resumary.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "##           PROJET D'IDENTIFICATION D’UNE LANGUE "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "####   “RÉSEAUX DE NEURONES POUR LA RECONNAISSANCE DE L’ORAL ET APPLICATION LINGUISTIQUES”\n",
    "#####                                Fabienne Auffret, Marya Borovikova & Sheherazade Nineb\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "En plus de notre étude appliquant ce qui a été vu en cours, pour étendre nos connaisances et en guise de recherche, deux modèles d'accès libres sur le net ont été testés: \n",
    "Deep Learning for Audio Classification (kapre) et Classification d'images (fleurs)\n",
    "\n",
    "\n",
    "L'un orienté classification audio et l'autre classification d'images. \n",
    "Cependant, une fois les audios traités ces deux problématiques deviennent équivalentes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Voici les liens vers les sources:\n",
    "* https://github.com/seth814/Audio-Classification#training\n",
    "* et https://github.com/Momotoculteur/Image-classification\n",
    "\n",
    "Ils ont été adaptés à notre corpus afin de pouvoir détecter une langue parmis: l'arabe, le français, le hindi et le russe.\n",
    "On retrouve les codes lancés et les résultats sous git https://github.com/ShNineb/CNN/tree/master ."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ce modèle bien que plus complexe que celui vu en cours (mais très bien expliqué dans les tutos) effectue un prétraitement des audios\n",
    "en considérant l'enveloppe et un seuil en-dessous duquel il ne prend plus en considération les zones dites \"mortes\".\n",
    "Il nous a semblé intéressant de tester (même si le test a surtout été la prise en main ici de ces codes) ce type de prétraitement. \n",
    "\n",
    "<i>Ceci reste à cette échelle un travail exploratoire. </i>\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Le premier modèle donne les résultats suivants sur notre corpus:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* L'optimiseur utilisé dans les trois cas est <b><i>Adam</i></b> , sur des couches de types: <b><i>lstm, conv1D et conv2D</i></b>.\n",
    "\n",
    "\n",
    "<i>lstm</i> semble légèrement meilleur que <i>conv2D</i> mais il est trop tôt pour tirer une quelconque conclusion: celà fera partie des perspectives à savoir multiplier les simulations numériques modifiant le prétraitement, l'architecture et les hypermaramètres. \n",
    "\n",
    "<i>N.B.: Seuls les graphes ont ici un intérêt, le code de cette page permet de recharger les résultats et de les afficher (il a été fourni et peut être lancé sur une machine sans passer par <i>Colab</i>). </i>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import pandas as pd"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 186,
   "metadata": {},
   "outputs": [],
   "source": [
    "path_mod_1 = \"C:/Nineb/M2-TAL/CNN/S2_reconnaissanceVoix/audio_classifier_langues\"\n",
    "log_csvs = sorted(os.listdir(path_mod_1 + '/logs'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 187,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = ['Conv 1D', 'Conv 2D', 'LSTM']\n",
    "colors = ['r', 'm', 'c']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 188,
   "metadata": {
    "scrolled": true
   },
   "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"
    }
   ],
   "source": [
    "for i, (fn, label, c) in enumerate(zip(log_csvs, labels, colors)):\n",
    "    csv_hist.append(csv_path)\n",
    "\n",
    "df0 = pd.read_csv(csv_hist[0])\n",
    "df1 = pd.read_csv(csv_hist[1])\n",
    "df2 = pd.read_csv(csv_hist[2])\n",
    "plt.plot(df0.accuracy, label='train')\n",
    "plt.plot(df0.val_accuracy, ls='--',label='test')\n",
    "plt.plot(df1.accuracy, label='train')\n",
    "plt.plot(df1.val_accuracy, ls='--', label='test')\n",
    "plt.plot(df2.accuracy, label='train')\n",
    "plt.plot(df2.val_accuracy, ls='--', label='test')\n",
    "plt.legend([\"train conV1D\",\"test conv1D\",\"train conV2D\",\"test conv2D\",\"train lstm\",\"test lstm\"])\n",
    "\n",
    "plt.title(\"Comparison of model's accuracy\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.xlabel(\"Epoch\")\n",
    "plt.tick_params(axis='both', which='major', labelsize=12)\n",
    "plt.show()\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 189,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.metrics import roc_auc_score, roc_curve, auc, confusion_matrix\n",
    "from glob import glob\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from tensorflow.keras.utils import to_categorical\n",
    "import os\n",
    "import matplotlib.pyplot as plt\n",
