GenerarPesos_Clasificadores.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "import numpy as np\n",
    "import cv2 as cv\n",
    "import matplotlib.pyplot as plt\n",
    "import sys\n",
    "import os\n",
    "import ssl\n",
    "import random\n",
    "\n",
    "ssl._create_default_https_context = ssl._create_unverified_context"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0821 00:54:54.388634 140734889190848 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 28, 28, 32)        832       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 10, 10, 64)        51264     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 5, 5, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 800)               1280800   \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 120)               96120     \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 84)                10164     \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 36)                3060      \n",
      "=================================================================\n",
      "Total params: 1,442,240\n",
      "Trainable params: 1,442,240\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "modelo_combinado =  keras.models.Sequential()\n",
    "modelo_combinado.add(keras.layers.Conv2D(32, (5, 5), activation='relu', kernel_initializer='he_uniform', padding='valid', input_shape=(32, 32, 1)))\n",
    "modelo_combinado.add(keras.layers.MaxPooling2D((2, 2)))\n",
    "modelo_combinado.add(keras.layers.Conv2D(64, (5, 5), activation='relu', kernel_initializer='he_uniform', padding='valid'))\n",
    "modelo_combinado.add(keras.layers.MaxPooling2D((2, 2)))\n",
    "modelo_combinado.add(keras.layers.Flatten())\n",
    "modelo_combinado.add(keras.layers.Dense(800, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_combinado.add(keras.layers.Dense(120, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_combinado.add(keras.layers.Dense(84, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_combinado.add(keras.layers.Dense(36, activation='softmax'))\n",
    "\n",
    "modelo_combinado.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "32 -> W\n",
      "33 -> X\n",
      "1 -> 1\n",
      "4 -> 4\n",
      "3 -> 3\n",
      "2 -> 2\n",
      "34 -> Y\n",
      "31 -> V\n",
      "9 -> 9\n",
      "27 -> R\n",
      "16 -> G\n",
      "20 -> K\n",
      "18 -> I\n",
      "25 -> P\n",
      "17 -> H\n",
      "26 -> Q\n",
      "19 -> J\n",
      "28 -> S\n",
      "15 -> F\n",
      "10 -> A\n",
      "7 -> 7\n",
      "29 -> T\n",
      "5 -> 5\n",
      "0 -> 0\n",
      "6 -> 6\n",
      "30 -> U\n",
      "35 -> Z\n",
      "8 -> 8\n",
      "13 -> D\n",
      "12 -> C\n",
      "24 -> O\n",
      "21 -> L\n",
      "22 -> M\n",
      "23 -> N\n",
      "11 -> B\n",
      "14 -> E\n"
     ]
    }
   ],
   "source": [
    "dataset = \"dataset/\"\n",
    "\n",
    "imagenesTrain = []\n",
    "imagenesTest = []\n",
    "\n",
    "clases = [\"0\", \"1\", \"2\", \"3\", \"4\", \"5\", \"6\", \"7\", \"8\", \"9\", \"A\", \"B\", \"C\", \"D\", \"E\", \"F\", \"G\", \"H\", \"I\", \"J\", \"K\", \"L\", \"M\", \"N\", \"O\", \"P\", \"Q\", \"R\", \"S\", \"T\", \"U\", \"V\", \"W\", \"X\", \"Y\", \"Z\"]\n",
    "\n",
    "samples = os.listdir(dataset)\n",
    "\n",
    "#Dividimos entre el 80% y el 20%\n",
    "totalImagenes = 0\n",
    "nTrain = 0\n",
    "nTest = 0\n",
    "\n",
    "for s in range(len(samples)):\n",
    "    carpetas = os.listdir(dataset+samples[s]+\"/\")\n",
    "    claseStr = samples[s][7:]\n",
    "    clase = clases[int(claseStr)-1]\n",
    "    print(int(claseStr)-1,\"->\",clase)\n",
    "    for d in range(len(carpetas)):\n",
    "        if carpetas[d] == 'gen' or carpetas[d] == 'bmp':\n",
    "            archivos = os.listdir(dataset+samples[s]+\"/\"+carpetas[d]+\"/\")\n",
    "            nArchivosLocal = len(archivos)\n",
    "            nTrainLocal = round(nArchivosLocal * 0.8)\n",
    "            nTestLocal = nArchivosLocal - nTrainLocal\n",
    "            totalImagenes += nArchivosLocal\n",
    "            nTrain += nTrainLocal\n",
    "            nTest += nTestLocal\n",
    "            for f in range(len(archivos)):\n",
    "                imagen = cv.imread(dataset+samples[s]+\"/\"+carpetas[d]+\"/\"+archivos[f])\n",
    "                imagen = cv.cvtColor(imagen, cv.COLOR_BGR2GRAY)\n",
    "                imagen = cv.resize(imagen, (32, 32))\n",
    "                if f < nTrainLocal:\n",
    "                    imagenesTrain.append([imagen,int(claseStr)-1,archivos[f]])\n",
    "                else:\n",
    "                    imagenesTest.append([imagen,int(claseStr)-1,archivos[f]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total de imagenes para entrenar: 19377\n",
      "Total de imagenes para comprobar: 4846\n",
      "--------------------------------------------\n",
      "nTrain: 19377\n",
      "nTest: 4846\n",
      "nImgs 24223\n"
     ]
    }
   ],
   "source": [
    "print(\"Total de imagenes para entrenar:\",len(imagenesTrain))\n",
    "print(\"Total de imagenes para comprobar:\",len(imagenesTest))\n",
    "print(\"--------------------------------------------\")\n",
    "print(\"nTrain:\",nTrain)\n",
    "print(\"nTest:\",nTest)\n",
    "print(\"nImgs\",totalImagenes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Antes de desordernar las imagenes (19377, 3) - (4846, 3)\n",
      "Despues de desordernar las imagenes (19377, 3) - (4846, 3)\n"
     ]
    }
   ],
   "source": [
    "print(\"Antes de desordernar las imagenes\",np.shape(imagenesTrain),\"-\",np.shape(imagenesTest))\n",
    "random.shuffle(imagenesTrain)\n",
    "random.shuffle(imagenesTest)\n",
    "print(\"Despues de desordernar las imagenes\",np.shape(imagenesTrain),\"-\",np.shape(imagenesTest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Comprobemos que quedo el arreglo desordenado\n",
      "Clase\tImagen\tTRAIN\n",
      "14 \t bmp15-0119.png\n",
      "10 \t A_0_rot3hsvall.png\n",
      "12 \t C_5_rot0lab.png\n",
      "18 \t I_16_rot3luvall.png\n",
      "5 \t 5_10_rot2luv.png\n",
      "2 \t 2_5_rot5hsv.png\n",
      "23 \t bmp24-0271.png\n",
      "30 \t U_41_thresh1_7.png\n",
      "31 \t V_42_rot5hsv.png\n",
      "25 \t P_30_con_6.png\n",
      "18 \t I_16_rot0xyzall.png\n",
      "4 \t 4_8335.png\n",
      "10 \t A_3470.png\n",
      "0 \t 0_0_rot2hsvall.png\n",
      "12 \t C_5_canny1.png\n",
      "7 \t 7_3390.png\n",
      "2 \t 2_3176.png\n",
      "26 \t Q_8170.png\n",
      "20 \t K_20_rot0luvall.png\n",
      "25 \t bmp26-0150.png\n",
      "35 \t Z_50_rot6hsv.png\n",
      "21 \t L_22_thresh1_2.png\n",
      "16 \t bmp17-0053.png\n",
      "7 \t 7_1204.png\n",
      "3 \t bmp04-0040.png\n",
      "33 \t X_2866.png\n",
      "33 \t X_47_rot7hlsall.png\n",
      "26 \t Q_32_con_6.png\n",
      "8 \t 8_2233.png\n",
      "14 \t E_9_rot5laball.png\n",
      "33 \t X_47_rot6hsvall.png\n",
      "29 \t T_1775.png\n",
      "15 \t F_10_rot7xyz.png\n",
      "10 \t A_5982.png\n",
      "15 \t F_4630.png\n",
      "14 \t bmp15-0124.png\n",
      "29 \t bmp30-0051.png\n",
      "27 \t bmp28-0039.png\n",
      "14 \t E_9611.png\n",
      "14 \t E_8_rot4bgr.png\n",
      "26 \t bmp27-0019.png\n",
      "30 \t bmp31-0062.png\n",
      "10 \t bmp11-0117.png\n",
      "9 \t 9_900.png\n",
      "0 \t 0_0_rot7laball.png\n",
      "14 \t E_8_rot0ycrcb.png\n",
      "23 \t N_8726.png\n",
      "7 \t bmp08-0011.png\n",
      "16 \t G_13_rot3hlsall.png\n",
      "14 \t bmp15-0083.png\n"
     ]
    }
   ],
   "source": [
    "print(\"Comprobemos que quedo el arreglo desordenado\")\n",
    "print(\"Clase\\tImagen\\tTRAIN\")\n",
    "for i in range(50):\n",
    "    print(imagenesTrain[i][1],\"\\t\",imagenesTrain[i][2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Clase\tImagen\tTEST\n",
      "27 \t bmp28-0292.png\n",
      "8 \t 8_17_canny6.png\n",
      "27 \t R_4270.png\n",
      "16 \t bmp17-0109.png\n",
      "24 \t O_29_canny7.png\n",
      "24 \t O_29_rot1lab.png\n",
      "28 \t bmp29-0135.png\n",
      "23 \t N_27_rot6hls.png\n",
      "7 \t 7_14_rot7bgr.png\n",
      "21 \t L_22_gauss17_2.png\n",
      "2 \t 2_4_rot6ycrcb.png\n",
      "7 \t 7_128.png\n",
      "14 \t E_8_rot0luvall.png\n",
      "25 \t P_5538.png\n",
      "24 \t O_29_rot1hls.png\n",
      "7 \t 7_14_rot2luvall.png\n",
      "18 \t I_9014.png\n",
      "11 \t B_2_rot7hlsall.png\n",
      "7 \t 7_2751.png\n",
      "23 \t N_26_rot5luv.png\n",
