3Ic9ApZL59HzP30n9ea4lcMlsy18IReDAW4z/66KOr3GGNGuI3ajSFNPFquatWrdIjjzwiqQ+ckEYTZKTN+Z577pE0nAyTUxFkQPs6Dqml/kRFg5uoWwveAAmyHgCnvduvM5gi8+s70oByVGpJ7XjKp/6sRI0MJfX7QSeQPSvRuHUlK8MkgtYsGzyHMWNNoa94wknSWWedNXQPfclEGU6sA4ifdQh83ugv3/EcrDqsh48ZLixTcHGvJwBlfvExSL0A6O5cYPabazP5qVPqfzLBqweGeeDRUsNyG+I3ajSF1H74jRpNIU08Hn/r1q0DpVgt5x6KIFimVJJ5jnlYSVhw2CGucRPROEcIWGT64g4wsIeZrTZz/tcy/6Jsy0ywzorBNhMnD8uKsinZSKlnpxFVUsxxsQMxAEcgsuvA0iJy8XzvX+YZ5B5fMwjnKjIgZekvN1dRz4B5RjSCHfZ8Alm+m7GmGdcVWnkPc5wOVOxBf3aW0OJa33O4/maJLvYR61Mr95358NgjblrM8mCMkWvYC7XcjVI9A1CNGuI3ajSFNFHEX7FihY499thBMISfTgRTcOJzImPOwwTlii4cgRJVa84gnLoZCgvRhjtGuBlKGs1Zxylcc2rJgJ7Mu+bP4uQHNdLRplYGmvZ4DgjqJZ1Bh3vvvVdSP3e0i3LVx5wZZBbrf6IrzwbVsx9Sj2RZZ4D3znFlcVK4mgwV9n2UuQ1BYOYwcyF6e7wyxhrHlW7hcFiYCdlfzuWkaS6du5xjScTPbEwZpuv9nJmZGansM44a4jdqNIU0UcTfvXu3fvrTnw7kUw8hBdkTRTjZQEN3IEHecgcUadTUJY2WqU7nk0QRqecKQAtOdxA68+F7O+lKW3PG4b5M1kFfM0Os/3/nnXdK6kNua8kqQC76CVeV4cY1l1ruZcysjyMN7XEtawfiMw43bV1xxRVDfUoXYTdtsRY82+V/aTRHXm2szCH6hww3rn3G+1rSDoh5Yc2yHp4jPv3kmixx7uubQUC0z9xmnn2/f9myZc1lt1GjRuNpooi/d+9ebd26daCp9Gq2nOopt3FacnK7ZpQTE1kfuQgtpzvjZLBDVkDh1WVM+gRlGCUnrSMOn51yyilD7WaCBWk0AUdmrV0stRTcEhmFQTh3gOF/5iGtBbVgKZ4Jh8WYQUPnrphn5H8oMxYzF1KPZK4pl0YDWbwd5OmUvdkTbmlgLrMi8CWXXCKpR35H8XTBTtnb9xx6knTmyr3giJ8cQ+plfH3TaYt9A5fAuniKNawzmZpsMWqI36jRFNLEE3GsW7duKBAGSnkX4gSsnWogSmpNQX5PqpGnLEjMaQsCnX322YN7sp4efcx0VC5X0U4G08BxODqCmCmX0Sf66roQEAAEI6gp3WSlUdly48aNknrkJKjG5zTt9Iu57CZKcy1BLukC68+kn+h70B34WqZW3/U7Tl7bjjVL3QFIef7550sa5gZTz5O1+rz/rC97gdf0S/C9V5P7pXq13ExnBseSFgYP9rr44osHz2R+90cN8Rs1mkKaeCKOubm5wanm8igolEkqM4DB5a30qOM05CR19Mh68pkKuhZEw7PhUDLxRC25ZCaayOSSTnAOcAO0DxpxqvscZBKKTFBSQ2+Q9+677x66FyJQSernML3MQDDvC/NOv+lLInMtGaT7Jkij6bT8Gu5PK0r6BHj/0KXwHetB33ye0v8Dj820ofv9icD0NX0/pFErE8ifHojSqPWHPZZVeN264tV9x6WHS2qI36jRFFL74TdqNIU0UVZ/bm5Ol1xyyUBBQvUUadShhvewyrBsrjTJPGfJwroyJUs6wz6m0s3NbVkoMhU62cda/2HjULq4ci+z2mSWFcblTk2Z0y2r2LhpMQtE4h7LvNSy7DIW2F3EkcxjIPUKM0Qh5iXdVzFB+j2M2RVz0jCrn27PWUiSPrpIx1gyr0Oa23xO6Qv9zdoELt4wFsaRZsnMbiSNlhrnlfXw+WEO0yktTciuvPTqO61oZqNGjcbSRBH/xRdf1G233TYwQWFakXqzFCjEKQxSgsy12nmc4rxijqmFy2YIL+9BWVeaQBkolFleXPmTpi2+Q5Hj/eekz9p5GZLpKE57mTuu5jiSGVmzXl1mMPL/MzSVe115lVmNUzFL0I7fwzwwp+na7ESADWOiT7l23v9UotLHVIp6KCv9RclJnQaudQendOBJbo3nuMI0M/ymwtFNcKkYhZPjd43i4oYAABTaSURBVIETlnMscCrpSLUYNcRv1GgKaeJBOk888cTA+cRPugy44VRPc4kjGtl0UxasmTQ4ZZGBQIY0mXnYKddk1VnuqYWFjjMPZm01J8ZWq1YjDaMTcm1mCYbLqWXMzfmgb8jXtCn1CMN68B73XkfmRNd062WsOF85EWTCcxiHIxlhw+Oq1MBJ+JyCrnAU6HbGBVxJ/dwxH+kq7BltcZQat18xF3v1He6nD7mPnGNJUy7Ec8g27fewJtu2bRsxw46jhviNGk0hLQnxSym/I+kzkjpJd0n6dUnHSfqKpCMl3SrpU13XLXrczM3N6aKLLtIdd9whqa9+KvWn4Nvf/nZJ0gMPPCCp12JzyruzBp8h92SNO05y/wyU4oTOBB9+0iLboeXlO9CPe2vVbHNcIL7n+ueZyIN52oNk7ugEB5RZXhPZ/JkgDnLiuCQhUq87QKYHaWoJHtKqAgpmuignkJ5+pmzve4K5ylDqrKxc09DT/wxNTicdqV8HOAy4hVqaMfRSOEPBJTBPtO+pvdDlZCVlyC1VaUVhzehjVgiSesvCSy+99Npp9Usp6yX9lqSLuq47W9JySZ+Q9KeS/qLrulMl/UzSp5f0xEaNGh1wWqqMPyNpdSllt6RDJG2R9H5Jv7rw/Zcl/ZGkLy7WyM6dO/XQQw8NQgz95CPQACQG8UE/UMvRCUTjNMxQUr82k11mYkVOd0cgTl1CIEEJ+g9H4CG/oA+vaUt3DXfWvwNVM02Uoy0nerqeMi7vS+oVQCf0JJmwU+rRB6TJhKY+PyTcgEugT/TfURVK/wDPWZ+EVYD5z2Qg5Ov359Bv5pl1YG8wVre/g9JwI2kBchTNgCE09XAncGe1ZJhwIdSGwJJVs2iwX7JqNHvOU6x5GHSGn4+j/SJ+13U/kfRnkh7R/A9+m6RbJD3bdR2je0zS+tr9pZTPllI2lVI21VinRo0aTZ6WwuofIemjkk6S9GZJc5I+VLm0muWv67qru667qOu6i2pBLY0aNZo8LYXVv0zSw13XPS1JpZSvSbpU0uGllJkF1D9e0uOLtCFpnqV66qmnBqzUO9/5zsF3KMxuvvlmST1LhuLotNNOkzSs6IJdg9WDtYftrblNws7RHmwpbJ0rx2DrMpMNZsS8ThqN2GOsmTVVGmXl6S+iRBZbdMqMvFzrz08zHtfwyvM8PwJmKD6DhWU9auZI+kA/6Vst4i7nNJ2ILr/88sG1sMiIFGkaPfnkkyUNs/rMXRbAzPliv0m9CMSYET9yPD4W+gL7jjIPRbAreWkHlp73mOZ8zdycLPXzhZiDmOZzijgzMzPzmrrsPiJpYynlkDI/4x+QdK+kb0v6+MI1V0n6+pKe2KhRowNO+0X8rutuKqV8VfMmuz2SbpN0taT/I+krpZQ/XvjsS/tra+XKldqwYcPgRHUzzHe/+11J0m233SZJI3n5agUkMf2kWQpUXyzffSr1sjy0t8spzunLvbWMpmk25GTOABxplNvIABLudRRI5Q19yGy+3n4qD5mDDDCRpM2bNw9dU8v9DtFuFp2Eq8nsNFJvgsusxmecccZIn+D+QDGeQ59AaM+aRLtwKnBpmExR+tUCe7JUegYz+X04JTFW+kZ5d3fzhYvM/P2M1R3Z6B8cCfs9MxS5a7nfv1Q92pK0+l3X/aGkP4yPH5J08ZKe0qhRo4OKJp6B54gjjhggs59UoHQG0WC2yBBQabTmGacvCI0MKI26S2YJbE59N62ATrxmFtZaFtwMm+QVOdLl9XH57HitBfYw5qziA0K7jMdnKZ/z3KyP5595rnZvt1ZGHKRPnUIt11zmR8xsuJRDd0p36+RCnKNIzi7NqRmg5O3kvGegktTvAfYwnEVm+HHX6axHyLMza7Bfy++A+UmnKOf8/P7M6zeOmstuo0ZTSBNF/H379un5558fIDI5zqUerTkxcdoA8TkdHXHgEnCIwGUUTauf6nx34YUXShp1muA0dpddwjOR10D+zNRby98GimTGWedYcLvl2uRGaqc37YM0yHqJDFI9F77PBdpwl9+Z/8wkzPq4VYJrM0EJHAXPdZfmrBybXJq7r2ZSk5TxudaryuB2C4fFNYyVOXf9TLpGp1ydyUJqY8eagPu4JyxhHuhntufzn8lF2HOpM3KXXa7ZuXNny7nXqFGj8TRxxH/hhRcG8rZX0kFORquJPZmTDtQlLFIaTa/EvQQyeK012kNbim0YROYUJkjI70/05nmcvo7MWW2WdjPZg9TbcZmPrH3OnLhba4Zdgt70oeb+SfuZ7RX0dY5lnN6k5tLMNaAP14DM9N/nh8+4FgSmDcbjBKLRt7TMeKos9/PwexhrBtX4NVhTkKPZc85l0n/2GD4di4U6pxsxOoRa4BMcClwH88Lz6JNbP5z7aKm3GjVqNJYmivizs7PasGHDQAZ32S+TaoKYnOYENNRCGNEHcCpy+jo6UlWW+/Hcg4PgVHaPKLgNar9hhaB9TmW3zYPaGWLLtY4ePJN+g1apzfcxZ1IK0JT2XdsLImdiztQduFUCFAKxuJcx1ioQs46sIWOuhaFmQkt0HjX7NJpzxpQpynJ8Uq8LSr8P+sjauSyciT9plzG7TA5qZxLPfF7NupIJZVkHt0pAifyse3I9fq1U5yJq1BC/UaMppPbDb9RoCmniDjxr1qyp5krLwoWwggQnZPFJqXemgNXnXhR3rsBBKQIrCasGu/We97xn6LlS7wqKGyssGbkCYJFrbriwdTwnlTTeP9hF+k8bNaefdC+FrUYBRoy6NFo7AEoFkLOHKNeYB76D1XRWf5zjTmYQ8nWm32mqo61aQAwiDyayzKLjmZaSrWYOENdSPPH2/Nl+LyZCqWerWcd020Y8cTfrdKdOZyt/bippWU8+R3ytOfCkc9Ri1BC/UaMppIki/o4dO3T//fcPkNORAFTDzPOjH/1IUq/oAjFdMYVLLicdedBALXdyAPFREOEYhJKPU98ROSvP4JyBCQdEcBMXKM3JTHu04WYYHDo4vUGpzMDjnEvmsqcPzKU7jsDdoCAaV665pnxLYl08l1yiaiJaBgdJ/Twnl5Alsb0d1izHyrUeuMX8YjbNOgc112z+zyo/aX6TRjPs0G4qNr0iU4Yy0yfG42Pmu3EOVJAjvheYrQWO1aghfqNGU0gTRfyVK1fqlFNOGZz6nlEVcx2nI6ctSICc5YiclUg4QRPhpOHEC9JwFR+/x9sHPZDXcA3m2uQIpF4vgN6BPiGHep9AH0xamQ8OLsUdnTKhBAhDHz3PINemU0/K63BX3gf0F1m3zk1PWakHSvm9ZjrL0tQZdurXIudmaHDqf/wa9gQITR947wlXGDNjY82ypoPUzzeInIlXauZC2skS2nzu+fN4dnID6QpeM2vv2rVrJMvzOGqI36jRFNJEEX/Xrl165JFHRtBd6k9DEAuNLkjNqeZVWZDlkZU5JZHJPGCFE/OCCy4Yeg+BLs6F0F5WLyFA5v3vf7+kYXkLTTNacWRuPncHoUz4kHXemB8PQuFZuCCDQHlvbYzjKga7DAtlH9CCZ4UXaTQZCHPJ81yDnhWCITgWRzJclUFEzx/v5HoJ5od7SYxBG3ACrnVPN1xCg2vuz8xHhkNDzK3rcrgmA9FqOqJ0esoKuzUdTM3len/UEL9RoymkidfOe+qpp0bCRKUeFUDV1E6C9B7EwUkHYnHaca+HwJLaKW3koAf3OBKhGUYGRqbcuHGjpP70rdWeo70bb7xRUo+UfmJnGGvKZ8ijnlDkoYcektQjCtcsVikVrgO5FrRLt1ypR7KU0xmXzw+6ghrHUBuP1Mv0cCygLBxSDbG4hvWgDbgHl6dBRvqdFYx4dds5fck6B+hlXKNOu+l7kajuc5LzkFWEPLUX8wvH4tyeVE8OAi0V7aWG+I0aTSW1H36jRlNIE2X1u67Tjh07RvK6S6Nmi8xLBlvq98BewTKjJMtXaZQtzZx49InIQW8fFj+zodSy1MLqcQ/PO+uss4bG4/fDqtKHFD98HJicEH1g/VCO1koojTNl1XLkJ1tKHzKW3/9PN+KM6a85CKVrczrPSP1aM5ewv9yTOfj8O+5lPLlWLkqmcpL5p10358HaZ96/mrMSlOITe6TGmuf9mXuA59YUmi+HGuI3ajSFNFHEf/HFF3XnnXcOUKt2UieBkJySfk+6S2aWWncvBeU4ZdOpIpFC6k17nKgoJTEJZW42aTSbK0rF733ve5KGHVRAuayUw+me7qZSnyWGQCECiFBWvvWtbx1pHzRKJyWe45mKUJ4mimSmW2nU5AQapjORK6LSMSuVqm4upF/pBk3fall7UkHHejAXWQXJKR2/UkEojSLyOAck36c5l6mY8z2Rpd6zihPfeyYnz6LcXHYbNWo0liaK+NK8nJNBFtKovMPpzgkGKvqJluYcZChecRSSpLe97W2S+hMa5OHemnsp7YAsyIlpvnIZlpOYfpJJGPRz7iBz+nNPmidd7ubadGfNMGOp11dkZiI4GfqKg5LPQ2bgzfny9kD6LItdC+VNB62s7uNtePCJ1HMS3APXU3PZTZ1NZvx1pyKeCXKCwOwv1/uwRhk0A7EeHqpNf/ksg6Q8yIj+swcyECqzFPvYXnjhhYb4jRo1Gk8T1+rv3bt3cOK57FOT4aVRDa6jH/dnYATvHT1ol5MexAGVUrPr32W9NIg+uSaadkB2UBV08iywtM+1yLe4l951110jzyV3ICieTjku7+J0wzN5xYmFeXJ0quXJk0a15FLvIMVzshINffIaiemQgu6g5kCFFSVr2kHMra8zY+GZrDfPoa/uOk3eRZ7NXkwnJh9j5iBkfhifc7AZapth2B7owzO5NnP48epjdnf3lle/UaNGY2miiL9s2TLNzc0NkMfllFrtN6dE/tq1WXfMvyeEFnkNNMFGnAkzpP6k5l5QA64g7dVOuGEi4+e9Ui+XZ6qq1C14CClI4+mm/PMa0oAouMVmoIejbNqJaQ8buqMTiUngXJA1GWPm85d6SwuIm9mBa2G5tJvyLq8eKpw17eEkMsGLczmEUNM+HABI6mHR7Nl0G85xOPLyPwFImWSjVhsxw6EzWMeDmeCEjjvuuCbjN2rUaDxNFPFXrFihdevWDU5NP+lSpky7PidcTR7N+vK89/a9Tp80mnQR9PUAIuRl0Bu5CtmVvjiCIndi6we1QLpaiG2G1iY5Z8QYQWDaryWepN/cj2wPSmEfd9s5SImMzBjhFhyRUxOfFYAyh740mgAFroA+uexK/7J+QXqz1TwD01bOva6hh5gfnsMeIHjH+8QzQdbsU3r0+XdpCaglI+X+DMbhXua/Vj9hqTn1pYb4jRpNJbUffqNGU0gTd+BZtmxZ1XSXjhew9umk4+6tmbcNqgW3oEzCeQWWFlY/TWvSaN58roWFzoAffzasGoqnWsZcWEjY06wTcNNNN0mSzjvvvME9qVRKF1GP7WYu6V8t/j77T7+znFeWjJJ6sYb26BN9rLGgKNdgo3GhrYlytJv1B3Jdai612RfWLgOKpN41mGsQB2pBTNxH+9yTpbXc9JhiRwYX+T7N/Pm5vsxTrezWzp07WwmtRo0ajaeJO/Ds2rWrWhVnqaGFtRMts73yWqs2gksliIazTAb4SKN52zOQhFPZkYETGvNgOsS4Aw/uq6A0CjWUh5zqHliCCS2VnzUFYebNG/fqKMv/zBdtoEx0c2G6r6Y7aS3IBUSkfeYgnaWcEnmTw6uFCmeAE0pETHUejJVBXWlGdS4tw60hOAH2Vy0PXiruamHFEJ+l049n1IXcvNyy7DZq1GgslZdjAnjVDyvlaUkvSPrp/q49SOhovXH6Kr2x+vtG6qv0xunvW7que9P+LproD1+SSimbuq67aKIPfYX0Ruqr9Mbq7xupr9Ibr7/7o8bqN2o0hdR++I0aTSEdiB/+1Qfgma+U3kh9ld5Y/X0j9VV64/V3UZq4jN+oUaMDT43Vb9RoCmliP/xSypWllM2llB+WUj4/qeculUopJ5RSvl1Kua+Uck8p5XMLnx9ZSrm2lPLgwusR+2trUlRKWV5Kua2U8s2F9yeVUm5a6OvflVJefsL114lKKYeXUr5aSrl/YY4vOVjntpTyOwt74O5Syv8spaw6mOf2ldBEfvillOWS/qukD0k6U9InSylnTuLZL4P2SPrdruvOkLRR0m8u9PHzkq7ruu5USdctvD9Y6HOSPN74TyX9xUJffybp0wekV3X6K0nXdF13uqTzNN/vg25uSynrJf2WpIu6rjtb0nJJn9DBPbcvn7que93/JF0i6R/t/RckfWESz34Vff66pMslbZZ03MJnx0nafKD7ttCX4zX/Y3m/pG9KKpp3MJmpzfkB7utaSQ9rQadknx90cytpvaRHJR2peZf2b0r64ME6t6/0b1KsPpMJPbbw2UFJpZQNki6QdJOkdV3XbZGkhddjDlzPhugvJf2eJJznj5L0bNd1OIkfTHN8sqSnJf3Ngmjy16WUOR2Ec9t13U8k/ZmkRyRtkbRN0i06eOf2FdGkfvi1RGAHpTmhlHKopH+Q9Ntd1z23v+sPBJVSPiLpqa7rbvGPK5ceLHM8I+ltkr7Ydd0FmnfbPuBsfY0W9AwflXSSpDdLmtO8iJp0sMztK6JJ/fAfk3SCvT9e0uMTevaSqZSyQvM/+r/tuu5rCx8/WUo5buH74yQ9Ne7+CdI7Jf1CKeXHkr6ieXb/LyUdXkohVO1gmuPHJD3Wdd1NC++/qvmD4GCc28skPdx13dNd1+2W9DVJl+rgndtXRJP64d8s6dQFzeis5pUl35jQs5dEZT5+80uS7uu67s/tq29Iumrh/6s0L/sfUOq67gtd1x3fdd0Gzc/lP3dd92uSvi3p4wuXHRR9laSu656Q9Ggp5bSFjz4g6V4dhHOreRZ/YynlkIU9QV8Pyrl9xTRBpcmHJT0g6UeS/tOBVm5U+vdvNM++3Snp9oW/D2tedr5O0oMLr0ce6L5Gv98r6ZsL/58s6QeSfijpf0laeaD7Z/08X9Kmhfn935KOOFjnVtJ/lnS/pLsl/Q9JKw/muX0lf81zr1GjKaTmudeo0RRS++E3ajSF1H74jRpNIbUffqNGU0jth9+o0RRS++E3ajSF1H74jRpNIbUffqNGU0j/D8eEMqCA8kcAAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.imshow(images[77], cmap='gray')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1951, 100, 100)"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "images = np.array(images)\n",
    "images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "images = images.reshape(-1, 100, 100, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1951, 100, 100, 1)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "images.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1951, 5)"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label_num_ohe.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "import keras\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, MaxPool2D, Flatten, Dense, Dropout\n",
    "# print(\"GPU Available: \", tf.test.is_gpu_available())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "x_train, x_test, y_train, y_test = train_test_split(images, label_num_ohe, test_size=0.20, random_state =42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1560, 100, 100, 1)\n",
      "(1560, 5)\n",
      "(391, 100, 100, 1)\n",
      "(391, 5)\n"
     ]
    }
   ],
   "source": [
    "print(x_train.shape)\n",
    "print(y_train.shape)\n",
    "\n",