    "import pylab\n",
    "import seaborn as sns"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = np.load(path_mod_1 + '/logs/y_pred.npy')\n",
    "y_true = np.load(path_mod_1 + '/logs/y_true.npy')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = sorted(os.listdir(path_mod_1 + '/wavfiles'))\n",
    "n_classes = len(set(y_true))\n",
    "y_pred_le = np.argmax(y_pred, axis=1)\n",
    "y_pred_labels = [classes[x] for x in y_pred_le]\n",
    "y_true_labels = [classes[x] for x in y_true]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 192,
   "metadata": {},
   "outputs": [],
   "source": [
    "wav_paths = glob('{}/**'.format(path_mod_1 + '/wavfiles'), recursive=True)\n",
    "wav_paths = sorted([x.replace(os.sep, '/') for x in wav_paths if '.wav' in x])\n",
    "\n",
    "classes = sorted(os.listdir(path_mod_1 + '/wavfiles'))\n",
    "labels = [os.path.split(x)[0].split('/')[-1] for x in wav_paths]\n",
    "\n",
    "unique_labels = np.unique(labels)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* ####  La matrice de confusion: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x576 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 432x288 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "conf_mat = confusion_matrix(y_true_labels, y_pred_labels, labels=unique_labels)\n",
    "df_cm = pd.DataFrame(conf_mat, index = unique_labels,\n",
    "                     columns = unique_labels)\n",
    "plt.figure(figsize = (10,8))\n",
    "plt.title('Confusion Matrix')\n",
    "sns.heatmap(df_cm, annot=True, cmap='viridis')\n",
    "plt.show()\n",
    "plt.savefig('matrixConfusion')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Le second modèle donne les résultats suivants sur notre corpus"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ce modèle est plus classsique dans son prétraitement et utilise comme optimizer <b><i>Adamax</i></b>. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 49, 49, 32)        544       \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 24, 24, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 24, 24, 32)        0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 18432)             0         \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 128)               2359424   \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_10 (Dense)             (None, 4)                 516       \n",
      "=================================================================\n",
      "Total params: 2,360,484\n",
      "Trainable params: 2,360,484\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "\n",
    "def plotArchi(modelPath):\n",
    "    \"\"\"\n",
    "    Affichage de l'architecture\n",
    "    \"\"\"\n",
    "    # Chargement du modele\n",
    "    #print(\"Chargement du modèle :\\n\")\n",
    "    model = load_model(modelPath)\n",
    "    #print(\"\\nModel chargé.\")\n",
    "    print(model.summary())\n",
    "    \n",
    "    \n",
    "Path = 'C:\\\\Nineb\\\\M2-TAL\\\\CNN\\\\S2_reconnaissanceVoix\\\\Image-classification-master\\\\audio_classifier'    \n",
    "modelPath = Path +'\\\\trainedModel\\\\moModel.hdf5'\n",
    "\n",
    "plotArchi(modelPath)\n",
    "    \n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "path_mod_2 = 'C:\\\\Nineb\\\\M2-TAL\\\\CNN\\\\S2_reconnaissanceVoix\\\\Image-classification-master\\\\audio_classifier'\n",
    "\n",
    "pathLogs = path_mod_2 + '\\\\logs\\\\log_moModel.csv'\n",
    "pathSaveGraph =  path_mod_2 + '\\\\graph'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [],
   "source": [
    "def plot(X, Y, Y2, title, xLabel, yLabel, legendLoc, pathSaveGraph):\n",
    "    \"\"\"\n",
    "    # Fonction d'affichage de graph\n",
    "    :param X: correspond au nombre d'époch\n",
    "    :param Y: correspond a la courbe accuracy\n",
    "    :param Y2: correspond a la courbe loss\n",
    "    :param title: titre du graphique\n",
    "    :param xLabel: label des abcisses\n",
    "    :param yLabel: label des ordonnees\n",
    "    :param legendLoc: legende\n",
    "    :param pathSaveGraph: chemin de sauvegarde pour les graphiques\n",
    "    \"\"\"\n",
    "\n",
    "   #On trace nos differentes courbes\n",
    "    plt.plot(Y)\n",
    "    plt.plot(Y2)\n",
    "   #titre du graph, legende...\n",
    "    plt.title(title)\n",
    "    plt.xlabel(xLabel)\n",
    "    plt.ylabel(yLabel)\n",
    "    plt.legend(['train', 'val'], loc=legendLoc)\n",
    "   #Pour avoir un courbe propre qui demarre à 0\n",
    "    plt.xlim(xmin=0.0, xmax=max(X))\n",
    "    plt.savefig(pathSaveGraph +'\\\\' + title)\n",
    "    plt.figure()\n",
