05-2_video-action-recognition-train-rnn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<center><div style=\"direction:rtl;font-family:B Lotus, B Nazanin, Tahoma\">به نام خدا</div></center>\n",
    "<img src=\"./logo.png\" alt=\"class.vision\" style=\"width: 200px;\"/>\n",
    "<h1><center><div style=\"direction:rtl;font-family:B Lotus, B Nazanin, Tahoma\">طبقه بندی ویدیو با شبکه‌های بازگشتی</div></center></h1>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">لود کتابخانه‌های مورد استفاده</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2.0.0\n",
      "2.2.4-tf\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import tensorflow.keras as keras\n",
    "from keras.layers import Dense, Flatten, Dropout, ZeroPadding3D\n",
    "from keras.layers.recurrent import GRU\n",
    "from keras.models import Sequential, load_model\n",
    "from keras.optimizers import Adam, RMSprop\n",
    "\n",
    "import os\n",
    "import csv\n",
    "import numpy as np\n",
    "\n",
    "print(tf.__version__)\n",
    "print(keras.__version__)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">تعریف متغیرها</div>\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "تعیین کردن مسیر مجموعه داده، تعداد فریم‌ها برای تشخیص یک ویدیو (تعداد step های زمانی)، نهایت تعداد فریمی که انتظار داریم یک ویدیو داشته باشد و با بیشتر از این تعداد فریم کار نمیکنیم</div>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_path = \"D:/dataset/RNN-Video\"\n",
    "seq_length = 40\n",
    "max_frames = 300"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">خواندن مجموعه داده</div>\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "قبلا  6 کلاس از دیتاست  UCF-101 را به عنوان نمونه انتخاب و فریم‌های ویدیوهای متعلق به این 6 کلاس از این مجموعه داده را استخراج کرده ایم و اطلاعات هر ویدیو نظیر اسم - کلاس و تعداد فریم را در یک فایل متنی قرار داده ایم.\n",
    "    <br/>\n",
    "    در اینجا آن ویدیوهایی که حداقل 40 فریم و حداکثر 300 فریم دارند را لود می‌کنیم.\n",
    "<br/>\n",
    "    دیتاست اصلی شامل 101 کلاس مختلف را می‌توانید از لینک زیر دانلود کنید:\n",
    "</div>\n",
    "\n",
    "<strong>UCF-101</strong>\n",
    "[https://www.crcv.ucf.edu/data/UCF101.php](https://www.crcv.ucf.edu/data/UCF101.php)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(os.path.join(base_path, 'data_file_5class.csv'), 'r') as fin:\n",
    "    reader = csv.reader(fin)\n",
    "    data = list(reader)\n",
    "    \n",
    "train_path = os.path.join(base_path, 'train')\n",
    "classes =os.listdir(train_path)\n",
    "classes = sorted(classes)\n",
    "\n",
    "data_clean = []\n",
    "for item in data:\n",