    "print(x_test.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train = x_train/255.0\n",
    "x_test = x_test/255.0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W1003 10:11:36.133229 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:74: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
      "\n",
      "W1003 10:11:36.422283 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:517: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
      "\n",
      "W1003 10:11:36.461241 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:4138: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
      "\n",
      "W1003 10:11:36.525193 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3976: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
      "\n",
      "W1003 10:11:36.611793 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/optimizers.py:790: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
      "\n",
      "W1003 10:11:36.639865 139737121220416 deprecation_wrapper.py:119] From /home/gunnika/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3295: The name tf.log is deprecated. Please use tf.math.log instead.\n",
      "\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "\n",
    "model.add(Conv2D(64, (3,3), input_shape=x_train.shape[1:], activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(32, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Conv2D(16, (3,3), activation='relu'))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "\n",
    "model.add(Flatten())\n",
    "\n",
    "model.add(Dense(8, activation='relu'))\n",
    "#model.add(Dropout(0.3))\n",
    "model.add(Dense(5, activation='softmax'))\n",
    "\n",
    "model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_1 (Conv2D)            (None, 98, 98, 64)        640       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 49, 49, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 47, 47, 32)        18464     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 23, 23, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 21, 21, 16)        4624      \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 10, 10, 16)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 8)                 12808     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 5)                 45        \n",
      "=================================================================\n",
      "Total params: 36,581\n",
      "Trainable params: 36,581\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "weights = np.array(df.groupby('label').size())\n",
    "w = weights.sum()/weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "class_weight = {0: w[0],\n",
    "               1: w[1],\n",
    "               2: w[2],\n",
    "               3: w[3],\n",
    "               4: w[4]}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 1248 samples, validate on 312 samples\n",
      "Epoch 1/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1419 - acc: 0.9968 - val_loss: 7.2145 - val_acc: 0.8173\n",
      "Epoch 2/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1441 - acc: 0.9968 - val_loss: 7.0799 - val_acc: 0.8045\n",
      "Epoch 3/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1261 - acc: 0.9984 - val_loss: 7.1673 - val_acc: 0.8013\n",
      "Epoch 4/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1660 - acc: 0.9960 - val_loss: 6.7993 - val_acc: 0.7949\n",
      "Epoch 5/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1538 - acc: 0.9952 - val_loss: 7.6007 - val_acc: 0.8013\n",