    "    #plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [],
   "source": [
    "def displayGraph(pathLog,pathSaveGraph):\n",
    "    \"\"\"\n",
    "    # Fonction permettant de creer nos graph de suivi de metriques\n",
    "    :param pathLog: chemin du CSV contenant nos metrics\n",
    "    :param pathSaveGraph: chemin de destination pour sauvegarder nos 4 graphiques en jpg\n",
    "    \"\"\"\n",
    "\n",
    "    data = pd.read_csv(pathLog)\n",
    "    # split into input (X) and output (Y) variables\n",
    "    #plot(data['epoch'], data['accuracy'], data['val_accuracy'], 'TRAIN_VAL_Accuracy', 'Epoch', 'Accuracy', 'upper left',pathSaveGraph)\n",
    "    #plot(data['epoch'], data['loss'], data['val_loss'], 'TRAIN_VAL_Loss', 'Epoch', 'Loss', 'upper left',pathSaveGraph)\n",
    "    \n",
    "    plt.plot(data['accuracy'], label='train_acc  ')\n",
    "    plt.plot(data['val_accuracy'], ls='--',label='val_acc')\n",
    "    plt.plot(data['loss'], label='train_loss')\n",
    "    plt.plot(data['val_loss'], ls='--', label='val_loss')\n",
    "\n",
    "    plt.legend([\"train_acc\",\"val_acc\",\"train_loss\",\"val_loss\"])\n",
    "\n",
    "    plt.title(\"TRAIN_VAL Accuracy/Loss\")\n",
    "    plt.ylabel(\"Accuracy\")\n",
    "    plt.xlabel(\"Epoch\")\n",
    "    plt.tick_params(axis='both', which='major', labelsize=12)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {
    "scrolled": false
   },
   "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"
    }
   ],
   "source": [
    "displayGraph(pathLogs,pathSaveGraph)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generateMatrix(model, datasetTestPath, imageSize, destinationMatrix):\n",
    "\t\"\"\"\n",
    "\t# Fonction qui va construire notre matrice de confusion\n",
    "\t:param model: chemin du modèle à charger pour realiser la prediction\n",
    "\t:param datasetTestPath: chemin du dataset contenant nos images de test\n",
    "\t:param imageSize: definit la taille de l'ensemble de nos images\n",
    "\t:param destinationMatrix: définit le chemin ou va être sauvegardé notre matrice sous format d'image\n",
    "\t:return:\n",
    "\t\"\"\"\n",
    "\n",
    "\t#Les tableaux contenanrt les predictions\n",
    "\ty_true = []\n",
    "\ty_pred = []\n",
    "\n",
    "\ttotal = 0\n",
    "\tsuccess = 0\n",
    "\tindex = 0\n",
    "\n",
    "\tprint('\\nEvaluation :')\n",
    "\t#On parcours notre dataset de test\n",
    "\tfor root, dirs, files in os.walk(datasetTestPath):\n",
    "\t\tfor mydir in dirs:\n",
    "\t\t\tfor sample in tqdm(os.listdir(root + '\\\\' + mydir), \"Prediction de la classe '{}'\".format(mydir)):\n",
    "\n",
    "\t\t\t\tsample_path = root + '\\\\' + mydir + '\\\\' + sample\n",
    "\t\t\t\t#Chargement et traitement de l'image\n",
    "\t\t\t\timg = Image.open(sample_path)\n",
    "\t\t\t\timg.load()\n",
    "\t\t\t\timg = img.resize(size=imageSize)\n",
    "\t\t\t\timg = np.asarray(img) / 255.\n",
    "\t\t\t\t#On reshape pour etre de la forme (nbImage,hauteurImage,largeurImage,nbCanaux)\n",
    "\t\t\t\timg = img.reshape(1, img.shape[0], img.shape[1], img.shape[2])\n",
    "\t\t\t\t#Prediction de notre modele\n",
    "\t\t\t\tpred = np.argmax(model.predict(img))\n",
    "\n",
    "\t\t\t\ttotal += 1\n",
    "\t\t\t\tif pred == index:\n",
    "\t\t\t\t\tsuccess += 1\n",
    "\n",
    "\t\t\t\ty_true.append(index)\n",
    "\t\t\t\ty_pred.append(pred)\n",
    "\n",
    "\t\t\tindex += 1\n",
    "\n",
    "\t#Precision de notre modele sur notre jeu de test en entier\n",
    "\taccuracy = (success / total) * 100.\n",
    "\tprint('\\nPrecision : {0:.3f}%'.format(accuracy))\n",
    "\n",
    "\n",
    "\tcnf_matrix = confusion_matrix(y_true, y_pred)\n",
    "\tnp.set_printoptions(precision=2)\n",
    "\n",
    "\t# Plot normalized confusion matrix\n",
    "\tplt.figure()\n",
    "\n",
    "\tcmap = plt.cm.Blues\n",
    "#\tclasses = ['marguerite', 'pissenlit', 'rose', 'tournesol', 'tulipe']\n",
    "\tclasses = ['ar', 'fr', 'hi', 'ru']\n",
    "    # classes = ['ar', 'fr', 'rose', 'hi', 'ru']\n",
    "\ttitle = 'Confusion matrix'\n",
    "\n",
    "\tcnf_matrix = cnf_matrix.astype('float') / cnf_matrix.sum(axis=1)[:, np.newaxis]\n",
    "\n",
    "\t#Legende de notre matrice\n",
    "\tplt.imshow(cnf_matrix, interpolation='nearest', cmap=cmap)\n",
    "\tplt.title(title)\n",
    "\tplt.colorbar()\n",
    "\ttick_marks = np.arange(len(classes))\n",
    "\tplt.xticks(tick_marks, classes, rotation=45)\n",
    "\tplt.yticks(tick_marks, classes)\n",
    "\n",
    "\tfmt = '.2f'\n",
    "\tthresh = cnf_matrix.max() / 2.\n",
    "\tfor i, j in itertools.product(range(cnf_matrix.shape[0]), range(cnf_matrix.shape[1])):\n",