      "8 \t 8_16_thresh2_0.png\n",
      "21 \t L_23_rot7laball.png\n",
      "2 \t 2_5_rot4hsv.png\n",
      "11 \t B_3_rot7hls.png\n",
      "10 \t A_1_rot4rgball.png\n",
      "19 \t J_18_thresh1_7.png\n",
      "14 \t bmp15-0200.png\n",
      "9 \t 9_18_rot1luvall.png\n",
      "21 \t L_23_rot7luvall.png\n",
      "7 \t 7_6126.png\n",
      "25 \t P_2332.png\n",
      "19 \t J_19_rot3gray.png\n",
      "24 \t O_28_rot2luv.png\n",
      "18 \t bmp19-0088.png\n",
      "16 \t G_12_canny4.png\n",
      "22 \t M_24_rot1grayall.png\n",
      "17 \t bmp18-0098.png\n",
      "8 \t 8_1322.png\n",
      "25 \t P_30_rot6laball.png\n",
      "20 \t bmp21-0016.png\n",
      "8 \t 8_1278.png\n",
      "27 \t R_2364.png\n",
      "18 \t I_16_rot2hls.png\n",
      "27 \t R_34_thresh1_1.png\n",
      "11 \t B_3_rot7luv.png\n",
      "13 \t bmp14-0164.png\n",
      "28 \t bmp29-0294.png\n",
      "17 \t H_14_rot7luv.png\n",
      "15 \t F_7421.png\n",
      "27 \t bmp28-0082.png\n"
     ]
    }
   ],
   "source": [
    "print(\"Clase\\tImagen\\tTEST\")\n",
    "for i in range(50):\n",
    "    print(imagenesTest[i][1],\"\\t\",imagenesTest[i][2])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(19377, 32, 32, 1) (19377,) (4846, 32, 32, 1) (4846,)\n",
      "[14 10 12 18  5  2 23 30 31 25 18  4 10  0 12  7  2 26 20 25 35 21 16  7\n",
      "  3 33 33 26  8 14 33 29 15 10 15 14 29 27 14 14 26 30 10  9  0 14 23  7\n",
      " 16 14]\n",
      "[27  8 27 16 24 24 28 23  7 21  2  7 14 25 24  7 18 11  7 23  8 21  2 11\n",
      " 10 19 14  9 21  7 25 19 24 18 16 22 17  8 25 20  8 27 18 27 11 13 28 17\n",
      " 15 27]\n"
     ]
    }
   ],
   "source": [
    "xTrain = []\n",
    "yTrain = []\n",
    "xTest = []\n",
    "yTest = []\n",
    "\n",
    "#Tomo las de entrenar\n",
    "for i in range(len(imagenesTrain)):\n",
    "    xTrain.append(imagenesTrain[i][0])\n",
    "    yTrain.append(imagenesTrain[i][1])\n",
    "\n",
    "#Tomo las de comprobar\n",
    "for i in range(len(imagenesTest)):\n",
    "    xTest.append(imagenesTest[i][0])\n",
    "    yTest.append(imagenesTest[i][1])\n",
    "    \n",
    "xTrain = np.array(xTrain)\n",
    "yTrain = np.array(yTrain)\n",
    "xTest = np.array(xTest)\n",
    "yTest = np.array(yTest)\n",
    "\n",
    "xTrain = xTrain.reshape([-1, 32, 32, 1])\n",
    "xTest = xTest.reshape([-1, 32, 32, 1])\n",
    "xTrain = xTrain.astype('float32')\n",
    "xTest = xTest.astype('float32')\n",
    "\n",
    "xTrain /= 255\n",
    "xTest /= 255\n",
    "\n",
    "print(xTrain.shape, yTrain.shape, xTest.shape, yTest.shape)\n",
    "\n",
    "print(yTrain[0:50])\n",
    "\n",
    "print(yTest[0:50])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(32, 32, 1)\n"
     ]
    }
   ],
   "source": [
    "print(np.shape(xTrain[0]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 19377 samples, validate on 4846 samples\n",
      "Epoch 1/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 2.9803 - acc: 0.2442 - val_loss: 2.1838 - val_acc: 0.4757\n",
      "Epoch 2/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.8392 - acc: 0.5548 - val_loss: 1.4910 - val_acc: 0.6457\n",
      "Epoch 3/50\n",
      "19377/19377 [==============================] - 37s 2ms/sample - loss: 1.2799 - acc: 0.6900 - val_loss: 1.1329 - val_acc: 0.7109\n",
      "Epoch 4/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.9940 - acc: 0.7611 - val_loss: 0.9473 - val_acc: 0.7641\n",
      "Epoch 5/50\n",
      "19377/19377 [==============================] - 47s 2ms/sample - loss: 0.8240 - acc: 0.8007 - val_loss: 0.8340 - val_acc: 0.7924\n",
      "Epoch 6/50\n",
      "19377/19377 [==============================] - 41s 2ms/sample - loss: 0.7068 - acc: 0.8297 - val_loss: 0.7177 - val_acc: 0.8341\n",
      "Epoch 7/50\n",
      "19377/19377 [==============================] - 45s 2ms/sample - loss: 0.6208 - acc: 0.8510 - val_loss: 0.6442 - val_acc: 0.8463\n",
      "Epoch 8/50\n",
      "19377/19377 [==============================] - 46s 2ms/sample - loss: 0.5585 - acc: 0.8644 - val_loss: 0.6357 - val_acc: 0.8549\n",
      "Epoch 9/50\n",
      "19377/19377 [==============================] - 44s 2ms/sample - loss: 0.4985 - acc: 0.8797 - val_loss: 0.5724 - val_acc: 0.8636\n",
      "Epoch 10/50\n",
      "19377/19377 [==============================] - 53s 3ms/sample - loss: 0.4526 - acc: 0.8899 - val_loss: 0.5760 - val_acc: 0.8527\n",
      "Epoch 11/50\n",
      "19377/19377 [==============================] - 67s 3ms/sample - loss: 0.4224 - acc: 0.8976 - val_loss: 0.5442 - val_acc: 0.8737\n",
      "Epoch 12/50\n",
      "19377/19377 [==============================] - 46s 2ms/sample - loss: 0.3865 - acc: 0.9064 - val_loss: 0.5288 - val_acc: 0.8758\n",
      "Epoch 13/50\n",
      "19377/19377 [==============================] - 45s 2ms/sample - loss: 0.3533 - acc: 0.9119 - val_loss: 0.5201 - val_acc: 0.8756\n",
      "Epoch 14/50\n",
      "19377/19377 [==============================] - 46s 2ms/sample - loss: 0.3218 - acc: 0.9217 - val_loss: 0.5005 - val_acc: 0.8828\n",
      "Epoch 15/50\n",
      "19377/19377 [==============================] - 45s 2ms/sample - loss: 0.2996 - acc: 0.9295 - val_loss: 0.4936 - val_acc: 0.8846\n",
      "Epoch 16/50\n",
      "19377/19377 [==============================] - 45s 2ms/sample - loss: 0.2830 - acc: 0.9313 - val_loss: 0.5042 - val_acc: 0.8900\n",
      "Epoch 17/50\n",
      "19377/19377 [==============================] - 42s 2ms/sample - loss: 0.2675 - acc: 0.9355 - val_loss: 0.4813 - val_acc: 0.8888\n",
      "Epoch 18/50\n",
      "19377/19377 [==============================] - 43s 2ms/sample - loss: 0.2439 - acc: 0.9421 - val_loss: 0.5025 - val_acc: 0.8906\n",
      "Epoch 19/50\n",
      "19377/19377 [==============================] - 36s 2ms/sample - loss: 0.2330 - acc: 0.9463 - val_loss: 0.4450 - val_acc: 0.9055\n",
      "Epoch 20/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.2198 - acc: 0.9489 - val_loss: 0.4584 - val_acc: 0.8948\n",
      "Epoch 21/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.1992 - acc: 0.9548 - val_loss: 0.4529 - val_acc: 0.8970\n",
      "Epoch 22/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.1898 - acc: 0.9572 - val_loss: 0.4364 - val_acc: 0.9082\n",
      "Epoch 23/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.1761 - acc: 0.9601 - val_loss: 0.4276 - val_acc: 0.9106\n",
      "Epoch 24/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1680 - acc: 0.9619 - val_loss: 0.4341 - val_acc: 0.8995\n",
      "Epoch 25/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1516 - acc: 0.9656 - val_loss: 0.4289 - val_acc: 0.9119\n",
      "Epoch 26/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1429 - acc: 0.9693 - val_loss: 0.4484 - val_acc: 0.9051\n",
      "Epoch 27/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.1368 - acc: 0.9704 - val_loss: 0.4269 - val_acc: 0.9059\n",
      "Epoch 28/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1314 - acc: 0.9728 - val_loss: 0.4135 - val_acc: 0.9168\n",
      "Epoch 29/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 0.1233 - acc: 0.9748 - val_loss: 0.4014 - val_acc: 0.9241\n",
      "Epoch 30/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1224 - acc: 0.9751 - val_loss: 0.4099 - val_acc: 0.9218\n",
      "Epoch 31/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.1187 - acc: 0.9757 - val_loss: 0.4498 - val_acc: 0.9131\n",
      "Epoch 32/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.1100 - acc: 0.9780 - val_loss: 0.4182 - val_acc: 0.9236\n",
      "Epoch 33/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.1041 - acc: 0.9787 - val_loss: 0.4477 - val_acc: 0.9154\n",
      "Epoch 34/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.1001 - acc: 0.9805 - val_loss: 0.4169 - val_acc: 0.9247\n",
      "Epoch 35/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.0957 - acc: 0.9806 - val_loss: 0.4207 - val_acc: 0.9247\n",
      "Epoch 36/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.0913 - acc: 0.9820 - val_loss: 0.4211 - val_acc: 0.9243\n",
      "Epoch 37/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.0845 - acc: 0.9831 - val_loss: 0.4340 - val_acc: 0.9230\n",
      "Epoch 38/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.0902 - acc: 0.9828 - val_loss: 0.4199 - val_acc: 0.9307\n",
      "Epoch 39/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.0782 - acc: 0.9858 - val_loss: 0.4218 - val_acc: 0.9226\n",
      "Epoch 40/50\n",