    "    if int(item[3]) >= seq_length and int(item[3]) <= max_frames:\n",
    "        data_clean.append(item)\n",
    "data = data_clean"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_class_one_hot(class_str):\n",
    "    \"\"\"Given a class as a string, return its number in the classes\n",
    "    list. This lets us encode and one-hot it for training.\"\"\"\n",
    "    # Encode it first.\n",
    "    label_encoded = classes.index(class_str)\n",
    "\n",
    "    # Now one-hot it.\n",
    "    label_hot = keras.utils.to_categorical(label_encoded, len(classes))\n",
    "\n",
    "    return label_hot\n",
    "    \n",
    "def get_extracted_sequence(sample):\n",
    "    \"\"\"Get the saved extracted features.\"\"\"\n",
    "    filename = sample[2]\n",
    "    path = os.path.join('sequences' , filename + '-' + str(seq_length) + \\\n",
    "        '-features.npy')\n",
    "    return np.load(path)\n",
    "\n",
    "def split_train_test(data):\n",
    "    \"\"\"Split the data into train and test groups.\"\"\"\n",
    "    train = []\n",
    "    test = []\n",
    "    for item in data:\n",
    "        if item[0] == 'train':\n",
    "            train.append(item)\n",
    "        else:\n",
    "            test.append(item)\n",
    "    return train, test\n",
    "\n",
    "def get_all_sequences_in_memory(data, train_test):\n",
    "    # Get the right dataset.\n",
    "    train, test = split_train_test(data)\n",
    "    data = train if train_test == 'train' else test\n",
    "\n",
    "    print(\"Loading %d samples into memory for %sing.\" % (len(data), train_test))\n",
    "\n",
    "    X, y = [], []\n",
    "    for row in data:\n",
    "        sequence = get_extracted_sequence(row)\n",
    "        X.append(sequence)\n",
    "        y.append(get_class_one_hot(row[1]))\n",
    "\n",
    "    return np.array(X), np.array(y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loading 302 samples into memory for training.\n",
      "Loading 137 samples into memory for testing.\n"
     ]
    }
   ],
   "source": [
    "# Get data.\n",
    "X, y = get_all_sequences_in_memory(data, 'train')\n",
    "X_test, y_test = get_all_sequences_in_memory(data, 'test')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "nb_classes = len(classes)\n",
    "features_length = X.shape[-1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = Sequential()\n",
    "model.add(GRU(120, return_sequences=True,\n",
    "               input_shape=(seq_length, features_length)))\n",
    "model.add(GRU(120, return_sequences=False))\n",
    "model.add(Dropout(0.2))\n",