    "\t\tplt.text(j, i, format(cnf_matrix[i, j], fmt), horizontalalignment=\"center\", color=\"white\" if cnf_matrix[i, j] > thresh else \"black\")\n",
    "\n",
    "\tplt.ylabel('True label')\n",
    "\tplt.xlabel('Predicted label')\n",
    "\tplt.tight_layout()\n",
    "\n",
    "\t#On sauvegarde notre matrice en image\n",
    "\tplt.savefig(destinationMatrix + '\\\\' + 'confusionMatrix')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import load_model\n",
    "from tqdm import tqdm\n",
    "from PIL import Image\n",
    "import itertools"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* ####  La matrice de confusion: "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\r",
      "Prediction de la classe 'ar':   0%|          | 0/5 [00:00<?, ?it/s]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Evaluation :\n",
      "WARNING:tensorflow:5 out of the last 24 calls to <function Model.make_predict_function.<locals>.predict_function at 0x000002AF526C41F8> triggered tf.function retracing. Tracing is expensive and the excessive number of tracings could be due to (1) creating @tf.function repeatedly in a loop, (2) passing tensors with different shapes, (3) passing Python objects instead of tensors. For (1), please define your @tf.function outside of the loop. For (2), @tf.function has experimental_relax_shapes=True option that relaxes argument shapes that can avoid unnecessary retracing. For (3), please refer to https://www.tensorflow.org/guide/function#controlling_retracing and https://www.tensorflow.org/api_docs/python/tf/function for  more details.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Prediction de la classe 'ar': 100%|██████████| 5/5 [00:00<00:00, 27.10it/s]\n",
      "Prediction de la classe 'fr': 100%|██████████| 5/5 [00:00<00:00, 34.34it/s]\n",
      "Prediction de la classe 'hi': 100%|██████████| 5/5 [00:00<00:00, 32.76it/s]\n",
      "Prediction de la classe 'ru': 100%|██████████| 5/5 [00:00<00:00, 33.42it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Precision : 70.000%\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAToAAAEYCAYAAADMJjphAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMywgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/Il7ecAAAACXBIWXMAAAsTAAALEwEAmpwYAAAv20lEQVR4nO3deXwV5dXA8d9JQiIKJOyQBMpuSDSygyibQFFWN0BAK9UWWyti9a22WhGxbkXr8rq0KFaLaBDwlYDKoogUK7KoKATUVKIkYZdFFhMI5/1jJuEmIckNuTd3yfnymQ93Zp555gw3nDwzz8wzoqoYY0w4iwh0AMYY42+W6IwxYc8SnTEm7FmiM8aEPUt0xpiwZ4nOGBP2LNGFMRGpLSKLROSgiMyrQj0TRGSZL2MLFBHpIyJfBToOU73E7qMLPBEZD9wOJAE/Ap8DD6rq6irWex0wGeitqieqGmewExEF2qtqZqBjMcHFWnQBJiK3A08CDwFNgZbAc8AoH1T/M+DrmpDkvCEiUYGOwQSIqtoUoAmIBQ4Do8spE4OTCHPd6Ukgxl3XH8gG7gB2AzuAX7rr7gfygePuPm4EpgGvetTdClAgyp2fCHyL06rcBkzwWL7aY7vewDrgoPt3b491K4EHgI/cepYBjco4tsL47/SI/3JgKPA18ANwt0f5HsDHwAG37DNAtLtulXssR9zjHetR/13ATmB24TJ3m7buPrq48/HAHqB/oH82bPLx/7VAB1CTJ+BS4ERhoimjzHRgDdAEaAz8B3jAXdff3X46UMtNEEeB+u76komtzEQHnAMcAs511zUHUtzPRYkOaADsB65ztxvnzjd0168E/gt0AGq784+UcWyF8U914/+1m2heA+oCKcAxoLVbvivQy91vK2ALcJtHfQq0O039j+L8wqjtmejcMr8GMoCzgaXAY4H+ubDJ95OdugZWQ2Cvln9qOQGYrqq7VXUPTkvtOo/1x931x1X1HZzWzLlnGM9J4DwRqa2qO1R182nKDAO+UdXZqnpCVV8HtgIjPMr8U1W/VtVjwBtAp3L2eRzneuRxIA1oBDylqj+6+88ALgBQ1Q2qusbdbxbwD6CfF8d0n6rmufEUo6ovAJnAJzjJ/Z4K6jMhyBJdYO0DGlVw7Sge+M5j/jt3WVEdJRLlUaBOZQNR1SM4p3u/AXaIyNsikuRFPIUxJXjM76xEPPtUtcD9XJiIdnmsP1a4vYh0EJHFIrJTRA7hXNdsVE7dAHtU9acKyrwAnAf8r6rmVVDWhCBLdIH1MZCHc12qLLk4nQqFWrrLzsQRnFO0Qs08V6rqUlUdjNOy2YqTACqKpzCmnDOMqTKex4mrvarWA+4GpIJtyr2tQETq4Fz3nAVME5EGPojTBBlLdAGkqgdxrk89KyKXi8jZIlJLRC4Tkb+6xV4H/iwijUWkkVv+1TPc5edAXxFpKSKxwJ8KV4hIUxEZJSLn4CTfwzinfSW9A3QQkfEiEiUiY4FkYPEZxlQZdXGuIx52W5u/LbF+F9CmknU+BaxX1V8BbwN/r3KUJuhYogswVX0c5x66P+NciN8O3AK85Rb5C7Ae+AL4EvjUXXYm+1oOzHXr2kDx5BThxpGL0xPZj9KJBFXdBwzH6endh9NjOlxV955JTJX0P8B4nN7cF3COxdM04BUROSAiYyqqTERG4XQIFR7n7UAXEZngs4hNULAbho0xYc9adMaYsGeJzhgTVETkJRHZLSKbylgvIvK0iGSKyBci0qWiOi3RGWOCzcs4107LchnQ3p0m4fTGl8sSnTEmqKjqKpwOsbKMAv6ljjVAnIg0L6/OoH7IuU5cA23QLKHigiGm8TkxgQ7B544dP92dKKHvcP7xQIfgF9u/2rRXVRv7qr7Iej9TPVHqwZPT0mN7NgOeN3HPVNWZldhdAs7dCYWy3WU7ytogqBNdg2YJ3DkrPdBh+NyverYOdAg+tzX3x0CH4Bert1fHXTPVb/LFbUo+3VIleuIYMedWeEcPAD99/uxPqtrNl/uvSFAnOmNMqBCQarsSlgO08JhPpIInc+wanTGm6gSIiPRuqrp04Bdu72sv4KCqlnnaCtaiM8b4ilT02LG31cjrOMNpNRKRbOA+nGG8UNW/4zyGOBRn1JmjwC8rqtMSnTHGB3x36qqq4ypYr8DvKlOnJTpjjG/4qEXnD5bojDFVJ1RnZ0SlWaIzxviA+KqjwS8s0RljfMNOXY0x4a1a76OrNEt0xpiqE6xFZ4ypAaxFZ4wJbwKR1hlhjAlndnuJMaZGsGt0xpjwZr2uxpiawFp0xpiwZy06Y0xYk+B+BCx4U7CPZKz5kAfGDeT+sQNYNrvslwV9vvJdJl/chu+3flG0bNns57h/7AAeGDeQLZ+sqo5wvbZs6RJSU84lJakdM/76SKn1eXl5XDt+LClJ7ejTuyffZWUVrZvx6MOkJLUjNeVcli9bWo1RV+yjlcu5fEAXRva9gJee+1up9bNfeIYrB3ZnzJALuWncCHKzvy9alz5/DiP7dWJkv06kz59TnWGXK1x/BksR8W4