      "19377/19377 [==============================] - 33s 2ms/sample - loss: 0.0733 - acc: 0.9863 - val_loss: 0.4203 - val_acc: 0.9284\n",
      "Epoch 41/50\n",
      "19377/19377 [==============================] - 39s 2ms/sample - loss: 0.0786 - acc: 0.9846 - val_loss: 0.4449 - val_acc: 0.9187\n",
      "Epoch 42/50\n",
      "19377/19377 [==============================] - 39s 2ms/sample - loss: 0.0769 - acc: 0.9840 - val_loss: 0.4364 - val_acc: 0.9173\n",
      "Epoch 43/50\n",
      "19377/19377 [==============================] - 37s 2ms/sample - loss: 0.0732 - acc: 0.9862 - val_loss: 0.4172 - val_acc: 0.9298\n",
      "Epoch 44/50\n",
      "19377/19377 [==============================] - 36s 2ms/sample - loss: 0.0705 - acc: 0.9871 - val_loss: 0.4114 - val_acc: 0.9259\n",
      "Epoch 45/50\n",
      "19377/19377 [==============================] - 36s 2ms/sample - loss: 0.0721 - acc: 0.9857 - val_loss: 0.4350 - val_acc: 0.9298\n",
      "Epoch 46/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.0635 - acc: 0.9882 - val_loss: 0.4433 - val_acc: 0.9251\n",
      "Epoch 47/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.0678 - acc: 0.9878 - val_loss: 0.4342 - val_acc: 0.9272\n",
      "Epoch 48/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.0621 - acc: 0.9878 - val_loss: 0.4210 - val_acc: 0.9298\n",
      "Epoch 49/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 0.0584 - acc: 0.9888 - val_loss: 0.4238 - val_acc: 0.9323\n",
      "Epoch 50/50\n",
      "19377/19377 [==============================] - 34s 2ms/sample - loss: 0.0588 - acc: 0.9889 - val_loss: 0.4364 - val_acc: 0.9311\n"
     ]
    }
   ],
   "source": [
    "opt = keras.optimizers.SGD(lr=0.001, momentum=0.9)\n",
    "modelo_combinado.compile(optimizer=opt, loss='sparse_categorical_crossentropy',metrics=['accuracy'])\n",
    "\n",
    "history = modelo_combinado.fit(xTrain, yTrain, epochs=50, batch_size=64, validation_data=(xTest, yTest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1032d1908>"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13ff546a0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['acc'], label='Train Accuracy')\n",
    "plt.plot(history.history['val_acc'], label='Validation Accuracy')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelo guardado\n"
     ]
    }
   ],
   "source": [
    "model_json = modelo_combinado.to_json()\n",
    "with open(\"model.json\", \"w\") as json_file:\n",
    "    json_file.write(model_json)\n",
    "# serialize weights to HDF5\n",
    "modelo_combinado.save_weights(\"modeloFinal.h5\")\n",
    "print(\"Modelo guardado\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W0821 01:30:33.051874 140734889190848 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:97: calling Zeros.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
      "W0821 01:30:33.675268 140734889190848 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:97: calling GlorotUniform.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelo cargado desde el computador\n",
      "acc: 93.11%\n"
     ]
    }
   ],
   "source": [
    "json_file = open('model.json', 'r')\n",
    "loaded_model_json = json_file.read()\n",
    "json_file.close()\n",
    "loaded_model = keras.models.model_from_json(loaded_model_json)\n",
    "#Abriendo pesos del modelo creado\n",
    "loaded_model.load_weights(\"modeloFinal.h5\")\n",
    "print(\"Modelo cargado desde el computador\")\n",
    " \n",
    "#Evaluar el modelo cargado\n",
    "opt = keras.optimizers.SGD(lr=0.001, momentum=0.9)\n",
    "loaded_model.compile(optimizer=opt, loss='sparse_categorical_crossentropy',metrics=['accuracy'])\n",
    "score = loaded_model.evaluate(xTest, yTest, verbose=0)\n",
    "print(\"%s: %.2f%%\" % (loaded_model.metrics_names[1], score[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W0821 01:30:39.587463 140734889190848 lazy_loader.py:50] \n",
      "The TensorFlow contrib module will not be included in TensorFlow 2.0.\n",
      "For more information, please see:\n",
      "  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md\n",
      "  * https://github.com/tensorflow/addons\n",
      "  * https://github.com/tensorflow/io (for I/O related ops)\n",
      "If you depend on functionality not listed there, please file an issue.\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 43ms/sample\n",
      "Caracter: ' 6 '\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1414747f0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = cv.cvtColor(cv.imread(\"pruebas/Letras/6.png\"),cv.COLOR_BGR2RGB)\n",
    "img = cv.resize(img, (32, 32))\n",
    "imgG = cv.cvtColor(img, cv.COLOR_BGR2GRAY)\n",
    "imgG = tf.contrib.keras.preprocessing.image.img_to_array(imgG)\n",
    "imgG = np.expand_dims(imgG, axis=0);\n",
    "plt.imshow(img,cmap='gray');\n",
    "\n",
    "features_res_classes = loaded_model.predict_classes(imgG, verbose=1)\n",
    "print(\"Caracter: '\",clases[features_res_classes[0]],\"'\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x14f6237f0>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYUAAAEKCAYAAAD9xUlFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAIABJREFUeJzt3XmcXGWV8PHfqa2r9707SyfpkATIvhADCAECKAGEiKKAoAOikYwjosLAOL7K4DLIOICIgwMqwohEX3ZlibwYCYyyJCFkJSRkId1JOt2d3veqOu8f93Z1ddJJKklXV3fX+X4+9blL3bp1bqdSp57lPo+oKsYYYwyAJ9kBGGOMGTwsKRhjjImypGCMMSbKkoIxxpgoSwrGGGOiLCkYY4yJsqRgjDEmypKCMcaYKEsKxhhjonzJDuBoFRUVaXl5ebLDMMaYIWXVqlU1qlp8pOOGXFIoLy9n5cqVyQ7DGGOGFBHZGc9xVn1kjDEmypKCMcaYKEsKxhhjohLWpiAiQWAFkOa+zxOq+r0DjkkDHgVOAWqBK1R1R6JiMsbEr6uri4qKCtrb25MdijkKwWCQsrIy/H7/Mb0+kQ3NHcC5qtosIn7gdRF5UVXfiDnmeqBOVSeKyJXAj4ErEhiTMSZOFRUVZGdnU15ejogkOxwTB1WltraWiooKxo8ff0znSFj1kTqa3U2/+zhwRp9FwCPu+hPAeWKfPmMGhfb2dgoLCy0hDCEiQmFh4XGV7hLapiAiXhFZA+wDXlbVNw84ZDSwC0BVQ0ADUJjImIwx8bOEMPQc779ZQpOCqoZVdRZQBswTkWnHch4RWSwiK0VkZXV19THF8t7eRn6ybDN1LZ3H9HpjjEkFA9L7SFXrgeXAwgOeqgTGAIiID8jFaXA+8PUPqupcVZ1bXHzEG/L6tKOmlfuXb6Wyvu2YXm+MGTi1tbXMmjWLWbNmMWLECEaPHh3d7uyM74fdddddx+bNm+N+z1/+8pfcdNNNxxrysJHI3kfFQJeq1otIOvAxnIbkWM8B/wD8Hbgc+IuqHtju0C8KMgMA1LVaScGYwa6wsJA1a9YAcPvtt5OVlcXNN9/c6xhVRVXxePr+bfvwww8nPM7hKJElhZHAchFZC7yN06bwJxG5Q0QudY/5FVAoIluBbwK3JSqY7qSw36qPjBmytm7dypQpU7j66quZOnUqe/bsYfHixcydO5epU6dyxx13RI8988wzWbNmDaFQiLy8PG677TZmzpzJ6aefzr59++J+z9/+9rdMnz6dadOm8e1vfxuAUCjE5z//+ej+++67D4B77rmHKVOmMGPGDK655pr+vfgBkrCSgqquBWb3sf+7MevtwGcSFUMsSwrGHLt/++MGNu5u7NdzThmVw/cumXrUr3vvvfd49NFHmTt3LgB33nknBQUFhEIhFixYwOWXX86UKVN6vaahoYGzzz6bO++8k29+85v8+te/5rbbjvwbtKKigu985zusXLmS3Nxczj//fP70pz9RXFxMTU0N69atA6C+vh6Au+66i507dxIIBKL7hpqUuaM5L92PRywpGDPUTZgwIZoQAB5//HHmzJnDnDlz2LRpExs3bjzoNenp6Vx44YUAnHLKKezYsSOu93rzzTc599xzKSoqwu/387nPfY4VK1YwceJENm/ezI033siyZcvIzc0FYOrUqVxzzTU89thjx3zzWLINuVFSj5XHI+RnBKi1pGDMUTuWX/SJkpmZGV3fsmULP/3pT3nrrbfIy8vjmmuu6bOPfiAQiK57vV5CodBxxVBYWMjatWt58cUX+fnPf86TTz7Jgw8+yLJly3j11Vd57rnn+NGPfsTatWvxer3H9V4DLWVKCgD5mQHrkmrMMNLY2Eh2djY5OTns2bOHZcuW9ev5Tz31VJYvX05tbS2hUIilS5dy9tlnU11djarymc98hjvuuIPVq1cTDoepqKjg3HPP5a677qKmpobW1tZ+jWcgpExJAZx2BSspGDN8zJkzhylTpnDyySczbtw4zjjjjOM6369+9SueeOKJ6PbKlSv5/ve/zznnnIOqcskll3DxxRezevVqrr/+elQVEeHHP/4xoVCIz33uczQ1NRGJRLj55pvJzs4+3ksccJKgHqAJM3fuXD3WSXZu+J9VfFDdzMvfPLufozJm+Nm0aROTJ09OdhjmGPT1byciq1R17iFeEpVS1UcFWQFraDbGmMNIqaRQmBmgrrWTSGRolY6MMWagpFRSyM8IEFFoaOtKdijGGDMopVRSKMxyuqVZY7MxxvQtpZJCfoaNf2SMMYeTUkmhe6iL2mZLCsYY05eUSgrd1UfWA8mYwW3BggUH3Yh27733smTJksO+LisrC4Ddu3dz+eWX93nMOeecw5G6td977729bjy76KKL+mUso9tvv52f/OQnx32eREqppGDVR8YMDVdddRVLly7ttW/p0qVcddVVcb1+1KhRvW5CO1oHJoUXXniBvLy8Yz7fUJJSSSHo95IZ8Fr1kTGD3OWXX87zzz8fnVBnx44d7N69m/nz59Pc3Mx5553HnDlzmD59Os8+++xBr9+xYwfTpjkTPba1tXHllVcyefJkLrvsMtraeibaWrJkSXTY7e9973sA3HfffezevZsFCxawYMECAMrLy6mpqQHg7rvvZtq0aUybNo177703+n6TJ0/my1/+MlOnTuXjH/94r/c5kr7O2dLSwsUXX8zMmTOZNm0av//97wG47bbbosNzHzjHRH9IqWEuwB3/yEoKxhydF2+Dvev695wjpsOFd/b5VEFBAfPmzePFF19k0aJFLF26lM9+9rOICMFgkKeffpqcnBxqamo47bTTuPTSSw85N/EDDzxARkYGmzZtYu3atcyZMyf63A9/+EMKCgoIh8Ocd955rF27lhtvvJG7776b5cuXU1RU1Otcq1at4uGHH+bNN99EVTn11FM5++yzyc/PZ8uWLTz++OM89NBDfPazn+XJJ5+Ma06FQ51z27ZtjBo1iueffx5whv+ura3l6aef5r333kNEEjI8d0qVFMC5gc26pBoz+MVWIcVWHakq3/72t5kxYwbnn38+lZWVVFVVHfI8K1asiH45z5gxgxkzZkSf+8Mf/sCcOXOYPXs2GzZs6HPY7Vivv/46l112GZmZmWRlZfGpT32K1157DYDx48cza9Ys4OiG5z7UOadPn87LL7/MrbfeymuvvUZubi65ubkEg0Guv/56nnrqKTIyMuJ6j6ORciWFgswA1c0dyQ7DmKHlEL/oE2nRokV84xvfYPXq1bS2tnLKKacA8Nhjj1FdXc2qVavw+/2Ul5f3OVz2kWzfvp2f/OQnvP322+Tn53Pttdce03m6paWlRde9Xu9RVR/15cQTT2T16tW88MILfOc73+G8887ju9/9Lm+99RavvPIKTzzxBPfffz9/+ctfjut9DpRyJQVn+Gy7o9mYwS4rK4sFCxbwxS9+sVcDc0NDAyUlJfj9fpYvX87OnTsPe56zzjqL3/3udwCsX7+etWvXAs