      "Epoch 6/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1308 - acc: 0.9968 - val_loss: 7.7801 - val_acc: 0.7917\n",
      "Epoch 7/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1208 - acc: 0.9976 - val_loss: 7.9993 - val_acc: 0.8013\n",
      "Epoch 8/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1174 - acc: 0.9976 - val_loss: 8.0615 - val_acc: 0.8013\n",
      "Epoch 9/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1153 - acc: 0.9976 - val_loss: 8.3098 - val_acc: 0.8077\n",
      "Epoch 10/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1128 - acc: 0.9976 - val_loss: 8.1553 - val_acc: 0.7981\n",
      "Epoch 11/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1107 - acc: 0.9976 - val_loss: 8.2628 - val_acc: 0.8045\n",
      "Epoch 12/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1089 - acc: 0.9976 - val_loss: 8.2604 - val_acc: 0.7981\n",
      "Epoch 13/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1070 - acc: 0.9976 - val_loss: 8.2968 - val_acc: 0.7917\n",
      "Epoch 14/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1051 - acc: 0.9976 - val_loss: 8.3068 - val_acc: 0.7917\n",
      "Epoch 15/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1034 - acc: 0.9976 - val_loss: 8.4067 - val_acc: 0.8013\n",
      "Epoch 16/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1017 - acc: 0.9976 - val_loss: 8.4398 - val_acc: 0.8045\n",
      "Epoch 17/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1006 - acc: 0.9976 - val_loss: 8.5141 - val_acc: 0.8109\n",
      "Epoch 18/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1169 - acc: 0.9976 - val_loss: 8.2798 - val_acc: 0.8173\n",
      "Epoch 19/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1556 - acc: 0.9912 - val_loss: 9.3518 - val_acc: 0.7853\n",
      "Epoch 20/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1549 - acc: 0.9920 - val_loss: 7.3048 - val_acc: 0.7724\n",
      "Epoch 21/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.3815 - acc: 0.9647 - val_loss: 8.4826 - val_acc: 0.7436\n",
      "Epoch 22/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.5637 - acc: 0.9559 - val_loss: 5.3192 - val_acc: 0.7853\n",
      "Epoch 23/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.2108 - acc: 0.9856 - val_loss: 6.3353 - val_acc: 0.7885\n",
      "Epoch 24/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1485 - acc: 0.9936 - val_loss: 6.3774 - val_acc: 0.7788\n",
      "Epoch 25/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.4862 - acc: 0.9615 - val_loss: 5.9962 - val_acc: 0.7564\n",
      "Epoch 26/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.2927 - acc: 0.9704 - val_loss: 5.7863 - val_acc: 0.7147\n",
      "Epoch 27/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.1427 - acc: 0.9920 - val_loss: 7.3012 - val_acc: 0.8109\n",
      "Epoch 28/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.1244 - acc: 0.9944 - val_loss: 7.0658 - val_acc: 0.7981\n",
      "Epoch 29/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0715 - acc: 0.9952 - val_loss: 7.6622 - val_acc: 0.8013\n",
      "Epoch 30/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0415 - acc: 0.9976 - val_loss: 7.6409 - val_acc: 0.8045\n",
      "Epoch 31/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0357 - acc: 0.9984 - val_loss: 8.0898 - val_acc: 0.8141\n",
      "Epoch 32/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0338 - acc: 0.9984 - val_loss: 8.0024 - val_acc: 0.8077\n",