KgLBOdCcLCpj3t/v47WP/5J5Xl7LhvUXs2PZNqXI/HT3Mynkv0yq5U9GyHdu+YcN7i7l79hJ++/jLvPH4VE4WFFRj9GUrKCjgtlt/x8JF7/LZFxnMS3udLRkZxcq8/NIs6sfVZ/PWTCZP+T333H0XAFsyMpg3N41PN24mffESpky+mYIgOq5H7r2DZ15ZwIL31rEkfT7//XprsTJJKanMWfwhbyz9mIFDR/HUw1MBOHjgB2Y++SizF67g1fQPmPnkoxw6uD8Qh1FMuP4MluZ2RngzBUBYJ7rvtmykUeLPaJTQkqha0XQdNJwvVy8vVe7tF/7GoAk3ERV96u1cX65eTtdBw6kVHUOj+BY0SvwZ323ZWJ3hl2nd2rW0bduO1m3aEB0dzeix17B40cJiZRYvWsiE664H4MqrrmblivdRVRYvWsjosdcQExNDq9atadu2HevWrg3EYZSy6fP1tGjVhsSWrakVHc2QEVexcvnbxcp0792X2rXPBiC1c3d27XBeFfCfD9+nV58BxMY1oF5sfXr1GcBHK9+r9mMoKVx/Bk/LWnSBcWDPTuo3OfW6x7jGzTmwZ1exMtu/2sT+3Ts4r/clJbbdRf0m8R7bNuPAnp3+DdhLubk5JCaeejdIQkIiOTk5pcu0cMpERUVRLzaWffv2kZNTetvc3HLfK1Jtdu/cQdPmiUXzTZvHs2dnbpnl35r7Ly7qPxiAPTt30LT5qVdjNmkWz56d5b5GoFqE689gKYU3DAdpi65Gd0acPHmSN//3Qa69Z0agQzGV9PabaWR8+Rkvzn030KFUSfj8DAZ3Z0RAE52IRKqq3y46xDVuxv7dp36rH9izg7jGTYvm844eZse2r3l6sjNE/aEf9vCPuyZx06MziWvclP27cz223Ulc42b+CrVS4uMTyM4+9f7enJxsEhISSpfZvp3ExEROnDjBoYMHadiwIQkJpbeNjw+Ol4Q3adacXTuyi+Z37cilcbP4UuXWrP6AWc88xotvvEt0jHOq17hZczasWV1UZvfOXLr2utj/QVcgXH8GTyuIby/xa2Qi8paIbBCRzSIyyV12WEQeF5GNwIX+3H/LpFT2bM9ib+52ThzPZ8N7izn/okFF62vXqccjb2/g/vn/5v75/6ZVcmduenQmLZNSOf+iQWx4bzHH8/PYm7udPduz+FnHC/wZrte6de9OZuY3ZG3bRn5+PvPmpjFs+MhiZYYNH8mc2a8A8OaC+fQbcAkiwrDhI5k3N428vDyytm0jM/MbuvfoEYjDKCXlgq58v+1bcr7P4nh+PksXLaD/4KHFymzdtJEH/zSFJ2al0aDRqRfN9+43kI9XreDQwf0cOrifj1etoHe/gdV9CKWE68/gaQXxNTp/t+huUNUfRKQ2sE5EFgDnAJ+o6h2n28BNiJMA6jct/du8MiKjohh9+zSeu/169ORJeg0bTfM2HXj7xSdomXQ+5188qMxtm7fpQJdLhvHQtUOIiIxk9O33ExEkozNERUXxxFPPMGLYEAoKCrh+4g0kp6QwfdpUunTtxvARI5l4w43cMPE6UpLaUb9+A2bPSQMgOSWFq0aPoXNqMlFRUTz59LNEBtFx3TV9Bjf/4gpOFhQwasx1tO3Qkece/wvJqV3oP3goTzx0L0ePHuHOm52OlmbxiTw1ay6xcQ349a13cu2I/gBMmnIXsXENAng0jnD9GSxFgvsRMHHeHOanykWmAVe4s62AIcBqIMabU9aWSefrnbPS/RZfoPyqZ+tAh+BzW3N/DHQIfrF6+95Ah+AXky9us0FVu/mqvoj6rTRmwL1elf3p/37l0317w28tOhHpDwwCLlTVoyKyEjgL+Mmf1+WMMdVPgIiI4G3R+fPUNRbY7ya5JKCXH/dljAkkcacg5c9EtwT4jYhsAb4C1vhxX8aYgBKkJo5eoqp5wGWnWVXHX/s0xgROjUx0xpiaxRKdMSa8CUiEJTpjTBiTmnqNzhhTs1iiM8aEPUt0xpiwZ4nOGBPegvyG4eB9ZsMYEzIEISIiwqupwrpELhWRr0QkU0T+eJr1LUXkAxH5TES+EJGhp6vHkyU6Y4xPiIhXUwV1RALP4jxskAyME5HkEsX+DLyhqp2Ba4DnKorNEp0xxjfEy6l8PYBMVf1WVfOBNGBUiTIK1HM/xwJlj7fvsmt0xpiqk0p1RjQSkfUe8zNVdab7OQHY7rEuG+hZYvtpwDIRmYwzvmXZg/q5LNEZY3yiEolubxXHoxsHvKyqj4vIhcBsETlPVU+WtYElOmNMlRV2RvhADtDCYz7RXebpRuBSAFX9WETOAhoBu8uq1K7RGWN8wzfX6NYB7UWktYhE43Q2lBxm/HtgIICIdMQZ0HdPeZVai84YU3WVu0ZXJlU9ISK3AEuBSOAlVd0sItOB9aqaDtwBvCAiv8fpmJioFbwTwhKdMcYnfPVkhKq+A7xTYtlUj88ZwEWVqdMSnTHGJ+wRMGNM2LPx6IwxYc2bpx4CyRKdMcYnLNEZY8KeJbozVCe6Fhe3aBToMHzuqhfXBjoEn3twaMdAh+AXv+rZOtAh+MVkf1QavHkuuBOdMSZECL56MsIvLNEZY6pMgCA+c7VEZ4zxBet1NcbUAEGc5yzRGWN8w1p0xpjwJtaiM8aEOQEiI4M301miM8b4hJ26GmPCm526GmPCnXMfXfBmOkt0xhgfsPvojDE1QISNR2eMCWt2jc4YE+7sGp0xpkYI4jxnic4Y4xvWojPGhDexzghjTJiz8eiMMTVAcN9HF7xjH/vIRyuXc/mALozsewEvPfe3Uutnv/AMVw7szpghF3LTuBHkZn9ftC59/hxG9uvEyH6dSJ8/pzrDrlDXFrH845rzeWFcKqM7NT9tmYvbNuD5Mefz3Jjz+MPAtkXLB3ZoxMxxqcwcl8rADsH1To5w/L6WLV1Casq5pCS1Y8ZfHym1Pi8vj2vHjyUlqR19evfku6ysonUzHn2YlKR2pKacy/JlS6sx6soT8W4KhLBu0RUUFPDIvXfw/JyFNG2WwISR/ek3aChtOyQVlUlKSWXO4g+pXfts3pj9Ik89PJVHn32Zgwd+YOaTjzJn8UpEhPHD+tF/8FDqxdYP4BE5IgR+e/HP+PPir9h7JJ8nrkxhzXf72b7/p6Iy8bExjOncnD+8lcHh/AJiz3K+6joxkYzvFs+UBZtB4amrU/gkaz+H8wsCdThFwvH7Kigo4LZbf8fb7y4nITGRi3t1Z/jwkXRMTi4q8/JLs6gfV5/NWzN5Y24a99x9F6++NpctGRnMm5vGpxs3syM3l6GXDuLLjK+JjIwM4BGVzVp0AbLp8/W0aNWGxJatqRUdzZARV7Fy+dvFynTv3Zfatc8GILVzd3btyAHgPx++T68+A4iNa0C92Pr06jOAj1a+V+3HcDodmtQh91AeO3/M48RJZdV/99GrVfH/0EM6NmHxpt1FCezgTycApyX4WfYhDucVcDi/gM+yD9G1ZWy1H8PphOP3tW7tWtq2bUfrNm2Ijo5m9NhrWLxoYbEyixctZMJ11wNw5VVXs3LF+6gqixctZPTYa4iJiaFV69a0bduOdWuD8w1y4nZGeDMFQlgnut07d9C0eWLRfNPm8ezZmVtm+bfm/ouL+g8GYM/OHTRtnlC0rkmzePbs3OG/YCuh4Tm12Hs4r2h+7+F8Gp4TXaxMQuxZJMSdxYzLO/L4Fcl0bRHrbhvNnsP55W4bKOH4feXm5pCY2KJoPiEhkZycnNJlWjhloqKiqBcby759+8jJKb1tbm7xbYOJiHg1BUK1JDoRuVVEtohI8Fw4KeHtN9PI+PIzrr9pSqBD8YnICCE+NoY/pm/lr+9lMrlfK86JDs5TnjMRbt9XOAjma3TV1aK7GRisqhMKF4iI368PNmnWnF07sovmd+3IpXGz+FLl1qz+gFnPPMaTL84lOiYGgMbNmhedFgHs3plL42anv+hf3fYdOU6jOjFF843qRLPvSH6xMnsP5/NJ1gEKTiq7fswn58BPxMeexb4j+TSuE13utoESjt9XfHwC2dnbi+ZzcrJJSEgoXWa7U+bEiRMcOniQhg0bkpBQetv4+OLbBpMa3aITkb8DbYB3ReSgiMwWkY+A2f7ed8oFXfl+27fkfJ/F8fx8li5aQP/BQ4uV2bppIw/+aQpPzEqjQaPGRct79xvIx6tWcOjgfg4d3M/Hq1bQu99Af4fsla93HyYhNoamdaOJihD6tm3IJ1kHipVZk7Wf8+PrAVDvrCgS4s5i56E8Nmw/SOfEWOpER1InOpLOibFs2H4wAEdRWjh+X926dycz8xuytm0jPz+feXPTGDZ8ZLEyw4aPZM7sVwB4c8F8+g24BBFh2PCRzJubRl5eHlnbtpGZ+Q3de/QIxGFUzMvWXNj2uqrqb0TkUmAAcAswArhYVY+drryITAImATRPaHG6Il6LiorirukzuPkXV3CyoIBRY66jbYeOPPf4X0hO7UL/wUN54qF7OXr0CHfe7FwMbhafyFOz5hIb14Bf33on147oD8CkKXcRG9egSvH4ykmF51d/xwPDkogQWP7VHr7ff4xruyXwzZ4jfPLdgaKE9vyY8zmpyksfb+fHPKdDIm1DDk9clQLA6xtyOJwX+B5XCM/vKyoqiieeeoYRw4ZQUFDA9RNvIDklhenTptKlazeGjxjJxBtu5