    "model.add(Dense(50, activation='relu'))\n",
    "model.add(Dropout(0.5))\n",
    "model.add(Dense(nb_classes, activation='softmax'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_1\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "gru_1 (GRU)                  (None, 40, 120)           780840    \n",
      "_________________________________________________________________\n",
      "gru_2 (GRU)                  (None, 120)               86760     \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 120)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 50)                6050      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 50)                0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 6)                 306       \n",
      "=================================================================\n",
      "Total params: 873,956\n",
      "Trainable params: 873,956\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Now compile the network.\n",
    "model.compile(loss='categorical_crossentropy', optimizer='rmsprop',\n",
    "                   metrics=['accuracy'])\n",
    "\n",
    "print(model.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Train on 302 samples, validate on 137 samples\n",
      "Epoch 1/20\n",
      "302/302 [==============================] - 4s 14ms/step - loss: 1.8985 - accuracy: 0.2020 - val_loss: 1.7211 - val_accuracy: 0.1679\n",
      "Epoch 2/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 1.6718 - accuracy: 0.3146 - val_loss: 1.5909 - val_accuracy: 0.2336\n",
      "Epoch 3/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 1.4664 - accuracy: 0.3808 - val_loss: 1.3258 - val_accuracy: 0.4307\n",
      "Epoch 4/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 1.1278 - accuracy: 0.5464 - val_loss: 1.1444 - val_accuracy: 0.5401\n",
      "Epoch 5/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 1.1876 - accuracy: 0.5132 - val_loss: 1.0716 - val_accuracy: 0.5620\n",
      "Epoch 6/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 0.8632 - accuracy: 0.6887 - val_loss: 0.8342 - val_accuracy: 0.7299\n",
      "Epoch 7/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.6350 - accuracy: 0.7815 - val_loss: 1.3242 - val_accuracy: 0.5182\n",
      "Epoch 8/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 0.5963 - accuracy: 0.7980 - val_loss: 1.4299 - val_accuracy: 0.5255\n",
      "Epoch 9/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.6516 - accuracy: 0.8013 - val_loss: 1.0624 - val_accuracy: 0.5912\n",
      "Epoch 10/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.3705 - accuracy: 0.9040 - val_loss: 0.9659 - val_accuracy: 0.7007\n",