      "Epoch 33/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0317 - acc: 0.9984 - val_loss: 7.9812 - val_acc: 0.8109\n",
      "Epoch 34/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0307 - acc: 0.9984 - val_loss: 8.0218 - val_acc: 0.8173\n",
      "Epoch 35/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0300 - acc: 0.9984 - val_loss: 8.0699 - val_acc: 0.8173\n",
      "Epoch 36/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0294 - acc: 0.9984 - val_loss: 8.0835 - val_acc: 0.8141\n",
      "Epoch 37/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0289 - acc: 0.9984 - val_loss: 8.1952 - val_acc: 0.8173\n",
      "Epoch 38/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0283 - acc: 0.9984 - val_loss: 8.2394 - val_acc: 0.8109\n",
      "Epoch 39/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0279 - acc: 0.9984 - val_loss: 8.3361 - val_acc: 0.8109\n",
      "Epoch 40/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0274 - acc: 0.9984 - val_loss: 8.3940 - val_acc: 0.8077\n",
      "Epoch 41/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0269 - acc: 0.9984 - val_loss: 8.4604 - val_acc: 0.8141\n",
      "Epoch 42/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0265 - acc: 0.9992 - val_loss: 8.5408 - val_acc: 0.8141\n",
      "Epoch 43/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0261 - acc: 0.9992 - val_loss: 8.5009 - val_acc: 0.8141\n",
      "Epoch 44/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0257 - acc: 0.9992 - val_loss: 8.5723 - val_acc: 0.8141\n",
      "Epoch 45/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0254 - acc: 0.9992 - val_loss: 8.6733 - val_acc: 0.8141\n",
      "Epoch 46/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0253 - acc: 0.9992 - val_loss: 8.7484 - val_acc: 0.8141\n",
      "Epoch 47/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0249 - acc: 0.9992 - val_loss: 8.7171 - val_acc: 0.8109\n",
      "Epoch 48/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0247 - acc: 0.9992 - val_loss: 8.7306 - val_acc: 0.8109\n",
      "Epoch 49/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0245 - acc: 0.9992 - val_loss: 8.7842 - val_acc: 0.8109\n",
      "Epoch 50/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0244 - acc: 0.9992 - val_loss: 8.8100 - val_acc: 0.8109\n",
      "Epoch 51/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0243 - acc: 0.9992 - val_loss: 8.8546 - val_acc: 0.8109\n",
      "Epoch 52/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0241 - acc: 0.9992 - val_loss: 8.8467 - val_acc: 0.8077\n",
      "Epoch 53/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0241 - acc: 0.9992 - val_loss: 8.9279 - val_acc: 0.8109\n",
      "Epoch 54/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0240 - acc: 0.9992 - val_loss: 8.9520 - val_acc: 0.8109\n",
      "Epoch 55/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0239 - acc: 0.9992 - val_loss: 8.9622 - val_acc: 0.8109\n",
      "Epoch 56/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0239 - acc: 0.9992 - val_loss: 9.0133 - val_acc: 0.8141\n",
      "Epoch 57/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0238 - acc: 0.9992 - val_loss: 9.0501 - val_acc: 0.8141\n",
      "Epoch 58/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0238 - acc: 0.9992 - val_loss: 9.0617 - val_acc: 0.8109\n",
      "Epoch 59/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0237 - acc: 0.9992 - val_loss: 9.0659 - val_acc: 0.8013\n",
      "Epoch 60/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0237 - acc: 0.9992 - val_loss: 9.1075 - val_acc: 0.8109\n",