IaJ15GS1I769Rswe04aAMkpKVw1egydU5OJioriyaefDd4eVwLX0eANUVX/70QkC+iGk+hUVe/3Zrvk1C762uIP/RlaQNzzzpZAh+BzDw7tGOgQ/CIpvm6gQ/CL2rVkg6p281V99Vp21J53/tOrsu9NvtCn+/ZGIHpdjwRgn8YYP/PVqauIXCoiX4lIpoj8sYwyY0QkQ0Q2i8hrFdUZ1jcMG2Oqh5PEqn7qKiKRwLPAYCAbWCci6aqa4VGmPfAn4CJV3S8iTSqqN6zvozPGVJ8I8W6qQA8gU1W/VdV8IA0YVaLMr4FnVXU/gKrurqjSMlt0IvK/QJkX8FT11gpDPlW2lftxmrfbGGNCSyVadI1EZL3H/ExVnel+TgC2e6zLBnqW2L6Du7+PgEhgmqouKW+H5Z26ri9nnTHGFBEgwvtEt7eKnRFRQHugP5AIrBKR81X1QHkbnJaqvuI5LyJnq+rRKgRnjAljPrq7JAfwvK8s0V3mKRv4RFWPA9tE5GucxLeuzNgq2quIXCgiGcBWd/4CEXmuksEbY8KZl09FeHF6uw5oLyKtRSQauAZIL1HmLZzWHCLSCOdU9tvyKvWmM+JJYAiwD0BVNwJ9vdjOGFOD+OL2ElU9gXO/7VJgC/CGqm4WkekiUvhIyVJgn9sA+wD4g6ruK69er24vUdXtJTJxcNxKb4wJCpW8RlcuVX0HeKfEsqkenxW43Z284k2i2y4ivQEVkVrAFJxMa4wxRYL5ETBvTl1/A/wOp9s3F+jkzhtjDOD9aWvQPtSvqnuBCRWVM8bUbL46dfUHb3pd24jIIhHZIyK7RWShiLSpjuCMMaFDvJwCwZtT19eAN4DmQDwwD3jdn0EZY0JPqA+8ebaqzlbVE+70KnCWvwMzxoQOESEywrspEMp71rVw1MJ33aFS0nCefR1Lia5fY4wJ4kt05XZGbMBJbIXh3+SxTnGGSTHGGCC43+ta3rOuraszEGNM6HJuGA50FGXz6skIETkPSMbj2pyq/stfQRljQk9ItugKich9OA/QJuNcm7sMWA1YojPGAM71ucggTnTe9LpeDQwEdqrqL4ELgFi/RmWMCTkh/WQEcExVT4rICRGpB+ym+HhRxhgT2qeuwHoRiQNewOmJPQx87M+gjDGhJ4jznFfPut7sfvy7iCwB6qnqF/4NyxgTSgQJ6mddy7thuEt561T1U/+EFP4W/KpHoEPwuVa/nR/oEPwi6/mrAx1CaAjg9TdvlNeie7ycdQpc4uNYjDEhLJh7Xcu7YXhAdQZijAldQuh3RhhjTIVC/skIY4ypiCU6Y0xYc24GDt5M580IwyIi14rIVHe+pYiEX7ehMaZKIiO8mwLBm90+B1wIjHPnfwSe9VtExpiQU/i6Q2+mQPDm1LWnqnYRkc8AVHW/+wZtY4wpEqDGmle8SXTHRSQS5945RKQxcNKvURljQk4QX6LzKtE9Dfwf0EREHsQZzeTPfo3KGBNSJICnpd7w5lnXOSKyAWeoJgEuV9Utfo/MGBNSAtXR4A1vBt5sCRwFFnkuU9Xv/RmYMSZ0FHZGBCtvTl3f5tRLcs4CWgNfASl+jMsYE2KCOM95dep6vue8O6rJzWUUN8bURBJmT0ao6qci0tMfwRhjQpcQvJnOm2t0t3vMRgBdgFy/RWSMCTkCRIVyZwRQ1+PzCZxrdgv8E44xJlQF87Ou5SY690bhuqr6P9UUjzEmBAX7C6zLbGyKSJSqFgAXVWM8xphQ5OWrDr1p9InIpSLylYhkisgfyyl3lYioiHSrqM7yWnRrca7HfS4i6cA84EjhSlV9s+KQjTE1hS/uo3PPIp8FBgPZwDoRSVfVjBLl6gJTgE+8is2LMmcB+3DeETEcGOH+HRI+Wrmcywd0YWTfC3jpub+VWj/7hWe4cmB3xgy5kJvGjSA3+9R90Onz5zCyXydG9utE+vw51Rl2hZYtXUJqyrmkJLVjxl8fKbU+Ly+Pa8ePJSWpHX169+S7rKyidTMefZiUpHakppzL8mVLqzHqig1IacrqB4bw8YOXcsul55Zan9CgNgvu6Mvyewey4r5BDDyvWdG6yZedy8cPXsrqB4bQP6VpdYZdrnD9rjwJPhumqQeQqarfqmo+kAaMOk25B4BHgZ+8ia+83TZxe1w3AV+6f292/97kTeWBVlBQwCP33sEzryxgwXvrWJI+n/9+vbVYmaSUVOYs/pA3ln7MwKGjeOrhqQAcPPADM598lNkLV/Bq+gfMfPJRDh3cH4jDKKWgoIDbbv0dCxe9y2dfZDAv7XW2ZBT7hcfLL82iflx9Nm/NZPKU33PP3XcBsCUjg3lz0/h042bSFy9hyuSbKSgoCMRhlBIh8PD4zox/ajV9py7lih4t6NC8brEytw3rSPr6bAY/8D6/mfkJj0zoDECH5nW5vHsL+t23jPFP/ZtHxncOimtG4fpdlSZEeDkBjURkvcc0yaOiBGC7x3y2u+zUnpx7eVuo6tveRldeoosE6rhTXY/PhVPQ2/T5elq0akNiy9bUio5myIirWLm8+L9N9959qV37bABSO3dn144cAP7z4fv06jOA2LgG1IutT68+A/ho5XvVfgyns27tWtq2bUfrNm2Ijo5m9NhrWLxoYbEyixctZMJ11wNw5VVXs3LF+6gqixctZPTYa4iJiaFV69a0bduOdWvXBuIwSuncugHb9hzm+71HOF6gvLVuO0M6xRcrowp1a9cCnL93HnB+oQ/pFM9b67aTf+Ik3+89yrY9h+ncukG1H0NJ4fpdleS8HMfra3R7VbWbxzTT6/2IRAB/A+6oTHzlXaPboarTK1NZsNm9cwdNmycWzTdtHs+mz9aXWf6tuf/iov6DAdizcwdNm5/6RdKkWTx7du7wX7CVkJubQ2Jii6L5hIRE1q79pHSZFk6ZqKgo6sXGsm/fPnJycujZs1exbXNzc6on8Ao0j6tN7g/HiuZ37D9GlxLJ6rFFGcy9rQ83XNKWs6OjGPO3VUXbbvj2h2LbNo+rXT2BlyNcv6tSfPdkRA7QwmM+0V1WqC5wHrDSvZ2lGZAuIiNVtcz/3OW16Koctoi0EpFSp7kiMl1EBlW1fl96+800Mr78jOtvmhLoUEw5rujRgrn/yaLLne8w4enVPHNjj6B+xrIm8dEIw+uA9iLS2h3g9xogvXClqh5U1Uaq2kpVWwFrgHKTHJSf6AZ6dXRnQFWnqqrfzwObNGvOrh3ZRfO7duTSuFl8qXJrVn/ArGce48kX5xIdEwNA42bNi05jAXbvzKVxs+b+Dtkr8fEJZGefuoyRk5NNQkJC6TLbnTInTpzg0MGDNGzYkISE0tvGxxffNlB2HDhGfINTrbDm9Wuz48CxYmXGX9yK9PXOd7rh2x+IqRVBwzoxXm0bCOH6XZVUyVPXMqnqCeAWYCmwBXhDVTe7jaORZxpfmYlOVX8oa10lRYrICyKyWUSWiUhtEXlZRK72Uf1lSrmgK99v+5ac77M4np/P0kUL6D94aLEyWzdt5ME/TeGJWWk0aNS4aHnvfgP5eNUKDh3cz6GD+/l41Qp69/Nb7q+Ubt27k5n5DVnbtpGfn8+8uWkMG178Z2DY8JHMmf0KAG8umE+/AZcgIgwbPpJ5c9PIy8sja9s2MjO/oXuP4HjX0edZ+2nTpA4tG51NrUjh8u4tWLax+OWCnH3H6NOxCQDtm9UlplYke3/MY9nGHVzevQXRURG0bHQ2bZrU4bNtvvoRPnPh+l2dTmSEeDVVRFXfUdUOqtpWVR90l01V1fTTlO1fUWsOqud1h+2Bcar6axF5A7iqvMJuD8wkgOYJLcorWqGoqCjumj6Dm39xBScLChg15jradujIc4//heTULvQfPJQnHrqXo0ePcOfNzsXgZvGJPDVrLrFxDfj1rXdy7Yj+AEyachexcYG/uA3OcT3x1DOMGDaEgoICrp94A8kpKUyfNpUuXbsxfMRIJt5wIzdMvI6UpHbUr9+A2XPSAEhOSeGq0WPonJpMVFQUTz79LJGRkQE+IkfBSeXu1z7n9dv6ECnC6x9l8VXuIe4cmczn3+1n2cYdTJu3kcd+0ZVJg9qjwJR/Oj/jX+UeIn19Nqvu/zknTip/eu1zTmpgjwfC97sqSQjud0aIqv9+GkSkFbBcVdu783cBtYB2wGJVnV/e9smpXfS1xR/6Lb5ASYqvW3GhENPqt+V+lSEr63m/n3gERO1askFVK3yiwFutk1N12r+8u9tjYveWPt23N6qjRZfn8bkACHxXmDHG54K5T6g6Ep0xJsyFw1DqxhhToWB4EqUsfk10qpqFc3Nf4fxj/tyfMSZQJHTHozPGGG8Ee6+rJTpjjE9Yi84YE/aCN81ZojPG+IAIRFqLzhgT7uzU1RgT9oI3zVmiM8b4SBA36CzRGWOqzrm9JHgznSU6Y4wPeDWoZsBYojPG+EQQ5zlLdMaYqrNTV2NM+PNimPRAskRnjPEJS3TGmLAndupqjAlngj0CZoypAYI4z1miM8b4hp26GmPCmvPOiEBHUTZLdMYYHxBr0RljwpxYi+6M1a4VEZYvew5H4fqi5/rdbwl0CCHBXndojKkRgjfNWaIzxvhKEGc6S3TGGJ+wzghjTNizzghjTPizRGeMCWeCnboaY8JdkI9HFxHoAIwx4UG8nCqsR+RSEflKRDJF5I+nWX+7