6w25mZmeTm5lJVVcWLL74YfU12djZNTU0HnWv+/Pk888wztLa20tLSwtNPP838+fOP6zoPdc7du3eTkZHBNddcwy233MLq1atpbm6moaGBiy66iHvuuYd33333uN67LylXUnCqj6ykYMxQcNVVV3HZZZf16ol09dVXc8kllzB9+nTmzp3LySeffNhzLFmyhOuuu47JkyczefLkaIlj5syZzJ49m5NPPpkxY8b0GnZ78eLFLFy4kFGjRrF8+fLo/jlz5nDttdcyb948AL70pS8xe/bsuKuKAH7wgx9EG5PBmfKzr3MuW7aMW265BY/Hg9/v54EHHqCpqYlFixbR3t6OqnL33XfH/b7xSqmhswH+669bueulzWy6YyHpgaE1I5IxA8mGzh66bOjso1DYfVezlRaMMeYgKZcUCjKdxiC7q9kYYw6WgknBD1hSMCYeQ6162Rz/v1kKJgUrKRgTj2AwSG1trSWGIURVqa2tJRgMHvM5Uq73UUGGDYpnTDzKysqoqKiguro62aGYoxAMBikrKzvm1ycsKYjIGOBRoBRQ4EFV/ekBx5wDPAtsd3c9pap3JComgJx0Hz6PWFIw5gj8fj/jx49PdhhmgCWypBACvqWqq0UkG1glIi+r6oH3kb+mqp9IYBy9iAj5mQFLCsYY04eEtSmo6h5VXe2uNwGbgNGJer+jUZBhScEYY/oyIA3NIlIOzAbe7OPp00XkXRF5UUSmDkQ8BVZSMMaYPiU8KYhIFvAkcJOqNh7w9GpgnKrOBH4GPHOIcywWkZUisrI/Gr0KMgPst+GzjTHmIAlNCiLix0kIj6nqUwc+r6qNqtrsrr8A+EWkqI/jHlTVuao6t7i4+LjjspKCMcb0LWFJQZwZL34FbFLVPkdtEpER7nGIyDw3ntpExdQtPzNAfWsXoXAk0W9ljDFDSiJ7H50BfB5YJyJr3H3fBsYCqOovgMuBJSISAtqAK3UA7pTpHv+ovq2Loqy0IxxtjDGpI2FJQVVfB/qeH6/nmPuB+xMVw6EUZPbcwGZJwRhjeqTcMBfQOykYY4zpYUnBGGNMVEonhVpLCsYY00tKJoV8d1C8OksKxhjTS0omhYDPQ3bQZ9VHxhhzgJRMCmA3sBljTF8sKRhjjIlK3aSQEbCGZmOMOUDqJoXMgDU0G2PMAVI3KWQ51Uc2/6wxxvRI3aSQEaAzHKGlM5zsUIwxZtBI3aTQfVdzs1UhGWNMt5RPCrUtHUmOxBhjBo+UTwp1NgObMcZEpWxSKMx0hsyuteojY4yJSp2ksOttePJL0OJM7Jaf6QespGCMMbFSJym07Yd1/xf2fwBAVpqPgNdjN7AZY0yM1EkK+eXOsm4HACJCfqbfeh8ZY0yM1EkKeWOdpZsUAAoy06z6yBhjYqROUvCnQ/bIXkmhMNPGPzLGmFipkxTAqUKq29mzaeMfGWNMLymYFHZEN62kYIwxvaVeUmishJBzF3N+RoCm9hCdoUhy4zLGmEEi9ZICCvW7AGekVIB6a2w2xhggJZMC0Sqkwuj4R5YUjDEGEpgURGSMiCwXkY0iskFEvt7HMSIi94nIVhFZKyJzEhUPEJMUtjubGe74R5YUjDEGAF8Czx0CvqWqq0UkG1glIi+r6saYYy4EJrmPU4EH3GViZJWCL9hTUsiykoIxxsRKWElBVfeo6mp3vQnYBIw+4LBFwKPqeAPIE5GRiYoJEcgbF00K3SWF/ZYUjDEGGKA2BREpB2YDbx7w1GhgV8x2BQcnjv4Vc69CfoYzKJ4lBWOMcSQ8KYhIFvAkcJOqNh7jORaLyEoRWVldXX18AXXfq6CKz+shL8NvScEYY1wJTQoi4sdJCI+p6lN9HFIJjInZLnP39aKqD6rqXFWdW1xcfHxB5ZdDZxO07gecuZr3W5dUY4wBEtv7SIBfAZtU9e5DHPYc8AW3F9JpQIOq7klUTMBB3VILMgM2UqoxxrgS2fvoDODzwDoRWePu+zYwFkBVfwG8AFwEbAVagesSGI8jtltq2SnkZwb4sLY14W9rjDFDQcKSgqq+DsgRjlHgq4mKoU/545xlzA1sa3bVD2gIxhgzWKXWHc0AgUzILIF6pwdSgTtSqpOfjDEmtaVeUoBeo6UWZAYIRZTG9lBSQzLGmMHAkkKm3cBmjDHdUjcpNFRAuIv8aFLoSG5MxhgzCKRuUtAINOyKjpS6v6UruTEZY8wgkLpJAaBuB0VZaQBUNbYnLx5jjBkkjpgURORrIpI/EMEMmJikMCInSLrfywfVzUkNyRhjBoN4SgqlwNsi8gcRWejeqTy0ZY8EbwDqduDxCBNLsti6z5KCMcYcMSmo6ndw5jv4FXAtsEVEfiQiExIcW+J4PJA3NtoDaVJJFluqLCkYY0xcbQruncd73UcIyAeeEJG7EhhbYsV0S51YmsXexnYa262x2RiT2uJpU/i6iKwC7gL+F5iuqkuAU4BPJzi+xIlJCpNKsgGsCskYk/LiGfuoAPiUqu6M3amqERH5RGLCGgD55dDeAG11TCrJAmBrVTNzxg6vNnVjjDka8VQfvQjs794QkRwRORVAVTclKrCEi+mBNKYgg4DPw5Z9TUkNyRhjki2epPAAEFuv0uzuG9pikoLXI0wozmKLVR8ZY1JcPElBNGYIUVWNkNh5GAZGXvcQ2k6tmPVAMsaY+JLCNhG5UUT87uPrwLZEB5ZwwRzIKOzVLbWyvo2WDhst1RiTuuJJCjcAH8WZO7kCOBVYnMigBkxsD6RSpweS3dlsjEllR6wGUtV9wJUDEMvAyy+HytUATCp1eiBtqWpmRlleEoMyxpjkOWJSEJEgcD0wFQh271fVLyYwroGRXw4bn4VwiHEFGfi9Yo3NxpiUFk/10f8AI4ALgFeBMmB49N3ML4dICBor8Xk9nFCUxVbrlmqMSWHxJIWJqvp/gBZVfQS4GKddYeiL6ZYKznAXVlIwxqSyeJJC94BA9SIyDcgFShIX0gA6IClMKsniw/2ttHeFkxaSMcYkUzxJ4UF3PoXvAM8BG4EfJzSqgZIzGjy+XmMgqVoPJGNM6jpsQ7OIeIBGVa0DVgAnDEhUA8XjhdwxMd1S3TGQ9jUzdVRuEgMzxpjkOGxJwb17+Z8HKJbkiLlXobwwE69HeL/KGpuNMakpnuqj/yciN4vIGBEp6H4c6UUi8msR2Sci6w/x/Dki0iAia9zHd486+v4QkxQCPg/lhRk23IUxJmXFM4bRFe7yqzH7lCNXJf0GuB949DDHvKaqyR1+O78c2vY7w2gHc5lUkm0lBWNMyornjubxx3JiVV0hIuXH8toBFe2BtBNGzmBSaRZ/3riXjlCYNJ83qaEZY8xAi+eO5i/0tV9VD1cCiNfpIvIusBu4WVU3HCKGxbjjLY0dO7Yf3jZGbLfUkTOYWJJFRGF7TQsnj8jp3/cyxphBLp42hY/EPOYDtwOX9sN7rwbGqepM4GfAM4c6UFUfVNW5qjq3uLi4H946RtEk8PihchXQMzWntSsYY1JRPNVHX4vdFpE8YOnxvrGqNsasvyAi/yUiRapac7znPiqBTCibC9tXAHBCcSYewe5sNsakpHhKCgdqAY6pnSGWiIwQEXHX57mx1B7veY9J+XzYswbaGwj6vYwtyLAxkIwxKSmeNoU/4vQ2AueLewrwhzhe9zhwDlAkIhXA9wA/gKr+ArgcWCIiIaANuDJ2hrcBNf4sWHEX7PwbnHQhE0uyrfrIGJOS4umS+pOY9RCwU1UrjvQiVb3qCM/fj9NlNfnKPgK+IGx/DU66kEmlWfx18z66whH83mMpTBljzNAUT1L4ENijqu0AIpIuIuWquiOhkQ0kfxDGzIu2K0wqySIUUXbWtjDRbXg2xphUEM/P4P8LRGK2w+6+4WX8WVC1DlpqrQeSMSZlxZMUfKra2b3hrgcSF1KSlJ/lLHe+zoSSTMB6IBljUk88SaFaRKL3JYjIImBgu40OhNFzwJ8J21eQEfBRlp9uScEYk3LiaVO4AXhMRLobhSuAPu9yHtK8fhj30Wi7woml2WyxMZCMMSkmnpvXPgBOE5Esd3v4/nwePx9efhma9jKpJIvXt9YQCkfwWQ8kY0yKOOK3nYj8SETyVLVZVZtFJF9EfjAQwQ248W67wvbXmFiSRWcowq66tuTGZIwxAyien8AXqmp994Y7C9tFiQspiUbMgGAubH+VSaXdPZCsCskYkzriSQpeEUnr3hCRdCDtMMcPXR4vjDsTdjglBbAeSMaY1BJPQ/NjwCsi8jAgwLXAI4kMKqnGnwWbnyerbTfjCjNYtbMu2REZY8yAOWJJQVV/DPwAmAycBCwDxiU4ruSJaVc4f3Ipr2+poam9K7kxGWPMAIm3W00VzqB4nwHOBTYlLKJkK5kMGUWwfQULp42gMxxh+ebqZEdljDED4pBJQUROFJHvich7OJPgfAiIqi5wB7MbnkScrqnbVzBnTB5FWWksW7832VEZY8yAOFxJ4T2cUsEnVPVMVf0ZzrhHw1/5fGjajbd+OxdMLWX55n20d6XGpRtjUtvhksKngD3AchF5SETOw2loHv7Gn+0st7/KwmkjaO0M89qW4TeyhzHGHOiQSUFVn1HVK4GTgeXATUCJiDwgIh8fqACTonACZI+C7Ss47YRCcoI+Xly/J9lRGWNMwsXT+6hFVX+nqpcAZcA7wK0JjyyZou0Kr+H3COdPKeX/bayiKxw58muNMWYIO6pBfVS1TlUfVNXzEhXQoDH+LGitgX2bWDh1BI3tId7YlpwppI0xZqDYSG+HEr1fYQVnnVhMRsDLS9YLyRgzzFlSOJS8sVB8Mqz5LUGfhwUnlbBsQxXhiCY7MmOMSRhLCodzxk2wdx1sfoELpo2gprmD1R/asBfGmOHLksLhTP8M5I+HV3/MghOLCHg9VoVkjBnWLCkcjtcHZ90Me94l+8O/MH9SES+t34uqVSEZY4YnSwpHMuMKyBsHr97JBVNLqaxvY8PuxmRHZYwxCWFJ4Ui8fpj/Ldj9DhcG1+P1iN3IZowZthKWFETk1yKyT0TWH+J5EZH7RGSriKwVkTmJiuW4zbwKcseS/cZ/cmp5vrUrGGOGrUSWFH4DLDzM8xcCk9zHYuCBBMZyfHwBmP9NqFzJF0du44PqFrbus2k6jTHDT8KSgqquAPYf5pBFwKPqeAPIE5GRiYrnuM26GnLKOGv3rwG10oIxZlhKZpvCaGBXzHaFu+8gIrJYRFaKyMrq6iRNeOMLwPxvENizkmtH7OCpdyrtRjZjzLAzJBqa3fGW5qrq3OLi4uQFMvvzkD2KG71Psa26mSdXVSQvFmOMSYBkJoVKYEzMdpm7b/DypcGZ36CgdhXXlH7I3S+/b5PvGGOGlWQmheeAL7i9kE4DGlR18Pf1nPMFyB7Jbb7H2d/YxG/+tiPZERljTL9JZJfUx4G/AyeJSIWIXC8iN4jIDe4hLwDbgK3AQ8A/JiqWfuUPwsI7yapdy0NFv+e/lm+hvrUz2VEZY0y/8CXqxKp61RGeV+CriXr/hJr6Sdj7Lc5+7T+5tKuUB/46jn+5aHKyozLGmOM2JBqaB6UF34FJF/Bv/kdZ97cXqaxvS3ZExhhz3CwpHCuPBz79EJH88fzMew+/eX5FsiMyxpjjZknheARz8V+9lExfmEWb/5nNu6qSHZExxhwXSwrHq2gSoU8+xBTPTuqXfgVsWG1jzBBmSaEfZE2/mLfG/xOntiznwz/+KNnhGGPMMbOk0E9mXXU7L3vOoGz1fxDe8kqywzHGmGNiSaGfBAM+2i68l82RMjp/fy3U7Uh2SMYYc9QsKfSjS+ZO4tmT7qKzK0TjI1dCZ2uyQzLGmKNiSaEfiQg3ffYC7sm5hay692h68mvW8GyMGVIsKfSzoN/L9dfdwC88nyF78xN0/v0XyQ7JGGPiZkkhAcYUZDDtyh/wSng23j//K7rzb8kOyRhj4mJJIUHOOqmUD+bfzYeRItoe+zw0Dv4BYI0xxpJCAn3p/Nk8POaH0NFE02+vhi4bH8kYM7hZUkggj0f41jWf5K7g18jet4rOhz4ODTZbmzFm8LKkkGC56X6uuu4mbuRWOvdtJfTAWbDj9WSHZYwxfbKkMABOGpHNV2/4Gtf67uTD9iD6yKXwxi+su6oxZtCxpDBAThqRzX8uuZwl6XexPDIbXroVnlli7QzGmEElYTOvmYONK8zkkRvO5wu/zGR9/WPc+O7jsG8TzL0Oskohq8RZZhaDLy3Z4RpjUpAlhQE2IjfI0hvO4NqH/azfM477ax4k8MevH3xgMA9O/gRc8ENIzxv4QI0xKcmSQhIUZAZ47Eun8qVHvEzbMYN/PjOfa2dk4muthuYqaN7nDKj37uOwbTks+jlMWJDssI0xKUB0iDV2zp07V1euXJnsMPpFe1eYf3lqHU+/U8n00bnc/dmZTCrN7jmgchU89RWo3QLzvgLn3w6BjGSFa4wZwkRklarOPdJx1tCcREG/l3uumMUDV8+hsr6Ni3/2Ov/96geEI26iHn0KfGUFnHoDvPXf8N9nOYnCGGMSxEoKg0RNcwf/+vQ6lm2o4pRx+fznZ2ZSXpTZc8AHy+HZr0LTXpj8CUCcnktdrT1LbwBOugimfRqKJibtWowxg0+8JQVLCoOIqvLsmt1899n1dIWVWxeexBdOL8fjEeeAtnpY9q+w/VXwp7uPzJ711lr48A1AYeRMmHY5TPsU5JYl9bqMMck3KJKCiCwEfgp4gV+q6p0HPH8t8B9ApbvrflX95eHOOZyTQre9De3c9tRa/rq5mo+U53Pnp2cwoTgrvhc3VMKGp2H9E7D7HWffmFNh9FwoORmKJ0PxSRDMSdwFGGMGnaQnBRHxAu8DHwMqgLeBq1R1Y8wx1wJzVfWf4j1vKiQFcEoNT62u5I4/baStK8w3zj+RL88fj897FM1AtR/A+qdg8/PO/RCh9p7ncsqcJJE7puceiewRMeujwGud04wZLuJNCon8Xz8P2Kqq29yAlgKLgI2HfZUBnFncPn1KGfNPLOL/PLOeH7/0Hi+s28Ndl89g8sg4f+UXToCzb3EekbDTzbX6PSdBVL/nPHavgdaag1/rS4dRs5zG7rK5TkkjtwxE+vU6jTGDSyJLCpcDC1X1S+7254FTY0sFbknh34FqnFLFN1R11+HOmyolhViqygvr9vLdZ9fT0NbF4rNO4CtnTSA3w98/bxDugpaannskmvc6iaNiJex5F8IdznFZpVAyBTIKIaMA0gt61n1BaNrjjALbUAGNlc6yeR+UToHyM2HcmTDudAjm9n7/pirY9QZ8+CZUvAXZI+Hj34f88v65PmPMoKg+iicpFALNqtohIl8BrlDVc/s412JgMcDYsWNP2blzZ0JiHuz2t3Ty/T9t5Ol3KslO83HdmeO5/ozx/Zcc+hLqhKr1TlfYipXOPROt+51HR8PBx3v8kDPKqZbKHe0kjT3vQsXbEO4E8cCIGTDuo845dr3hlGAAvGkwarbzfpEwnHMbnP5V8Cbw+oxJEYMhKZwO3K6qF7jb/wKgqv9+iOO9wH5Vze3r+W6pWFI40KY9jdz3yhZeXL/XSQ5nlHP9mSckNjn0JRyCtjpo2+90ic0eCZkl4Omj3aMZEGHkAAATJElEQVSrzUkqO153HhVvQVoOjD3NaQgfe7rTY8oXcEoYL/yz0xZSOg0+cS+M+cjAXpsxw8xgSAo+nCqh83B6F70NfE5VN8QcM1JV97jrlwG3qupphzuvJYUeByaHz58+jk/NGc3EkuwjvzjZwiHweA/fRrHpT/DCLU611Eeuh/k3Q0s17P8AardC7TZnvXG30yU3mOskmmCOs0zPg8KJUDrV6XU1UHeDhzqdElXOKEjPH5j37KbqlOqySiFvTHyv6WqDzhbILEpsbCapkp4U3CAuAu7F6ZL6a1X9oYjcAaxU1edE5N+BS4EQsB9YoqrvHe6clhQO1p0cXtqwF1WYVJLFRdNHctH0kZxYmoUM5cbhjib4yw+dO7o10vu57FFOY3rOaAi1QXsjdDQ6y/YGaK93qqwAEOfY0qlO6SO/3Pni7O5xFcztnaBUnXO11TmPrjYIZPUknLRsp1pLFeq2Q+VqpyRUuap3O0zJFKc0NPZ0Z5k7JjGN9Z2tsHapM09HzWZn35jTnBsZp37S6VEWq70RtvwZNj0HW152SnpjToWpl8GURU5C67fYWmD/dme94IT+Tc6Nu50bO7cth+0rINQRM+JwiVNyzSqB/HEwao7z736sf//OFtj1FtR/CC37oNkdq6yl2mk78/icz1P0MdKJJbfMue6MgqN/z0jEaePbv935nBWddMyl5kGRFBLBksKhVTW2s2zDXp5fu4e3duxHFSYUZ3LR9JF8ek5Z7zukh5rd7zj/6fPGOV/uBSdA4AjXE4lA/Q6o2uA89q5zlnXbDz7WF3S+PLxpPYlAw4c/vy/d+SLobOrZ7u6xNWKG8+Xx4d+dL5LuY3JGO9VkBSc4pZjCCVAwwfkSPpYvq/pd8PZDsOoRJwmOnAkf+bLzRbL+Kdi30WnHGX+WkyAANv0Rtv3VSZhZpc5ovNkjYONzULXOOWbMaU6CmPQxJ67O7jvnW9z1VidJqwIas4w4d93v3w77tzmP5r29Y84d41x70SQonORcu4adDg+RkLvsctqVPF7nb+zxOe1VHq/zHrvecq6hOwFmFsP4s52SWcs+t8OE++j+24PTOWL0HOffaPQpzg+E7o4SB/79u9qdas7trzmfvcqVTnzd0nIhq7hnuPtIyLn2pr1Osoh09T5fer7z714woefHTCTkJLJQe8+yq9X57Ozf7rS3hWLmXDntH2FhnzXwR2RJIcXta2pn2YYqXly3hze21RJROO2EAq78yFgWThtB0O9NdojJ09Hs/MJsrup5dP9HDnc6/3nTC9yl+/AHnV+KHU3Oo7tUEupweleNPsWpourr3o5I2ElGH77hJIl9m5wvy+7SBIA/A/LHO79kez3GOV84rbUxvcP2OevVm+H9lwB1vthP+0enNBL75Va1ETY8Beue6EmGeeNg8iUw+VIo+0jvNqCaLbDhGdj4jNPgf6yyRrhfgCdAwXhnCU61X80WqHnfWe9sPrbz+9KdzgoTFsAJ50DJ1L7bssBJYrVbnNJc5SpnWb2pd8nTG3BKi90P8faU+MTjlDLGz4fy+U4yyyxxPhOHEok4bW1Ne90veLeqs/YD58u+YRfQx3evx+9UhXaXLvLLnb9f/nhnmTvmmDteWFIwUVWN7TyxqoLfv72LD/e3khP0cdns0Vw5b2z89zyY/hWJON12a7e6Xxbur+r6nc6vw67Ww7/e43O+eKddBvMWQ97Ywx+vCnvXOl9wpdPiK5VUv+8kMW/A+aIKZDrJK5DhLCWmTUgEEGeZUQRpcdyBr9qTkD0+58suunTXNez8mo6EnHaoSMj5Mi844fBfykfS2eJ86e/b5FY1NvRUObY3OKWEUbOdElZf3aiPV1e7c+3egDOhli/oLD2J+7FmScEcJBJR3they9K3dvHS+r10hiNMHZXDZbNHc+nMUZTkHMd/MtN/VJ166rodzqOlxmkEjtaRlzqll0P9MjamD5YUzGHVtXTy7JpKnn6nkncrGvAInDGxiMtmj+aCqSPITLMhLowZTiwpmLht3dccTRAVdW2k+718bEopl8wcxVknFpHmS+H2B2OGCUsK5qhFIsqqD+t4anUlL67fQ31rF9lBHwunjuCSmaP46ITCoxuQzxgzaFhSMMelKxzh9a01/PHd3fx5QxXNHSEKMwN8bEopp51QyLzxBYzKS092mMaYOFlSMP2mvSvMXzdX88e1u1nxfjVN7U5f7bL8dOaNL2BeeQFzywsYX5SJ1zOEb5QzZhizpGASIhxR3tvbyFvb90cftS3OXcPpfi8njshmyshsTh6Rw8kjnOWAj8lkjDmIJQUzIFSVD6pbWL2zjk17G3lvTxOb9jZS39pzN+eInCCTSrM4qTSbE0uzmVSaxaTSbLKsh5MxA2YwTLJjUoCIMLEki4klPTcrqSpVjR1s2tvIpj2NbKlq5v2qJv7njZ10hHruIj2xNIuPTiji9AmFnDa+0EoUxgwCVlIwAyYcUXbtb2VzVROb9zbx9o79vL1jP+1dEURg6qgcPjqhiOmjcxmdn05ZXjpFWWl4rJ3CmONmJQUz6Hg9QnlRJuVFmVwwdQQAnaEI71bU87ettfztgxp+87876Az3lCYCXg8j84KMyk2nLD/deX1hJuVFGZQXZtpNdsb0MyspmEGlvSvMjtoWKuva2F3fRkV9G7vr26msa2VXXRvVTR29ji/OTuOEokxmjclj9th85ozLoyTbhusw5kDW0GyGpZaOEDtqW9hZ28r2mhZ21rbwflUzG3c3RksYYwsymDM2jznj8hlXmElpThql2UHyMvxDe24JY46DVR+ZYSkzzcfUUblMHdV71Mr2rjAbdjewemc9q3bW8b8f1PLMmt29jgl4PZTkpDEiJ0hZfjrji7IYX5zJCW6VlvWGMsZKCmaYUlX2NLSzu76NqsYOqhrbqWpqZ19jB3sb2vlwfyuV9W29XlOSncaYggyKsgIUZqVRlJVGsbtemhPkxNIssoPWQ8oMTVZSMClNRBiVl37YoTi62y+2V7ewraaFbdUt7GloY3tNC2/vqKOutZMDfzOV5adz8ogcJo/M5qQR2ZxUms2I3CBZaT6rmjLDgiUFk7KCfq9753XfEw2FwhH2t3ZS29xJZV0bm6ua2LSnkc17m1i+eR/hSE/GSPd7Kc5OoyQ7LbrMzQiQm+7v9chJ95ET9JMd9JEZ8Fl3WzPoWFIw5hB8Xg8l2UFKsoNMHpnD+VNKo8+1d4XZuq+ZrfuaqWpsp7qpg31NHVQ3dfB+VRP/u7WGxvbQYc7uTFKWneYj200SeRl+CrPSKM5KozAzQFG2s8zPDOD1CF4RPCJ4POARwecRgn4vQb+X9ICXdL/Xxp4yx82SgjHHIOj3Mm10LtNGH3qaxnBEaWrvoqGt96OpPURTe/cyRKO7Xt/ayabdjaxo7ogOOni0Al4PQb+HgsyAW2IJUuyWXoqz0ijIDJATU2rJTfeT7vda1ZeJsqRgTIJ4PUJeRoC8jMBRv7YjFKa2uZOa5g7qW7sIq6KqhCNOslFVQhGlIxShrTNEW1eYts6IuwxR29JJdZMz1MiKLYdPMn6vkJXmIyPgIz3gJSP6cLazAj4y0rxkBnxkpvnIjK47x3QvnXN4yUn347d5N4YsSwrGDEJpPu8RG8qPRntXmOqmDupaO6Mllsa2kLNs76KpvYvWzjBtneHocl9TO60dYVo6Q9FlJM7OillpTnVYfkYguvR6hEa3pNTY3kVjWxeN7SEiqozOS2dsQQZjuh/56YzOT+9pf0nz9ZloQuEIzR2haKkroorPK/g8HvxewesR/F4PAa+H9ICXNJ/HSkVHYEnBmBQQ9HujX7jHStUpmbR0hGhxk0RLR4iWzjCtHSGaO5zthrYQ9W2d1Ld2UdfaSV1rFx/ubyUcUafaKuhnbEEGOe46QIV7x/qb2/fT3NF3qSbN5yE76JRo2rvCNLU7JaSjIeJ0CsgIeKPtMWk+DwGfx126214P4YhTGgtFIs56WAlHlJx0n1slF4xWyxVnp+HziPs3CdMas3Re41xrd0eDnHS/ey1egj7voOpwYEnBGBMXkZ6G7cKsIx9/LFSVutYudu1vZXd9G03tTrLpTjhNHSFaO0IE/V6yYhrpux8eEcIRpSuihMIR50s9rHSEwrR1hWnvLgl1OaWhtq4wnaEIneEIHV0RGtq6nO1QGJ/Hg9cjbsnDKX2IQGV9O2t2NVDb0nFQl+VjFfR73GTlI83vQQBVUPdvElFQlM/NG8eScyb0z5seQkKTgogsBH4KeIFfquqdBzyfBjwKnALUAleo6o5ExmSMGbxEhILMAAWZAWaOyUt2OIfV3WW52u11FlF12lh6tcE4jfhN7U51XU+1mVON1l1d197lLLvXERCcv4fHXfeIMKYg8VPgJiwpiIgX+DnwMaACeFtEnlPVjTGHXQ/UqepEEbkS+DFwRaJiMsaY/hLbZflIstJ8jDx0R7VBJZFdBOYBW1V1m6p2AkuBRQccswh4xF1/AjhPrBXIGGOSJpFJYTSwK2a7wt3X5zGqGgIagMIExmSMMeYwhkRnYhFZLCIrRWRldXV1ssMxxphhK5FJoRIYE7Nd5u7r8xgR8QG5OA3Ovajqg6o6V1XnFhcXJyhcY4wxiUwKbwOTRGS8iASAK4HnDjjmOeAf3PXLgb/oUBvL2xhjhpGE9T5S1ZCI/BOwDKdL6q9VdYOI3AGsVNXngF8B/yMiW4H9OInDGGNMkiT0PgVVfQF44YB9341Zbwc+k8gYjDHGxG9INDQbY4wZGENuOk4RqQZ2HuPLi4CafgxnKEnVa7frTi123Yc2TlWP2FNnyCWF4yEiK+OZo3Q4StVrt+tOLXbdx8+qj4wxxkRZUjDGGBOVaknhwWQHkESpeu123anFrvs4pVSbgjHGmMNLtZKCMcaYw0iZpCAiC0Vks4hsFZHbkh1PoojIr0Vkn4isj9lXICIvi8gWd5mfzBgTQUTGiMhyEdkoIhtE5Ovu/mF97SISFJG3RORd97r/zd0/XkTedD/vv3eHmhl2RMQrIu+IyJ/c7WF/3SKyQ0TWicgaEVnp7uu3z3lKJIWYCX8uBKYAV4nIlORGlTC/ARYesO824BVVnQS84m4PNyHgW6o6BTgN+Kr7bzzcr70DOFdVZwKzgIUichrOhFX3qOpEoA5nQqvh6OvAppjtVLnuBao6K6Ybar99zlMiKRDfhD/DgqquwBlHKlbsZEaPAJ8c0KAGgKruUdXV7noTzhfFaIb5tauj2d30uw8FzsWZuAqG4XUDiEgZcDHwS3dbSIHrPoR++5ynSlKIZ8Kf4axUVfe463uB0mQGk2giUg7MBt4kBa7drUJZA+wDXgY+AOrdiatg+H7e7wX+GYi424WkxnUr8GcRWSUii919/fY5T+iAeGbwUVUVkWHb5UxEsoAngZtUtTF2dtfheu2qGgZmiUge8DRwcpJDSjgR+QSwT1VXicg5yY5ngJ2pqpUiUgK8LCLvxT55vJ/zVCkpxDPhz3BWJSIjAdzlviTHkxAi4sdJCI+p6lPu7pS4dgBVrQeWA6cDee7EVTA8P+9nAJeKyA6c6uBzgZ8y/K8bVa10l/twfgTMox8/56mSFOKZ8Gc4i53M6B+AZ5MYS0K49cm/Ajap6t0xTw3raxeRYreEgIikAx/DaU9ZjjNxFQzD61bVf1HVMlUtx/n//BdVvZphft0ikiki2d3rwMeB9fTj5zxlbl4TkYtw6iC7J/z5YZJDSggReRw4B2fUxCrge8AzwB+AsTgjzH5WVQ9sjB7SRORM4DVgHT11zN/GaVcYttcuIjNwGha9OD/y/qCqd4jICTi/oAuAd4BrVLUjeZEmjlt9dLOqfmK4X7d7fU+7mz7gd6r6QxEppJ8+5ymTFIwxxhxZqlQfGWOMiYMlBWOMMVGWFIwxxkRZUjDGGBNlScEYY0yUJQVjXCISdkee7H702+B5IlIeO3KtMYOVDXNhTI82VZ2V7CCMSSYrKRhzBO749Xe5Y9i/JSIT3f3lIvIXEVkrIq+IyFh3f6mIPO3OcfCuiHzUPZVXRB5y5z34s3sHMiJyozsPxFoRWZqkyzQGsKRgTKz0A6qProh5rkFVpwP349wZD/Az4BFVnQE8Btzn7r8PeNWd42AOsMHdPwn4uapOBeqBT7v7bwNmu+e5IVEXZ0w87I5mY1wi0qyqWX3s34Ezkc02d9C9vapaKCI1wEhV7XL371HVIhGpBspih1dwh/N+2Z0EBRG5FfCr6g9E5CWgGWc4kmdi5kcwZsBZScGY+Ogh1o9G7Bg8YXra9C7GmRlwDvB2zCifxgw4SwrGxOeKmOXf3fW/4YzQCXA1zoB84EyHuASiE+DkHuqkIuIBxqjqcuBWIBc4qLRizECxXyTG9Eh3ZzDr9pKqdndLzReRtTi/9q9y930NeFhEbgGqgevc/V8HHhSR63FKBEuAPfTNC/zWTRwC3OfOi2BMUlibgjFH4LYpzFXVmmTHYkyiWfWRMcaYKCspGGOMibKSgjHGmChLCsYYY6IsKRhjjImypGCMMSbKkoIxxpgoSwrGGGOi/j8pY8vxo8IqdQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14f75ed30>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['loss'], label='Train Loss')\n",
    "plt.plot(history.history['val_loss'], label='Validation Loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x14fafd2b0>"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1491b8630>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['acc'], label='Train Accuracy')\n",
    "plt.plot(history.history['val_acc'], label='Validation Accuracy')\n",
    "plt.plot(history.history['loss'], label='Train Loss')\n",
    "plt.plot(history.history['val_loss'], label='Validation Loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d_2 (Conv2D)            (None, 28, 28, 32)        832       \n",
      "_________________________________________________________________\n",
      "max_pooling2d_2 (MaxPooling2 (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 10, 10, 64)        51264     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_3 (MaxPooling2 (None, 5, 5, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 1600)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 800)               1280800   \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 120)               96120     \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 84)                10164     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 84)                0         \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 36)                3060      \n",
      "=================================================================\n",
      "Total params: 1,442,240\n",
      "Trainable params: 1,442,240\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "modelo_2 =  keras.models.Sequential()\n",
    "modelo_2.add(keras.layers.Conv2D(32, (5, 5), activation='relu', kernel_initializer='he_uniform', padding='valid', input_shape=(32, 32, 1)))\n",
    "modelo_2.add(keras.layers.MaxPooling2D((2, 2)))\n",
    "modelo_2.add(keras.layers.Conv2D(64, (5, 5), activation='relu', kernel_initializer='he_uniform', padding='valid'))\n",
    "modelo_2.add(keras.layers.MaxPooling2D((2, 2)))\n",
    "modelo_2.add(keras.layers.Flatten())\n",
    "modelo_2.add(keras.layers.Dropout(0.2))\n",
    "modelo_2.add(keras.layers.Dense(800, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_2.add(keras.layers.Dense(120, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_2.add(keras.layers.Dense(84, activation='relu', kernel_initializer='he_uniform'))\n",
    "modelo_2.add(keras.layers.Dropout(0.2))\n",
    "modelo_2.add(keras.layers.Dense(36, activation='softmax'))\n",
    "\n",
    "modelo_2.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 19377 samples, validate on 4846 samples\n",
      "Epoch 1/50\n",
      "19377/19377 [==============================] - 36s 2ms/sample - loss: 3.3120 - acc: 0.1287 - val_loss: 2.6014 - val_acc: 0.3933\n",