      "Epoch 61/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0236 - acc: 0.9992 - val_loss: 9.1189 - val_acc: 0.8077\n",
      "Epoch 62/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0236 - acc: 0.9992 - val_loss: 9.1543 - val_acc: 0.8077\n",
      "Epoch 63/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0236 - acc: 0.9992 - val_loss: 9.2029 - val_acc: 0.8109\n",
      "Epoch 64/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0235 - acc: 0.9992 - val_loss: 9.1865 - val_acc: 0.8077\n",
      "Epoch 65/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0235 - acc: 0.9992 - val_loss: 9.1738 - val_acc: 0.8045\n",
      "Epoch 66/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0235 - acc: 0.9992 - val_loss: 9.1866 - val_acc: 0.8045\n",
      "Epoch 67/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0235 - acc: 0.9992 - val_loss: 9.2255 - val_acc: 0.8077\n",
      "Epoch 68/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.2531 - val_acc: 0.8077\n",
      "Epoch 69/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.2411 - val_acc: 0.8045\n",
      "Epoch 70/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.2627 - val_acc: 0.8045\n",
      "Epoch 71/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.2980 - val_acc: 0.8077\n",
      "Epoch 72/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.3143 - val_acc: 0.8077\n",
      "Epoch 73/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0234 - acc: 0.9992 - val_loss: 9.3278 - val_acc: 0.8077\n",
      "Epoch 74/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.3195 - val_acc: 0.8045\n",
      "Epoch 75/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.3301 - val_acc: 0.8045\n",
      "Epoch 76/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.3484 - val_acc: 0.8045\n",
      "Epoch 77/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.3963 - val_acc: 0.8077\n",
      "Epoch 78/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.4108 - val_acc: 0.8077\n",
      "Epoch 79/80\n",
      "1248/1248 [==============================] - 1s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.4135 - val_acc: 0.8077\n",
      "Epoch 80/80\n",
      "1248/1248 [==============================] - 2s 1ms/step - loss: 0.0233 - acc: 0.9992 - val_loss: 9.4112 - val_acc: 0.8077\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<keras.callbacks.History at 0x7f1698fc9358>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit(x_train,y_train, epochs= 80, validation_split= 0.20, batch_size=20, class_weight = class_weight)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "391/391 [==============================] - 0s 882us/step\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[1.8770249170415543, 0.7928388747565277]"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.evaluate(x_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['loss', 'acc']"
      ]
     },
     "execution_count": 31,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.metrics_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_image(url):\n",
    "    categories = ['facebook', 'google_old', 'google_new', 'miscellaneous', 'self' ]\n",
    "    response = requests.get(url)\n",
    "    img = Image.open(BytesIO(response.content))\n",
    "    img = img.resize((100,100))\n",
    "    img= img.convert('L')\n",
    "    img = np.array(img)\n",
    "    img = img.reshape(-1, 100, 100, 1)\n",
    "    predictions= model.predict(img)\n",