iGSIyBci8r6I/KyiOi3RGWN8QBDxbiq3FpFI4FngMiAZGCciySWKfQZ0U9VUYD7w14qis0RnjPEJEe+mCvQAMlX1W1XNB9KAUZ4FVPUDVT3qzq4BEiuq1BKdMabKvD1tdfNcIxFZ7zFN8qgqAdjuMZ/tLivLjcC7FcVnnRHGGN/wvjNir6p2q/LuRK4FugH9Kipric4Y4xM+ur0kB2jhMZ/oLiu+L5FBwD1AP1XNq6hSO3U1xviEj67RrQPai0hrEYkGrgHSi+9HOgP/AEaq6m5vYrMWnTGm6nx0H52qnhCRW4ClQCTwkqpuFpHpwHpVTQdmAHWAeW4v7veqOrK8ei3RGWN8wldPRqjqO8A7JZZN9fg8qLJ1WqIzxlSZENxPRliiM8b4RBDnOUt0xhgfCeJMF/a9rsuWLiE15VxSktox46+PlFqfl5fHtePHkpLUjj69e/JdVlbRuhmPPkxKUjtSU85l+bKl1Rh1xey4Que4/n7fBL57/2HWz7u7zDKP33k1mxbex9q5f6JT0qkb/SeM6MmXC6fy5cKpTBjRszrCPWMRIl5NAYktIHutJgUFBdx26+9YuOhdPvsig3lpr7MlI6NYmZdfmkX9uPps3prJ5Cm/55677wJgS0YG8+am8enGzaQvXsKUyTdTUFAQiMMoxY4rtI5r9qI1jPrds2WuH3JxMm1bNua8Ufdzy19e5+m7rwGgfr2zuWfSZfS97jH6XDuDeyZdRlzd2tUVdqX56qF+fwjrRLdu7Vratm1H6zZtiI6OZvTYa1i8aGGxMosXLWTCddcDcOVVV7NyxfuoKosXLWT02GuIiYmhVevWtG3bjnVr1wbiMEqx4wqt4/ro0//yw8GjZa4f3i+V1xY7sa79MovYurVp1qgeg3t35P01W9l/6CgHfjzG+2u28vOLSj7fHkSCONOFdaLLzc0hMfHUTdYJCYnk5OSULtPCKRMVFUW92Fj27dtHTk7pbXNzS92gHRB2XKF1XBWJbxJH9s79RfM5uw4Q3ySO+MZxZO/yWL77APGN4wIQYcUKB9705k8gWGeEMabqbOBNEEe1tx7j4xPIzj41EEJOTjYJCQmly2x3ypw4cYJDBw/SsGFDEhJKbxsfX94gCtXHjiu0jqsiubsPkNisftF8QtM4cncfIHfPARKbeixvEkfungMBiNA7PnoEzC/8lnxEpJU7Sui/gE1Agce6q0XkZX/tu1C37t3JzPyGrG3byM/PZ97cNIYNL/6kyLDhI5kz+xUA3lwwn34DLkFEGDZ8JPPmppGXl0fWtm1kZn5D9x49/B2yV+y4Quu4KvL2h18yfrgTa4/zW3Ho8DF27j3E8v9sYdCFScTVrU1c3doMujCJ5f/ZEuBoy+LtiWt4nrq2B65X1TUictibDdyxqSYBtGjZsko7j4qK4omnnmHEsCEUFBRw/cQbSE5JYfq0qXTp2o3hI0Yy8YYbuWHidaQktaN+/QbMnpMGQHJKCleNHkPn1GSioqJ48ulniYyMrFI8vmLHFVrH9crDE+nTtT2N4uqQueQBHvj7O9SKcmJ7cf5qlqzezJCLU9icfh9HfzrOTdNeBWD/oaM8/MISVr96JwAPzVzC/kNld2oEWjCfuoqq+qdikVbAB6ra2p0/rKp13M9XA8NVdWJ5dXTt2k0/+mS9X+Izxhv1u98S6BD84qfPn93gizHhCqV26qrp733kVdnWjWv7dN/e8HeL7ojHZ8+Mepaf92uMqW5B3KKrzg6CXSLS0e2UuKIa92uMqQbB/GREdd5e8kdgMbAHWI8znpQxJkwEcYPOf4lOVbOA8zzm5+O8mswYE26C/D46u2HYGOMjwZvpLNEZY6rMBt40xtQIEZbojDHhLlBPPXjDEp0xxjeCN89ZojPG+EYQ5zlLdMaYqgvkyCTesERnjPEJu0ZnjAl71qIzxoQ9S3TGmDAXuEE1vWGJzhhTZcH+ZERYvwXMGGPAWnTGGB8J1Fhz3rBEZ4ypOruPzhgT7gR7MsIYUxMEcaazRGeM8Ylgvr3Eel2NMT4RId5NFRGRS0XkKxHJFJE/nmZ9jIjMddd/4r5atfzYzuiIjDGmJPFyKq8KkUjgWeAyIBkYJyLJJYrdCOxX1XbAE8CjFYVmic4Y4xPi5Z8K9AAyVfVbVc0H0oBRJcqMAl5xP88HBoqU3+cb1NfoPv10w97ateS7atpdI2BvNe2ruoTjMYEdly/8zJeVffbphqVnR0sjL4ufJSLrPeZnqupM93MCsN1jXTbQs8T2RWVU9YSIHAQaUs6/XVAnOlVtXF37EpH1qtqtuvZXHcLxmMCOKxip6qWBjqE8dupqjAkmOUALj/lEd9lpy4hIFBAL7CuvUkt0xphgsg5oLyKtRSQauAZIL1EmHbje/Xw1sEJVtbxKg/rUtZrNrLhIyAnHYwI7rrDlXnO7BVgKRAIvqepmEZkOrFfVdGAWMFtEMoEfcJJhuaSCRGiMMSHPTl2NMWHPEp0xJuxZojPGhD1LdGFEHGH7nYpIkoh0DHQcJvSE7X+KyhKR3iJSP9BxVNE5qnoSQESuF5G+gQ7IV0SkFjAS2CUisYGOxx9EpGGgYwhXlugAEfk58DhwPNCxnCkRGQU86X4eBvwS2BrImHxFRCJV9TjwN6A58JCI9AhwWD4lIp2B50WkU6BjCUc1PtGJyLXADOBPqnrYvdM6pLgtgck4CWAc8EfgE1Xd7a4P3oHCvKCqBe7HRJz7pmKAn4tI98BF5XPHgY3ALSJyQaCDCTc1PtEBH+I8TjIaim5YDLV/l3zgBHAvMBXYApwrIhcDqKqGYrJzLydc436eDKwA/gB8A7QFLhWRkHw2tJDHd7QJeBP4CrhNRJICGliYCbX/0D4jIreIyN9xhnwZClwrIn8AUNWToZQYVPVHnCQwBvinqk4CNgHDRaS3WyYU7wyvDzwsItOAC4DBOImgFs7zji2Bq0WkS8AirAIRuRVYJSKvAqjqFpwnAhoAD1rHi+/UyEQnIjfjtOD+AkzHucg9Eec36TQIycQwFydp3yAiN+IMXvgTMF5ESg5zExJU9W1gEnCVM6v/BV4GtgF5wI9AHeD7QMV4pkTkCuA8IAnoICIvA6jqF8DXQBZwMFDxhZsa9wiYiNTDuah9L06yG4Yz8sExYAnwCNAdZwTTkPvHcVs3c3GS+Hs4Dz+/WHi9LhS5HS0vALeqapp7aWEi0BjnWcg9gYyvskQkAVgDrFbVcSJSB1gLfApsACYAl6tqdgDDDCs1LtGBM+Y8zm/SJ1V1gHuaegDnIv6r7qlgyHIvZq/A6aCY63ExP2S5PckPAw95JLs6qnoowKGdERG5EngeuF1V57g/kw/hDDb+stuyMz4Scj2MvqCqeSJyFIgSkfNxRltdArwT6kkOQFU3ikh/4Fg4JDlwTmNF5CQwU0ROqOp8ICSTHICqviki+Tg95ajqHOAOEYkovBfS+E6NbNFBUavuNmAQEA+MVtWMgAZlKiQig4H/quq3gY7FF0TkMpzhmW5T1QWBjidc1dhEB0V32zcDTqpqyVFMjakW4Za8g1GNTnTGmJqhRt5eYoypWSzRGWPCniU6Y0zYs0RnjAl7luiMMWHPEl2IEpECEflcRDaJyDwRObsKdb0sIle7n18UkeRyyvYvHCigkvvIEpFG3i4vUeZwJfc1TUT+p7IxmvBliS50HVPVTqp6Hs4wTb/xXHmm4+qp6q8quHG6P1DpRGdMIFmiCw//Btq5ra1/i0g6kCEikSIyQ0TWicgXInITFL1b4hkR+UpE3gOaFFYkIisLx3gTkUtF5FMR2Sgi74tIK5yE+nu3NdlHRBqLyAJ3H+tE5CJ324YiskxENovIizjPcJZLRN4SkQ3uNpNKrHvCXf6+iDR2l7UVkSXuNv+2MdxMWWrks67hxG25XYbzrC5AF+A8Vd3mJouDqtrdfeTtIxFZBnQGzgWSgaZABvBSiXob44wY0tetq4Gq/uCO4XdYVR9zy70GPKGqq0WkJc54ah2B+3BG55juPpB/oxeHc4O7j9rAOhFZoKr7gHNw3tL+exGZ6tZ9C86jU79R1W/coaieAy45g39GE+Ys0YWu2iLyufv538AsnFPKtaq6zV3+cyC18PobEAu0B/oCr7sP/OeKyIrT1N8LWFVYl6r+UEYcg4Bkj3FK67nDDvUFrnS3fVtE9ntxTLe647SBM+pze5whtE7iDD0F8CrwpruP3sA8j33HeLEPUwNZogtdx1S1k+cC9z/8Ec9FwGRVXVqi3FAfxhEB9FLVn04Ti9fc0VYGAReq6lERWQmcVUZxdfd7oOS/gTGnY9fowttS4Lfu4AWISAcROQdYBYx1r+E1BwacZts1QF8Rae1u28Bd/iNQ16PcMpxx73DLdXI/rgLGu8suwxkWvTyxOIOdHnWvtfXyWBcBFLZKx+OcEh8CtonIaHcfIvZSGVMGS3Th7UWc62+fisgm4B84rfj/w3nBTAbwL+Djkhu6o/ZOwjlN3MipU8dFwBWFnRHArUA3t7Mjg1O9v/fjJMrNOKewFQ13vgRnfMAtOKM8r/FYdwTo4R7DJTjD34MzEu+NbnybcYaSN6YUG73EGBP2rEVnjAl7luiMMWHPEp0xJuxZojPGhD1LdMaYsGeJzhgT9izRGWPC3v8DtkXclte7F68AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Path = 'C:\\\\Nineb\\\\M2-TAL\\\\CNN\\\\S2_reconnaissanceVoix\\\\Image-classification-master\\\\audio_classifier'\n",