      "Epoch 2/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 2.4458 - acc: 0.3757 - val_loss: 1.7939 - val_acc: 0.6121\n",
      "Epoch 3/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.8880 - acc: 0.5211 - val_loss: 1.3902 - val_acc: 0.6766\n",
      "Epoch 4/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.5447 - acc: 0.6032 - val_loss: 1.1488 - val_acc: 0.7332\n",
      "Epoch 5/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.3378 - acc: 0.6556 - val_loss: 1.0001 - val_acc: 0.7676\n",
      "Epoch 6/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.1736 - acc: 0.6916 - val_loss: 0.8922 - val_acc: 0.7959\n",
      "Epoch 7/50\n",
      "19377/19377 [==============================] - 35s 2ms/sample - loss: 1.0703 - acc: 0.7214 - val_loss: 0.8279 - val_acc: 0.7988\n",
      "Epoch 8/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.9803 - acc: 0.7411 - val_loss: 0.7470 - val_acc: 0.8205\n",
      "Epoch 9/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.9034 - acc: 0.7591 - val_loss: 0.7024 - val_acc: 0.8215\n",
      "Epoch 10/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.8300 - acc: 0.7792 - val_loss: 0.6492 - val_acc: 0.8411\n",
      "Epoch 11/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.7856 - acc: 0.7876 - val_loss: 0.6237 - val_acc: 0.8362\n",
      "Epoch 12/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.7421 - acc: 0.7988 - val_loss: 0.6033 - val_acc: 0.8438\n",
      "Epoch 13/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.7028 - acc: 0.8075 - val_loss: 0.5652 - val_acc: 0.8529\n",
      "Epoch 14/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.6707 - acc: 0.8171 - val_loss: 0.5526 - val_acc: 0.8576\n",
      "Epoch 15/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.6375 - acc: 0.8260 - val_loss: 0.5184 - val_acc: 0.8671\n",
      "Epoch 16/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.6221 - acc: 0.8286 - val_loss: 0.5099 - val_acc: 0.8675\n",
      "Epoch 17/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.5825 - acc: 0.8386 - val_loss: 0.5090 - val_acc: 0.8665\n",
      "Epoch 18/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.5663 - acc: 0.8433 - val_loss: 0.4810 - val_acc: 0.8774\n",
      "Epoch 19/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.5361 - acc: 0.8510 - val_loss: 0.4855 - val_acc: 0.8712\n",
      "Epoch 20/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.5197 - acc: 0.8556 - val_loss: 0.4495 - val_acc: 0.8867\n",
      "Epoch 21/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.5039 - acc: 0.8593 - val_loss: 0.4308 - val_acc: 0.8875\n",
      "Epoch 22/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4880 - acc: 0.8640 - val_loss: 0.4341 - val_acc: 0.8871\n",
      "Epoch 23/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4643 - acc: 0.8668 - val_loss: 0.4241 - val_acc: 0.8913\n",
      "Epoch 24/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4476 - acc: 0.8720 - val_loss: 0.4121 - val_acc: 0.8919\n",
      "Epoch 25/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4380 - acc: 0.8760 - val_loss: 0.4082 - val_acc: 0.8888\n",
      "Epoch 26/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4192 - acc: 0.8821 - val_loss: 0.3815 - val_acc: 0.8991\n",
      "Epoch 27/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.4148 - acc: 0.8813 - val_loss: 0.3865 - val_acc: 0.8960\n",
      "Epoch 28/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3963 - acc: 0.8872 - val_loss: 0.3863 - val_acc: 0.8923\n",
      "Epoch 29/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3893 - acc: 0.8866 - val_loss: 0.3820 - val_acc: 0.8983\n",
      "Epoch 30/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3755 - acc: 0.8939 - val_loss: 0.3657 - val_acc: 0.9038\n",
      "Epoch 31/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3684 - acc: 0.8951 - val_loss: 0.3601 - val_acc: 0.9059\n",
      "Epoch 32/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3536 - acc: 0.8969 - val_loss: 0.3568 - val_acc: 0.9045\n",
      "Epoch 33/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3476 - acc: 0.8984 - val_loss: 0.3571 - val_acc: 0.9059\n",
      "Epoch 34/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3311 - acc: 0.9037 - val_loss: 0.3387 - val_acc: 0.9094\n",
      "Epoch 35/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3244 - acc: 0.9048 - val_loss: 0.3487 - val_acc: 0.9084\n",
      "Epoch 36/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3148 - acc: 0.9078 - val_loss: 0.3310 - val_acc: 0.9152\n",
      "Epoch 37/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3046 - acc: 0.9104 - val_loss: 0.3280 - val_acc: 0.9146\n",
      "Epoch 38/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.3020 - acc: 0.9094 - val_loss: 0.3256 - val_acc: 0.9162\n",
      "Epoch 39/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2907 - acc: 0.9134 - val_loss: 0.3182 - val_acc: 0.9181\n",
      "Epoch 40/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2891 - acc: 0.9143 - val_loss: 0.3226 - val_acc: 0.9168\n",
      "Epoch 41/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2778 - acc: 0.9178 - val_loss: 0.3145 - val_acc: 0.9175\n",
      "Epoch 42/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2679 - acc: 0.9211 - val_loss: 0.3123 - val_acc: 0.9191\n",
      "Epoch 43/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2636 - acc: 0.9216 - val_loss: 0.3157 - val_acc: 0.9222\n",
      "Epoch 44/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2561 - acc: 0.9251 - val_loss: 0.3077 - val_acc: 0.9245\n",
      "Epoch 45/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2525 - acc: 0.9261 - val_loss: 0.3094 - val_acc: 0.9216\n",
      "Epoch 46/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2471 - acc: 0.9290 - val_loss: 0.2982 - val_acc: 0.9255\n",
      "Epoch 47/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2377 - acc: 0.9292 - val_loss: 0.2956 - val_acc: 0.9274\n",
      "Epoch 48/50\n",
      "19377/19377 [==============================] - 31s 2ms/sample - loss: 0.2358 - acc: 0.9290 - val_loss: 0.2951 - val_acc: 0.9288\n",
      "Epoch 49/50\n",
      "19377/19377 [==============================] - 30s 2ms/sample - loss: 0.2274 - acc: 0.9317 - val_loss: 0.2966 - val_acc: 0.9259\n",
      "Epoch 50/50\n",
      "19377/19377 [==============================] - 32s 2ms/sample - loss: 0.2214 - acc: 0.9335 - val_loss: 0.2923 - val_acc: 0.9307\n"
     ]
    }
   ],
   "source": [
    "opt = keras.optimizers.SGD(lr=0.001, momentum=0.9)\n",
    "modelo_2.compile(optimizer=opt, loss='sparse_categorical_crossentropy',metrics=['accuracy'])\n",
    "\n",
    "history2 = modelo_2.fit(xTrain, yTrain, epochs=50, batch_size=64, validation_data=(xTest, yTest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelo guardado\n"
     ]
    }
   ],
   "source": [
    "model_json = modelo_combinado.to_json()\n",
    "with open(\"model.json\", \"w\") as json_file:\n",
    "    json_file.write(model_json)\n",
    "# serialize weights to HDF5\n",
    "modelo_combinado.save_weights(\"modeloFinalDrop.h5\")\n",
    "print(\"Modelo guardado\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0821 12:02:31.893424 140734893671872 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:97: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
      "W0821 12:02:31.897029 140734893671872 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:97: calling Zeros.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
      "W0821 12:02:32.121144 140734893671872 deprecation.py:506] From /Library/Frameworks/Python.framework/Versions/3.6/lib/python3.6/site-packages/tensorflow/python/ops/init_ops.py:97: calling GlorotUniform.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Modelo cargado desde el computador\n"
     ]
    }
   ],
   "source": [
    "json_file = open('model.json', 'r')\n",
    "loaded_model_json = json_file.read()\n",
    "json_file.close()\n",
    "loaded_model = keras.models.model_from_json(loaded_model_json)\n",
    "#Abriendo pesos del modelo creado\n",
    "loaded_model.load_weights(\"modeloFinalDrop.h5\")\n",
    "print(\"Modelo cargado desde el computador\")\n",
    " \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "#Evaluar el modelo cargado\n",
    "opt = keras.optimizers.SGD(lr=0.001, momentum=0.9)\n",
    "loaded_model.compile(optimizer=opt, loss='sparse_categorical_crossentropy',metrics=['accuracy'])\n",