    "    return categories[predictions.argmax()]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [],
   "source": [
    "#self_image = 'https://scontent.fdel21-1.fna.fbcdn.net/v/t1.0-9/19145949_1901818590031524_5854858957661418728_n.jpg?_nc_cat=102&_nc_oc=AQmgLFYJ3IRbfvgww3t4vUDTc_k4K1brtI1ZTg8g1_0ZfuO1RLFhTnwqBp8JNp2dHvs&_nc_ht=scontent.fdel21-1.fna&oh=a28d00cff970b47fec915ca01bd1116e&oe=5DA6326A'\n",
    "#test_image(self_image)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/gunnika/anaconda3/lib/python3.7/site-packages/PIL/Image.py:968: UserWarning: Palette images with Transparency   expressed in bytes should be converted to RGBA images\n",
      "  ' expressed in bytes should be converted ' +\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'google_new'"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "google_new = 'https://secure.gravatar.com/avatar/bc6e493d17a9bc4ff13b664205a48609?d=https://storage.whmcs.community/monthly_2017_10/A.png.308193d82a68e42c7b4e71fa8a5272f4.png'\n",
    "test_image(google_new)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'google_old'"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "google_old = 'https://i.stack.imgur.com/34AD2.jpg'\n",
    "test_image(google_old)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "#facebook = 'https://www.colibrilife.com/wp-content/uploads/2018/01/no-avatar-female.png'\n",
    "#test_image(facebook)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "miscellaneous\n"
     ]
    }
   ],
   "source": [
    "misc = 'http://wisdomquotes.com/wp-content/uploads/inspirational-quotes-every-next-level-of-your-life-will-demand-a-different-you-wisdom-quotes-1.jpg'\n",
    "\n",
    "misc2 = 'https://cdn.lifehack.org/wp-content/uploads/2016/02/26042240/Lifehack_Quotes_251429d69c58da436ad9d6ee8bb751d2.jpg'\n",
    "img = convert_picture(misc2)\n",
    "plt.imshow(img, cmap='gray')\n",
    "plt.show()\n",
    "print(test_image(misc2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "#self2 = 'https://scontent.fdel21-1.fna.fbcdn.net/v/t1.0-9/37919976_2172114239483735_2750643600642539520_n.jpg?_nc_cat=103&_nc_oc=AQlu-b2cQaYYXJ1cFH9Iei7Ocoqc6MG8Kx8BF8tXLKSVeVOndOttmpcjpLs-LOopyH0&_nc_ht=scontent.fdel21-1.fna&oh=7c2119453b6c5ab7e30a6b8c1a42d5d4&oe=5DD5347F'\n",
    "#test_image(self2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      1.00      1.00         1\n",
      "           1       1.00      1.00      1.00         5\n",
      "           2       1.00      0.96      0.98       106\n",
      "           3       0.34      0.27      0.30        63\n",
      "           4       0.79      0.86      0.82       216\n",
      "\n",
      "   micro avg       0.79      0.79      0.79       391\n",
      "   macro avg       0.83      0.82      0.82       391\n",
      "weighted avg       0.78      0.79      0.78       391\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "\n",
    "pred = model.predict(x_test)\n",
    "print(classification_report(np.argmax(pred, axis=1), np.argmax(y_test, axis=1)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import confusion_matrix\n",
    "cm = confusion_matrix(np.argmax(pred, axis=1), np.argmax(y_test, axis=1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[  1,   0,   0,   0,   0],\n",
       "       [  0,   5,   0,   0,   0],\n",
       "       [  0,   0, 102,   2,   2],\n",
       "       [  0,   0,   0,  17,  46],\n",
       "       [  0,   0,   0,  31, 185]])"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "cm"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}