    "datasetTestPath = Path + '\\\\datasetTest'\n",
    "destinationMatrix = Path + '\\\\graph'\n",
    "modelPath = Path + '\\\\trainedModel\\\\moModel.hdf5'\n",
    "\n",
    "imageSize = (50, 50)\n",
    "model = load_model(modelPath)\n",
    "\n",
    "generateMatrix(model, datasetTestPath, imageSize, destinationMatrix)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 203,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def predict(modelPath,imagePath, imageSize, label):\n",
    "    \"\"\"\n",
    "    # Fonction qui permet de convertir une image en array, de charger le modele et de lui injecter notre image pour une prediction\n",
    "    :param modelPath: chemin du modèle au format hdf5\n",
    "    :param imagePath: chemin de l'image pour realiser une prediction\n",
    "    :param imageSize: défini la taille de l'image. IMPORTANT : doit être de la même taille que celle des images\n",
    "    du dataset d'entrainements\n",
    "    :param label: nom de nos 5 classes de sortie\n",
    "    \"\"\"\n",
    "\n",
    "    start = time.time()\n",
    "\n",
    "    # Chargement du modele\n",
    "    print(\"Chargement du modèle :\\n\")\n",
    "    model = load_model(modelPath)\n",
    "    print(\"\\nModel chargé.\")\n",
    "    \n",
    "  \n",
    "    #Chargement de notre image et traitement\n",
    "    data = []\n",
    "    img = Image.open(imagePath)\n",
    "    img.load()\n",
    "    img = img.resize(size=imageSize)\n",
    "    img = np.asarray(img) / 255.\n",
    "    data.append(img)\n",
    "    data = np.asarray(data)\n",
    "\n",
    "    #On reshape pour correspondre aux dimensions de notre modele\n",
    "    # Arg1 : correspond au nombre d'image que on injecte\n",
    "    # Arg2 : correspond a la largeur de l'image\n",
    "    # Arg3 : correspond a la hauteur de l'image\n",
    "    # Arg4 : correspond au nombre de canaux de l'image (1 grayscale, 3 couleurs)\n",
    "    dimension = data[0].shape\n",
    "\n",
    "    #Reshape pour passer de 3 à 4 dimension pour notre réseau\n",
    "    data = data.astype(np.float32).reshape(data.shape[0], dimension[0], dimension[1], dimension[2])\n",
    "\n",
    "    #On realise une prediction\n",
    "    prediction = model.predict(data)\n",
    "\n",
    "\n",
    "    #On recupere le numero de label qui a la plus haut prediction\n",
    "    maxPredict = np.argmax(prediction)\n",
    "\n",
    "    #On recupere le mot correspondant à l'indice precedent\n",
    "    word = label[maxPredict]\n",
    "    pred = prediction[0][maxPredict] * 100.\n",
    "    end = time.time()\n",
    "\n",
    "\n",
    "    #On affiche les prédictions\n",
    "    print()\n",
    "    print('----------')\n",
    "    print(\" Prediction :\")\n",
    "    for i in range(0, len(label)):\n",
    "        print('     ' + label[i] + ' : ' + \"{0:.2f}%\".format(prediction[0][i] * 100.))\n",
    "\n",
    "    print()\n",
    "    print('RESULTAT : ' + word + ' : ' + \"{0:.2f}%\".format(pred))\n",
    "    print('temps prediction : ' + \"{0:.2f}secs\".format(end-start))\n",
    "\n",
    "    \n",
    "    print('----------')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* ####  Test sur un \"audio\" particulier isolé et résultat trouvé par le modèle: "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Ici nous avons isolé une image russe pour test final.\n",
    "\n",
    "On constate que le modèle reconnait la langue avec un pourcentage de 85.03%"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 204,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Chargement du modèle :\n",
      "\n",
      "\n",
      "Model chargé.\n",
      "\n",
      "----------\n",
      " Prediction :\n",
      "     ar : 5.47%\n",
      "     fr : 0.01%\n",
      "     hi : 9.49%\n",
      "     ru : 85.03%\n",
      "\n",
      "RESULTAT : ru : 85.03%\n",
      "temps prediction : 0.16secs\n",
      "----------\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "\n",
    "modelPath = Path +'\\\\trainedModel\\\\moModel.hdf5'\n",
    "imagePath =  Path +'\\\\testImage\\\\common_voice_ru_23407397.png'\n",
    "imageSize = (50,50)\n",
    "label = ['ar', 'fr', 'hi', 'ru']\n",
    "\n",
    "predict(modelPath, imagePath,imageSize, label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 205,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "print()"
   ]
  },
  {
   "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": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}