    "score = loaded_model.evaluate(xTest, yTest, verbose=0)\n",
    "print(\"%s: %.2f%%\" % (loaded_model.metrics_names[1], score[1]*100))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 31ms/sample\n",
      "Caracter: ' 6 '\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAHmdJREFUeJztnWuMXdd13//rnPuY93D41JCiSIqhJSuyJBsDwUWEwE2QQDUCyAYKw/5g6IMRBm0M1ED6QXCB2gWKwglqG/5QuKArIUrh+NHYhoXCaOMICdS0qWzKliiJtMyHSPExM6T4mvfce89Z/XAvU5Le/z2XHM4dyvv/Awje2evuc/bd56xz7tn/u9Yyd4cQIj2y9R6AEGJ9kPMLkShyfiESRc4vRKLI+YVIFDm/EIki5xciUeT8QiSKnF+IRKmsprOZPQngawByAP/F3b8Ue//YiPn2LeR6Yxbb0y22A3Bu89u23UYfauHbAwCLzEeex67Z4Y06ykif2Dgituinu3WyyM6yjH/msuTjYL9gjc8gH0cZm8boAeUdWbcy+uvbsG3qYgtX58qYM/0jt+38ZpYD+E8Afg/AGQA/NbMX3P0w67N9S4a//LOh8PYiB9foMKu0T1lwW9GqUVuzyfu1Wnm4vRluB4DlFjWhiJy01Rofx9DwAN+ohXfYaC3RLpWcn5h5zJbx8WfswlAWtE9frY/aBvv5Z15eXKa2ZqMZbK/l/JgVLX4uLi1yvyo8vC8A8GyR2ppFeE4aLb69FtnXv/wP79I+N7Oar/2PAzjm7ifcvQHg2wCeWsX2hBA9ZDXOvwPA6ev+PtNpE0K8B1jzBT8z229mB83s4OUZRRAKcbewGuc/C2DndX/f22m7AXc/4O4T7j4xNtLVOoQQogesxvl/CmCfme0xsxqATwJ44c4MSwix1tz2ar+7t8zsswD+J9pS33Pu/ma8lyErw7uMLBwjs7DRjK+GmvFVZVRiq7J8Fdhr4WtlWUZWjo1PcRHZl+V8jNUqXzluFOGV7wLztE+lxsdYqXJlpNWMrHwX4X6ZDdI+l2f4vWhyqkFtF6auUNvFC5eD7a0mn9+i4CdjHlGlWmTVHgAKcNuuPSPB9m3jXP2gsugtfLlelc7v7j8C8KPVbEMIsT7oF35CJIqcX4hEkfMLkShyfiESRc4vRKKsarX/1slgTqSeSASTk4goIxIgAERM0VC7LCLJsKAZz7i+UqlGpL6IDNgsuLS1sMClvrwWlg+HhnigUDMSYNRs8cikRiQIyot6sD1DOLALAM6cuUptbx+7QG1nI/0uTIUlziaPBUIW0cvqXH1DXyTeamCIz/HY5vA9eOv2ftqnWmM+cZEP4iZ05xciUeT8QiSKnF+IRJHzC5Eocn4hEqXHq/0VwMaCFnO+ys5sFkkJ5ZFVe4ut6Edz3ZF8cM77lM5X7S1foLYsMh8AX6q2LLxCXK3ygJrGEl/dXo6mNRumtoW58LL47FV+v3n9TT4fr74yHdkXNaFohk/xWo0HLMWCd3yRH+u9m0apbc++DdS2YVNYAanUwooJANTq4cAky87RPjejO78QiSLnFyJR5PxCJIqcX4hEkfMLkShyfiESpcdSH+DscpPxgI+yDMsaRRmR0UgQTnsQ3Fa2+DZZhZpKJVauKyLLRcZfyfihGR7gc7XYCI/xymVesada30Jtg/1bqc0GeL+3j4eluf/10lHa59xZnovvKjchz3hETaUalioXC64P9tf58dy4kctvOyNy3r5HtlPbwEB4f1bjwUAFjVzrPomf7vxCJIqcX4hEkfMLkShyfiESRc4vRKLI+YVIlFVJfWZ2EsAsgAJAy90nYu93lGi2wjnVWs4lsTwn5boiNb6M5P0DgLLkUl8rIgNScSVyCc0iQYLOK0ahKHnHZsFtXgmXfuqvchmqVYQjLQFgcorP8RuH36C2o0fDueROvP0u7TNzhU9IWfBJrvdxGTOvhbc5GJHzdt7H5+rBh8apbcd94bkHgMENkQg9MsbMuE80lsNSZRmNBr2RO6Hz/1N350dUCHFXoq/9QiTKap3fAfy1mb1iZvvvxICEEL1htV/7n3D3s2a2FcCPzewX7v7S9W/oXBT2A8A9m3v+a2IhBGFVd353P9v5/zyAHwB4PPCeA+4+4e4TYyO8Hr0QorfctvOb2aCZDV97DeD3AfDlXyHEXcVqvodvA/ADM7u2nb909/8R71LCiXxRtmZ5N6LKVCOXrjyLfcuIROFZRCrJyTYjEXhwHoEXLVHmfIxlyffnrU3B9sYyl6Gu8ryZeOc0L4X1+uunqe3smbCk21jkCUEHB7kcFivN1mhwSayfRMzt2sVLYb3vAZ6Y9OHf5PM4HCnXVcsuU1tzgcnfXHauVsPnYvcxfatwfnc/AeDR2+0vhFhfJPUJkShyfiESRc4vRKLI+YVIFDm/EInS25/cucNJRJ2Xi7wfq8kXkcMyC9eKaxv5x86ZnAeATVfpfHuVCpeU8grXhvKSy15o8s+2SGyXrnDZ6CevHKe2Xx47T23nprlGOHM1HHlYFlxK3TjG52p4mM/HcoMn47xnW3ibH3psJ+2zezeX+rZt4uPwSFLQssXP7z5y+ngkXLRVkkhX2uNX0Z1fiESR8wuRKHJ+IRJFzi9Eosj5hUiUnq72m2Xoq4ZXX7Oc56XLEF4pzSLBHhbLc0drhgHwGjeV4aCUpvNglWXSpw1f3c6yQWpbanKV4NCRd4LtvyTlswDgrbfC+fYAYPoCX6VuRfLq5STwpFLhq/3z87wmV63GP/P7I3n19u0N5yfcdz8P0BnbwM+reoUrHMuNGWrzJlcC6gNhBcEiwWlLzfD5HQuAuhnd+YVIFDm/EIki5xciUeT8QiSKnF+IRJHzC5EoPZX6MsvRR8pJ1TMuiXkrfI0qCl7eqdWK5OmLSHMZuMSGjEgyeSR5Wx+XlGbm+RhnZxrUdubsOWr7PwfPBNuPvMWlvjISDdKMSEdlxmW7WiV8ag3288CYgTrf184dPNjmiSceoLbd94WPjbV4wFIOnk/Sm1z6tJL3q1qkpBjCQVeZcdm5nod9wmhRucD2u36nEOLXCjm/EIki5xciUeT8QiSKnF+IRJHzC5EoK0p9ZvYcgD8AcN7dH+60bQTwHQC7AZwE8Al35/WIOrgDBamsVMYuQwWJbuKBezDwiKiKcTkvM15eq/CwTNVocUlmnit2OHryXWo7feYCtR0+EpbzAOA82eRiwfW8pRafyDxSE22AlMICgMzC26z1cynq4UfuobYH3xcuQwYAI2NcRmuV4VJYFeMRhG6R+mXGpb5qJSIDZlyWzkjZtozlrgRQQficu9NS358DePKmtmcAvOju+wC82PlbCPEeYkXnd/eXAFy6qfkpAM93Xj8P4GN3eFxCiDXmdp/5t7n7ZOf1FNoVe4UQ7yFWveDn7g7wBw0z229mB83s4OUZnjteCNFbbtf5p81sHAA6/9MfSrv7AXefcPeJsZHe1ggRQnBu1/lfAPB05/XTAH54Z4YjhOgV3Uh93wLwEQCbzewMgC8A+BKA75rZZwCcAvCJbnZmyJAjXE4qjwylScLOCnAdrRWputUyHtXXIuMDgFYZjiy7PMelsmMRye6tE9w2Nc0jxE6e4xKQVcNyZN8olyMbc1yGKp0/qhU5l/ruuzcsze25bwPt89CDXOrbOc6PSyXjyTGLVljqy7Pbk/MsEp2XZfx8NESOGdOsI6pdTkp53YrUt6Lzu/uniOl3u96LEOKuQ7/wEyJR5PxCJIqcX4hEkfMLkShyfiESpce/uimBPCyjLEWScc4vhGWvwrjUlPfxpJp5P08GCRulposXw/rhL09cpX3+7h94BN656ZtDJv4/zYJLQ0sNLi0uz4fDJht8c7HgSGwY5RLb+A4u273/A/cF2x9631baZ2wwEsVW5xJbLZJIlMlo7pFajhG5rIxIdgA/h/NIvwxh6ZlFRgKAs8SqkWSsv7pfIUSSyPmFSBQ5vxCJIucXIlHk/EIkipxfiETpqdTnVqLMSMRUJBgp6wvLTVmkRl6ln0t2SwW3nb/E5ZWfv0bq4B29yLc3zSWqxQU+/aVzW0SlQoVczys1vr1qjUf8jY3whKZjw2PUNjI4Hmzvr2+mfSrs3ABgJY/Cs5xHaVar/cH22LnjLT6OVhHeHhCPFMzySCIbC0vIMVGRKcEkF2h4TN2/VQjx64ScX4hEkfMLkShyfiESRc4vRKL0drXfSywW4ZVU80iOtoHwqrIbD9BZaPGSXG+fCud1A4C3jvGqY6/8/HSw/dRpHtiz0IwEH1X4KnVe4UkIazm3NUmJp0aDr2AvzPPxz1yZprbzk3wcl86/E2w/fpQHA22OBBFtHuO2DSPhvIUAsGVrWMnYtJmPY6DO1aBKlc9juczPnYVlnpPRW+GAoHqdqzD9/eH5sIwrTzejO78QiSLnFyJR5PxCJIqcX4hEkfMLkShyfiESpZtyXc8B+AMA59394U7bFwH8IYBr9aY+7+4/WmlbpRuWGmF5KMt4wITVwrLM7DyXQs5N8YCanxwMS3YA8MbhSWq7cDFcjmlpmUfaZBUu9dEyTQBQ8H5EzQMAOLFZya/zVeeSYyzgqrnIjcfeejfYfipSoqxe5595oJ/bRof4ebBtW1gS271rhPZ5cN+91HbvPVwGHOzj48gqXHpeWp4JGyKxQBVSVs5v4X7ezTv/HMCTgfavuvtjnX8rOr4Q4u5iRed395cA8DSzQoj3JKt55v+smR0ys+fMjAd2CyHuSm7X+b8OYC+AxwBMAvgye6OZ7Tezg2Z28MpsLD2BEKKX3Jbzu/u0uxfernzwDQCPR957wN0n3H1iwzD/LbgQorfclvOb2fU5mj4O4I07MxwhRK/oRur7FoCPANhsZmcAfAHAR8zsMbSFoJMA/qibnblnaJJou0bBZZJLs2FJ7PRZvg55/PgVajv8Cx75FJMInZRVyiv8GloaL+FUluHSWkBbFmVk4N+gKhYeY8X4GC3nc1+J5PdDxscxOx+WrxrLYbkUACzjYywiyelmFvkxm74Ytp08yaPsTh7ntvfv4+XGHtjL8xNu3cQjUGu1sMxdlDz6dGYufF4VZfdJ/FZ0fnf/VKD52a73IIS4K9Ev/IRIFDm/EIki5xciUeT8QiSKnF+IROlpAs8sr2BgKCyVLFzmEW5nJ8OSx2uHeHLJI7/g0WPzC/yXhlkWiXCz8HRFFC9kOb++Nhs8bKts8fmIbBJ9JDKuaHLJkZV+AoCqRSTHmMQ5EI46W2pG5iMykPkmn49WJAKyj+yuydVBzM9ymfjCFJffLkzxBJ6PPryT2t7/QNiWV7h7zi2Gz+8yEih6M7rzC5Eocn4hEkXOL0SiyPmFSBQ5vxCJIucXIlF6KvUVBXB1PiznHP4Fj9D7vy9PBdunzi/QPpcux6Lz+DUvi0SWZSxCL+ORVJUa1162jQ9Q20AfPzTb7+HRYw/s+41g+8gIT1gZSwg6t8AlwiuzfI4vXAofm7MXeF3Ad86Fk34CwOwCl0VbJZf6nCRXXV7iEZWNJS7nNWL9lvn4s5wnqB0e3RRs37iJnwNVUk/QrPucGbrzC5Eocn4hEkXOL0SiyPmFSBQ5vxCJ0tPVfnfHMglYmVvkq8pzC+EV1sVlvgLskctaEQkgaSckDlOthFdSh4b5NI6NhQNcAODRR/dS2/btfEV/41id2gaIqVrhK/OVnAczlc731Sx4CarZpfAK9skzPJcdqjy/39G3w4oPAMxcXqS2rBVWVKoZV1qqdX48LeMK09wiVwLOv8vHeG4qrIBkNT6/o2PhY+bgysevbL/rdwohfq2Q8wuRKHJ+IRJFzi9Eosj5hUgUOb8QidJNua6dAP4CwDa0y3MdcPevmdlGAN8BsBvtkl2fcHeexAztUkJXF8JyzuVZLoVcJVJfK1KaKM95gENmt5cfr78WNt6/KyxrAcAHHhmntvv3bqS2DaM8EKRa5ZJSqxkOkMoiEmbsDlA6L9dVJ2WmAGBoMBxINDLA+2zd8AC1bernx/N//8Nb1DZbhKW5VuQcqMdKlBmXbuebXKp89yqXWt+ZDJeP6x/h53ff8FCwvftiXd3d+VsA/sTdHwLwYQB/bGYPAXgGwIvuvg/Ai52/hRDvEVZ0fnefdPefdV7PAjgCYAeApwA833nb8wA+tlaDFELceW7pmd/MdgP4IICXAWxz98mOaQrtxwIhxHuErp3fzIYAfA/A59z9hvrL7u4gjxtmtt/MDprZwZk5/nNcIURv6cr5zayKtuN/092/32meNrPxjn0cwPlQX3c/4O4T7j4xMtTTUAIhRIQVnd/MDMCzAI64+1euM70A4OnO66cB/PDOD08IsVZ0cyv+LQCfBvC6mb3aafs8gC8B+K6ZfQbAKQCfWGlDRemYXQhLHldmuUwyvxiWZVplRM7LeHRTHilBVckiUX3kUrlphEe+7djKpa2+bJbaigYvGeUt3g9lWDaqVPlnrkZk0WaD3x8cYbkJAGp5OHLS6hton82jfB43RSInN4/wqERH2La4zPssL/FzsZpzW38fF9oK51Grs/NzwfalBo/qg/G575YVnd/d/x6gcYK/u+oRCCHWBf3CT4hEkfMLkShyfiESRc4vRKLI+YVIlN4m8ISj2QpLQI0mT6pZkOi90rm0YpFEhnkWkQgjcVFsi/Uql436a1y+qmX8F49ZwSWl3CJSVC08j4N9/DNXMj73cy2+r1ZEBrQsXPKqFpEHhytc2hqKlLsaiSS6nCVD5DF2QCtyLlbAj1m9ys+dOjkuAAAmAxaRGL2CHE9XAk8hxArI+YVIFDm/EIki5xciUeT8QiSKnF+IROmp1GcAaha+3uQRiY3JbzFVIyb1xUxlyY0tYltu8kjAhSUuDVXrkUi7iHyYObeB1NZrRsboWSQRKngyS6tyGbOShfstN/hnbkb0t3KZ98sKPsZKHpbRMuNJUPOMR+CZRyTpRmwe+X22vxqW7fpq/DjX8/BntkjE6s3ozi9Eosj5hUgUOb8QiSLnFyJR5PxCJEqPV/sNNVIPyyKr7EYWUbPYsn2Egi98A5ESYA3S8cIVvnJ8ajJcLgoA7osEsgwP8UAWN37YSnJIlxt8KT2P5C3srw9TWzXnQTqWjQbb567y1fLT56ao7eTpSWo7M3mB2pok8WIljwRwRfL0ZZHVfm/y87GW82M2NBheue+v8zFWsvA5Z4id3DeiO78QiSLnFyJR5PxCJIqcX4hEkfMLkShyfiESZUWpz8x2AvgLtEtwO4AD7v41M/sigD8EcE1n+by7/yi2rSwDBurh683IIJc1RokiNrPAZReSKhAAUMQkwki00OJyOEjnzSPnaJ9LVxepbWJiL7Xdsy0slQHAxg1j1DY6FLaVLS45ZsblIcu41Fcs8nvH3Hx4rs6cnQm2A8CJty9R28W5SGkzUhoMAJoFC9Lhn7kvkouvVuHnx0g/z5N4zxZ+PHft2Bps37SBB05VScCSMV08QDc6fwvAn7j7z8xsGMArZvbjju2r7v4fu96bEOKuoZtafZMAJjuvZ83sCIAdaz0wIcTackvP/Ga2G8AHAbzcafqsmR0ys+fMjH8XFULcdXTt/GY2BOB7AD7n7jMAvg5gL4DH0P5m8GXSb7+ZHTSzgzNzPLGFEKK3dOX8ZlZF2/G/6e7fBwB3n3b3wt1LAN8A8Hior7sfcPcJd58YGeppKIEQIsKKzm/tvEDPAjji7l+5rn38urd9HMAbd354Qoi1optb8W8B+DSA183s1U7b5wF8ysweQ1v+Owngj1baUGbAAIlU2rmd52Gbn+sLtp94m0fTXb7KHzGaJd9XUYT3BQBlGZZymqx0EoB33uGS0oULx6itTiRRABgZ5bndNm8Oj390iPcZGeKS0sgwl9GWFvn8T0+fD7a/c+Yq7TM5FS7xBQBFyeejGZFnScpIWKxPpAzc6DA/1jvG+bmzZ88Gatu5c0uwfbCf5xIEkzC7V/q6Wu3/e4RTXkY1fSHE3Y1+4SdEosj5hUgUOb8QiSLnFyJR5PxCJEpPf3WTZY7+/nByxEcf2Un7jW0MSyH1Pp7U8fgJHj02Pc0TNFYqXMrJqwPB9sL5NBaRbKEXr/BIu8K5VFl5l+s5x0+FbXmVj6NS5bJXpcI/WysSOrm4EE4Y6pEEqbF0rLVI6SqrRaILyRQPD/Ptjd8TPs4AsGcnT7q6O2LbsZ0nO+0nCmHR5Odpi0h9sfm9Gd35hUgUOb8QiSLnFyJR5PxCJIqcX4hEkfMLkSg9lvpK9PeFtZdNgzzB4fDwxmD70CCPoto8xuu3HX7zXWq7eInLK/MLRD60Edonr/Oae7V+LjeVkZpwjSaPfms0wwlDy2Yko2nG5aEs45FleaQf22bGldQoS00uffb18SjNfQ9sDrZvj0hvu3by47lzPCLnbePHuq8SmUcLJyfNci5+Zgif+1nWff1K3fmFSBQ5vxCJIucXIlHk/EIkipxfiESR8wuRKL3Npe0F0AoncGxw9Qq1PCy97L6X15EbG7qX2vaShIkAcO4cH8jZc3PB9lPnwhFsAHBl5gq1lTFZxrhtkCRBBYDCw1rackQqK8qIDBghr/J7R5XY+gZ4stBaneuAsWjLLZu5NPebD+4Kto9v4ZF7I3xzGOrj0ZGDdT6P1uLnCFphebls8c/sZXgcHkk+ejO68wuRKHJ+IRJFzi9Eosj5hUgUOb8QibLiar+Z9QF4CUC98/6/cvcvmNkeAN8GsAnAKwA+7e48KgZAJc+waTS8Qr+4yLsWRVghGMj5yuvQFr5ku30Tt32ABIIAwGw4/gJT5/lK7sVLPE/f5Rleump5mZfCahR8rhpFeFW/FVnRr9b5CnzfAA9WyXJ++lhOVvv7eRBOH0tmB2BgMLI6P8T7jQ2EVZN6LZIjsRIOjgKAHPy4ZC1us5IfM1YeLM/5Z87ycFCYRVSiX9lGF+9ZBvA77v4o2uW4nzSzDwP4UwBfdfffAHAZwGe63qsQYt1Z0fm9zTWBu9r55wB+B8BfddqfB/CxNRmhEGJN6OqZ38zyToXe8wB+DOA4gCvu/5hf+gyAHWszRCHEWtCV87t74e6PAbgXwOMAHux2B2a238wOmtnBS1f5M7oQorfc0mq/u18B8LcA/gmADWZ2bcXnXgBnSZ8D7j7h7hMbRyUuCHG3sKI3mtkWM9vQed0P4PcAHEH7IvDPO297GsAP12qQQog7TzeBPeMAnjezHO2LxXfd/b+b2WEA3zazfw/g5wCeXWlD7kC5HP7qP1jlElCzEZapFuZ5nr7SL1FbZnxfWc5lo9H+sCQ2uJ0HYOyN2PLaGLVZJOjHIkEuBQn4WFrmUlMZLZTF99WMPMWVHr6vVKtcVswj0mFMqgS4NFexsIxmxuVZ84jUZ3wec/A8fVnGJyvPwp87QyRIh8iDt6D0rez87n4IwAcD7SfQfv4XQrwH0UO4EIki5xciUeT8QiSKnF+IRJHzC5Eodis5v1a9M7MLAE51/twMgNfN6h0ax41oHDfyXhvHLnfnSSqvo6fOf8OOzQ66+8S67Fzj0Dg0Dn3tFyJV5PxCJMp6Ov+Bddz39WgcN6Jx3Miv7TjW7ZlfCLG+6Gu/EImyLs5vZk+a2VtmdszMnlmPMXTGcdLMXjezV83sYA/3+5yZnTezN65r22hmPzazo53/ecjf2o7ji2Z2tjMnr5rZR3swjp1m9rdmdtjM3jSzf9Vp7+mcRMbR0zkxsz4z+4mZvdYZx7/rtO8xs5c7fvMds0h4aje4e0//oR0jehzA/QBqAF4D8FCvx9EZy0kAm9dhv78N4EMA3riu7c8APNN5/QyAP12ncXwRwL/u8XyMA/hQ5/UwgF8CeKjXcxIZR0/nBIABGOq8rgJ4GcCHAXwXwCc77f8ZwL9YzX7W487/OIBj7n7C26m+vw3gqXUYx7rh7i8BuDnhwFNoJ0IFepQQlYyj57j7pLv/rPN6Fu1kMTvQ4zmJjKOneJs1T5q7Hs6/A8Dp6/5ez+SfDuCvzewVM9u/TmO4xjZ3n+y8ngKwbR3H8lkzO9R5LFjzx4/rMbPdaOePeBnrOCc3jQPo8Zz0Imlu6gt+T7j7hwD8MwB/bGa/vd4DAtpXfiCWxmVN+TqAvWjXaJgE8OVe7djMhgB8D8Dn3H3melsv5yQwjp7Pia8iaW63rIfznwWw87q/afLPtcbdz3b+Pw/gB1jfzETTZjYOAJ3/z6/HINx9unPilQC+gR7NiZlV0Xa4b7r79zvNPZ+T0DjWa046+77lpLndsh7O/1MA+zorlzUAnwTwQq8HYWaDZjZ87TWA3wfwRrzXmvIC2olQgXVMiHrN2Tp8HD2YE2vXmHoWwBF3/8p1pp7OCRtHr+ekZ0lze7WCedNq5kfRXkk9DuDfrNMY7kdbaXgNwJu9HAeAb6H99bGJ9rPbZ9CuefgigKMA/gbAxnUax38F8DqAQ2g733gPxvEE2l/pDwF4tfPvo72ek8g4ejonAB5BOynuIbQvNP/2unP2JwCOAfhvAOqr2Y9+4SdEoqS+4CdEssj5hUgUOb8QiSLnFyJR5PxCJIqcX4hEkfMLkShyfiES5f8BN7z1l7GFiqkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x146684470>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = cv.cvtColor(cv.imread(\"pruebas/Letras/6.png\"),cv.COLOR_BGR2RGB)\n",
    "img = cv.resize(img, (32, 32))\n",
    "imgG = cv.cvtColor(img, cv.COLOR_BGR2GRAY)\n",
    "imgG = tf.contrib.keras.preprocessing.image.img_to_array(imgG)\n",
    "imgG = np.expand_dims(imgG, axis=0);\n",
    "plt.imshow(img,cmap='gray');\n",
    "\n",
    "features_res_classes = loaded_model.predict_classes(imgG, verbose=1)\n",
    "print(\"Caracter: '\",clases[features_res_classes[0]],\"'\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 9ms/sample\n",
      "Caracter: ' 0 '\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15162b240>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = cv.cvtColor(cv.imread(\"pruebas/Letras/0.png\"),cv.COLOR_BGR2RGB)\n",
    "img = cv.resize(img, (32, 32))\n",
    "imgG = cv.cvtColor(img, cv.COLOR_BGR2GRAY)\n",
    "imgG = tf.contrib.keras.preprocessing.image.img_to_array(imgG)\n",
    "imgG = np.expand_dims(imgG, axis=0);\n",
    "plt.imshow(img,cmap='gray');\n",
    "\n",
    "features_res_classes = loaded_model.predict_classes(imgG, verbose=1)\n",
    "print(\"Caracter: '\",clases[features_res_classes[0]],\"'\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 4ms/sample\n",
      "Caracter: ' C '\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x15216e588>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = cv.cvtColor(cv.imread(\"pruebas/Letras/C.png\"),cv.COLOR_BGR2RGB)\n",
    "img = cv.resize(img, (32, 32))\n",
    "imgG = cv.cvtColor(img, cv.COLOR_BGR2GRAY)\n",
    "imgG = tf.contrib.keras.preprocessing.image.img_to_array(imgG)\n",
    "imgG = np.expand_dims(imgG, axis=0);\n",
    "plt.imshow(img,cmap='gray');\n",
    "\n",
    "features_res_classes = loaded_model.predict_classes(imgG, verbose=1)\n",
    "print(\"Caracter: '\",clases[features_res_classes[0]],\"'\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 2ms/sample\n",
      "Caracter: ' I '\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x140d1c2b0>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = cv.cvtColor(cv.imread(\"pruebas/Letras/I.png\"),cv.COLOR_BGR2RGB)\n",
    "img = cv.resize(img, (32, 32))\n",
    "imgG = cv.cvtColor(img, cv.COLOR_BGR2GRAY)\n",
    "imgG = tf.contrib.keras.preprocessing.image.img_to_array(imgG)\n",
    "imgG = np.expand_dims(imgG, axis=0);\n",
    "plt.imshow(img,cmap='gray');\n",
    "\n",
    "features_res_classes = loaded_model.predict_classes(imgG, verbose=1)\n",
    "print(\"Caracter: '\",clases[features_res_classes[0]],\"'\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### CNN\n",
    "Resultados de la red convolucional"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x1516724e0>"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1512a6b00>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history2.history['acc'], label='Train Accuracy')\n",
    "plt.plot(history2.history['val_acc'], label='Validation Accuracy')\n",
    "plt.plot(history2.history['loss'], label='Train Loss')\n",
    "plt.plot(history2.history['val_loss'], label='Validation Loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.legend()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(19377, 32, 32, 1)\n"
     ]
    }
   ],
   "source": [
    "print(xTrain.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "19377 4846\n"
     ]
    }
   ],
   "source": [
    "print(len(imagenesTrain),len(imagenesTest))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "imagenes = []\n",
    "clases = []\n",
    "\n",
    "for i in imagenesTrain:\n",
    "    imgFlat = i[0].reshape(32*32)\n",
    "    imagenes.append(imgFlat)\n",
    "    clases.append(i[1])\n",
    "    \n",
    "for i in imagenesTest:\n",
    "    imgFlat = i[0].reshape(32*32)\n",
    "    imagenes.append(imgFlat)\n",
    "    clases.append(i[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "imagenes = np.array(imagenes)\n",
    "clases = np.array(clases)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(24223, 1024) (24223,)\n"
     ]
    }
   ],
   "source": [
    "print(imagenes.shape, clases.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[29 15  0 ...  9 11  0]\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "test_size = 0.2\n",
    "X_train, X_test, y_train, y_test = train_test_split(imagenes,clases,test_size=test_size)\n",
    "print(y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "RandomForestClassifier(bootstrap=True, class_weight=None, criterion='gini',\n",
       "                       max_depth=None, max_features='auto', max_leaf_nodes=None,\n",
       "                       min_impurity_decrease=0.0, min_impurity_split=None,\n",
       "                       min_samples_leaf=1, min_samples_split=2,\n",
       "                       min_weight_fraction_leaf=0.0, n_estimators=100,\n",
       "                       n_jobs=None, oob_score=False, random_state=None,\n",
       "                       verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestClassifier\n",
    "\n",
    "modelo_RF = RandomForestClassifier(n_estimators=100)\n",
    "\n",
    "modelo_RF.fit(X_train,y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision para las de Test con Random Forest: 0.8771929824561403\n",
      "Precision para las de Train con Random Forest: 0.9941686448549902\n"
     ]
    }
   ],
   "source": [
    "print(\"Precision para las de Test con Random Forest:\",modelo_RF.score(X_test, y_test))\n",
    "print(\"Precision para las de Train con Random Forest:\",modelo_RF.score(X_train, y_train))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## KNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precision para las de Test con KNN: 0.743859649122807\n",
      "Precision para las de Train con KNN: 0.8480235318402312\n"
     ]
    }
   ],
   "source": [
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "\n",
    "neigh = KNeighborsClassifier(n_neighbors=5)\n",
    "\n",
    "neigh.fit(X_train, y_train)\n",
    "print(\"Precision para las de Test con KNN:\",neigh.score(X_test, y_test))\n",
    "print(\"Precision para las de Train con KNN:\",neigh.score(X_train, y_train))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}