      "Epoch 11/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.4026 - accuracy: 0.8940 - val_loss: 1.0333 - val_accuracy: 0.5839\n",
      "Epoch 12/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 0.2528 - accuracy: 0.9272 - val_loss: 1.5315 - val_accuracy: 0.5401\n",
      "Epoch 13/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 0.4282 - accuracy: 0.8775 - val_loss: 0.9561 - val_accuracy: 0.6934\n",
      "Epoch 14/20\n",
      "302/302 [==============================] - 2s 7ms/step - loss: 0.1086 - accuracy: 0.9801 - val_loss: 1.0507 - val_accuracy: 0.7007\n",
      "Epoch 15/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.0880 - accuracy: 0.9901 - val_loss: 2.1897 - val_accuracy: 0.4745\n",
      "Epoch 16/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.4926 - accuracy: 0.8775 - val_loss: 1.6895 - val_accuracy: 0.6350\n",
      "Epoch 17/20\n",
      "302/302 [==============================] - 2s 8ms/step - loss: 0.1643 - accuracy: 0.9669 - val_loss: 1.0169 - val_accuracy: 0.6642\n",
      "Epoch 18/20\n",
      "302/302 [==============================] - 3s 9ms/step - loss: 0.0632 - accuracy: 0.9934 - val_loss: 2.5353 - val_accuracy: 0.4307\n",
      "Epoch 19/20\n",
      "302/302 [==============================] - 3s 8ms/step - loss: 0.1357 - accuracy: 0.9669 - val_loss: 1.3398 - val_accuracy: 0.6715\n",
      "Epoch 20/20\n",
      "302/302 [==============================] - 3s 10ms/step - loss: 0.0274 - accuracy: 1.0000 - val_loss: 1.5767 - val_accuracy: 0.7007\n"
     ]
    }
   ],
   "source": [
    "history = model.fit(X, y, validation_data=(X_test, y_test),\n",
    "                    batch_size=32, epochs=20)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEICAYAAABPgw/pAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4xLjEsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy8QZhcZAAAgAElEQVR4nO3deXhTZfYH8O8BKruAgIIgLSiD0FKgVCwDCqIioAVUlKWoKMiijtssMuAPFWVGkWEYXEBQ3Kiio6MiwoiOzCCjA5YtBQGhtWAFWSqrrG3P749zU0JI0pvmLkl6Ps/Tp2lyc3N60568973nfV9iZiillIp9VdwOQCmllDU0oSulVJzQhK6UUnFCE7pSSsUJTehKKRUnNKErpVSc0ISuAiKiqkR0hIhaWLmtm4joEiKyvE6XiK4hogKfn7cQ0RVmtq3Aa71MRBMq+vwQ+32KiF6zer/KWdXcDkBZg4iO+PxYC8AJACXGz2OYOTuc/TFzCYA6Vm9bGTBzGyv2Q0SjAAxn5p4++x5lxb5VfNKEHieYuSyhGi3AUcz8ebDtiagaMxc7EZtSyhna5VJJGKfU7xDR20R0GMBwIupKRP8jogNEtIuIZhJRgrF9NSJiIkoyfp5vPL6EiA4T0ddE1DLcbY3H+xLRd0R0kIieI6L/EtGIIHGbiXEMEW0jov1ENNPnuVWJ6K9EVEREeQD6hDg+jxLRAr/7XiCi6cbtUUS0yfh98ozWc7B9FRJRT+N2LSJ604htI4DOAV4339jvRiLqb9zfHsDzAK4wurP2+Rzbx32eP9b43YuI6EMiamrm2JSHiAYa8Rwgoi+IqI3PYxOIaCcRHSKizT6/awYRrTHu301Ez5p9PWURZtavOPsCUADgGr/7ngJwEkAm5IO8JoDLAFwOOVNrBeA7APcZ21cDwACSjJ/nA9gHIB1AAoB3AMyvwLbnAzgMYIDx2MMATgEYEeR3MRPjRwDqAUgC8LP3dwdwH4CNAJoDaAhgufzJB3ydVgCOAKjts+89ANKNnzONbQhALwDHAKQaj10DoMBnX4UAehq3pwH4N4AGABIBfOu37a0AmhrvyTAjhguMx0YB+LdfnPMBPG7c7m3E2BFADQAvAvjCzLEJ8Ps/BeA143ZbI45exns0wTjuCQCSAWwH0MTYtiWAVsbtbwAMNW7XBXC52/8Lle1LW+iVywpm/piZS5n5GDN/w8wrmbmYmfMBzAHQI8Tz32PmHGY+BSAbkkjC3fYGAOuY+SPjsb9Ckn9AJmP8MzMfZOYCSPL0vtatAP7KzIXMXATg6RCvkw9gA+SDBgCuBXCAmXOMxz9m5nwWXwD4F4CAFz793ArgKWbez8zbIa1u39d9l5l3Ge/JW5AP43QT+wWALAAvM/M6Zj4OYDyAHkTU3GebYMcmlCEAFjLzF8Z79DSAcyEfrMWQD49ko9vue+PYAfLB3JqIGjLzYWZeafL3UBbRhF65/OD7AxFdSkSfENFPRHQIwGQAjUI8/yef20cR+kJosG0v9I2DmRnSog3IZIymXgvSsgzlLQBDjdvDIB9E3jhuIKKVRPQzER2AtI5DHSuvpqFiIKIRRLTe6No4AOBSk/sF5Pcr2x8zHwKwH0Azn23Cec+C7bcU8h41Y+YtAH4LeR/2GF14TYxN7wTQDsAWIlpFRP1M/h7KIprQKxf/kr2XIK3SS5j5XACTIF0KdtoF6QIBABAR4cwE5C+SGHcBuMjn5/LKKt8BcI3Rwh0ASfAgopoA3gPwZ0h3SH0AS03G8VOwGIioFYBZAMYBaGjsd7PPfssrsdwJ6cbx7q8upGvnRxNxhbPfKpD37EcAYOb5zNwN0t1SFXJcwMxbmHkIpFvtLwDeJ6IaEcaiwqAJvXKrC+AggF+IqC2AMQ685iIAaUSUSUTVADwAoLFNMb4L4EEiakZEDQE8EmpjZt4NYAWAVwFsYeatxkPVAZwDYC+AEiK6AcDVYcQwgYjqk9Tp3+fzWB1I0t4L+WwbBWmhe+0G0Nx7ETiAtwGMJKJUIqoOSaxfMnPQM54wYu5PRD2N1/495LrHSiJqS0RXGa93zPgqgfwCtxFRI6NFf9D43UojjEWFQRN65fZbAHdA/llfgrRQbWUkzcEApgMoAnAxgLWQunmrY5wF6evOhVywe8/Ec96CXOR8yyfmAwAeAvAB5MLiIMgHkxmPQc4UCgAsAfCGz349AGYCWGVscykA337nzwBsBbCbiHy7TrzP/yek6+MD4/ktIP3qEWHmjZBjPgvyYdMHQH+jP706gKmQ6x4/Qc4IHjWe2g/AJpIqqmkABjPzyUjjUeaRdGEq5Q4iqgo5xR/EzF+6HY9SsUxb6MpxRNSHiOoZp+3/B6mcWOVyWErFPE3oyg3dAeRDTtv7ABjIzMG6XJRSJmmXi1JKxQltoSulVJxwbXKuRo0acVJSklsvr5RSMWn16tX7mDlgqa9rCT0pKQk5OTluvbxSSsUkIgo64lm7XJRSKk5oQldKqTihCV0ppeJEVK1YdOrUKRQWFuL48eNuh6JMqFGjBpo3b46EhGBTjSilnBRVCb2wsBB169ZFUlISZBI+Fa2YGUVFRSgsLETLli3Lf4JSynZR1eVy/PhxNGzYUJN5DCAiNGzYUM+mlIoiUZXQAWgyjyH6XikVXaIuoSullJ3+8Q9g5063o7BHuQmdiC4iomXGiucbieiBANv0JFnBfZ3xNcmecO1VVFSEjh07omPHjmjSpAmaNWtW9vPJk+amdb7zzjuxZcuWkNu88MILyM7ODrmNWd27d8e6dess2ZdS8e7wYeDmm4Hp092OxB5mLooWA/gtM68xlrhaTUSfMfO3ftt9ycw3WB9icNnZwMSJwI4dQIsWwJQpQFYE0/s3bNiwLDk+/vjjqFOnDn73u9+dsU3Z6tpVAn8Wvvrqq+W+zr333lvxIJVSFZZvLGe9YYO7cdil3Ba6sSL5GuP2YQCbEHoNSEdkZwOjRwPbtwPM8n30aLnfatu2bUNKSgrGjh2LtLQ07Nq1C6NHj0Z6ejqSk5MxefLksm29Lebi4mLUr18f48ePR4cOHdC1a1fs2bMHAPDoo49ixowZZduPHz8eXbp0QZs2bfDVV18BAH755RfcfPPN6NChA4YOHYr09PRyW+Lz589H+/btkZKSggkTJgAAiouLcdttt5XdP3PmTADAX//6V7Rr1w4dOnTA8OHDLT9mSkUjb0LfuNHdOOwSVh86ESUB6IQzl8ny6mqsXr6EiJKDPH80EeUQUc7evXvDDtbXxInA0aNn3nf0qNxvh2+//RYjR47E2rVr0axZMzz99NPIycnB+vXr8dlnn+Hbb/1PWICDBw+iR48eWL9+Pbp27Yp58+YF3DczY9WqVXj22WfLPhyee+45NGnSBOvXr8f48eOxdu3akPEVFhbi0UcfxbJly7B27Vr897//xaJFi7B69Wrs27cPubm52LBhA26//XYAwNSpU7Fu3TqsX78ezz//fIRHR6nYkJcn3wsLgQMH3I3FDqYTOhHVAfA+gAeZ+ZDfw2sAJDJzBwDPAfgw0D6YeQ4zpzNzeuPGodYFLt+OHeHdH6mLL74Yl112WdnPb7/9NtLS0pCWloZNmzYFTOg1a9ZE3759AQCdO3dGQUFBwH3fdNNNZ22zYsUKDBkyBADQoUMHJCcH/Iwss3LlSvTq1QuNGjVCQkIChg0bhuXLl+OSSy7Bli1b8MADD+DTTz9FvXr1AADJyckYPnw4srOzdWCQqjS8LXQACPAvG/NMJXRj5e/3AWQz8z/8H2fmQ8x8xLi9GEACETWyNFI/LVqEd3+kateuXXZ769at+Nvf/oYvvvgCHo8Hffr0CViPfc4555Tdrlq1KoqLiwPuu3r16mdtE+7CI8G2b9iwITweD7p3746ZM2dizJgxAIBPP/0UY8eOxapVq5Ceno6SkpKwXk+pWJSfD3jbkvHYj26myoUAvAJgEzMHvDZMRE2M7UBEXYz9FlkZqL8pU4Batc68r1Ytud9uhw4dQt26dXHuuedi165d+PTTTy1/je7du+Pdd98FAOTm5gY8A/CVkZGBZcuWoaioCMXFxViwYAF69OiBvXv3gplxyy234IknnsCaNWtQUlKCwsJC9OrVC88++yz27t2Lo/79V0rFofx8oEcPoHbt+EzoZqpcugG4DUAuEXmvyk0A0AIAmHk2gEEAxhFRMYBjAIawzWvbeatZrKxyMSstLQ3t2rVDSkoKWrVqhW7duln+Gr/5zW9w++23IzU1FWlpaUhJSSnrLgmkefPmmDx5Mnr27AlmRmZmJq6//nqsWbMGI0eOBDODiPDMM8+guLgYw4YNw+HDh1FaWopHHnkEdevWtfx3UCqalJQABQVStpicHJ8XRl1bUzQ9PZ39F7jYtGkT2rZt60o80aa4uBjFxcWoUaMGtm7dit69e2Pr1q2oVi2qpt/R90zFjO3bgaQkYM4c4OuvgU8+AXbvdjuq8BHRamZOD/RYdGUHVebIkSO4+uqrUVxcDGbGSy+9FHXJXKlY4r0g2qqVDDB69VVg797TferxQDNElKpfvz5Wr17tdhhKxQ1vQr/4YsBbn7BxI9Czp2shWU7nclFKVQp5eUC1akDz5kBKitwXb/3omtCVUpVCfj6QmChJ/cILgXr14q/SRRO6UqpSyM+X/nMAIJJWurbQlVIqBuXlSf+5V3KytNBdKvSzhSZ0Hz179jxrkNCMGTNwzz33hHxenTp1AAA7d+7EoEGDgu7bv0zT34wZM84Y4NOvXz8csGDCiccffxzTpk2LeD9KxaoDB4Cffz7dQgekhb5/P/DTT+7FZTVN6D6GDh2KBQsWnHHfggULMHToUFPPv/DCC/Hee+9V+PX9E/rixYtRv379Cu9PKSW+/16++yZ07/RI8dSPrgndx6BBg7Bo0SKcOHECAFBQUICdO3eie/fuZXXhaWlpaN++PT766KOznl9QUIAU4/L5sWPHMGTIEKSmpmLw4ME4duxY2Xbjxo0rm3r3scceAwDMnDkTO3fuxFVXXYWrrroKAJCUlIR9+/YBAKZPn46UlBSkpKSUTb1bUFCAtm3b4u6770ZycjJ69+59xusEsm7dOmRkZCA1NRU33ngj9u/fX/b67dq1Q2pqatmkYP/5z3/KFvjo1KkTDh8+XOFjq5SbfEsWveKx0iVq69AffBCweiGejh0BIxcG1LBhQ3Tp0gX//Oc/MWDAACxYsACDBw8GEaFGjRr44IMPcO6552Lfvn3IyMhA//79g66rOWvWLNSqVQsejwcejwdpaWllj02ZMgXnnXceSkpKcPXVV8Pj8eD+++/H9OnTsWzZMjRqdOa8ZqtXr8arr76KlStXgplx+eWXo0ePHmjQoAG2bt2Kt99+G3PnzsWtt96K999/P+T85rfffjuee+459OjRA5MmTcITTzyBGTNm4Omnn8b333+P6tWrl3XzTJs2DS+88AK6deuGI0eOoEaNGmEcbaWih3fa3JYtT993/vlAo0baQo9rvt0uvt0tzIwJEyYgNTUV11xzDX788UfsDjFuePny5WWJNTU1FampqWWPvfvuu0hLS0OnTp2wcePGcifeWrFiBW688UbUrl0bderUwU033YQvv/wSANCyZUt07NgRQOgpegGZn/3AgQPo0aMHAOCOO+7A8uXLy2LMysrC/Pnzy0akduvWDQ8//DBmzpyJAwcO6EhVFbPy84GGDaVU0Ve8VbpE7X9oqJa0nQYOHIiHH34Ya9aswbFjx8pa1tnZ2di7dy9Wr16NhIQEJCUlBZwy11eg1vv333+PadOm4ZtvvkGDBg0wYsSIcvcTar4d79S7gEy/W16XSzCffPIJli9fjoULF+LJJ5/Exo0bMX78eFx//fVYvHgxMjIy8Pnnn+PSSy+t0P6VcpNvyaKv5GTgjTek0iXIyXZM0Ra6nzp16qBnz5646667zrgYevDgQZx//vlISEjAsmXLsH379pD7ufLKK8sWgt6wYQM8Hg8AmXq3du3aqFevHnbv3o0lS5aUPadu3boB+6mvvPJKfPjhhzh69Ch++eUXfPDBB7jiiivC/t3q1auHBg0alLXu33zzTfTo0QOlpaX44YcfcNVVV2Hq1Kk4cOAAjhw5gry8PLRv3x6PPPII0tPTsXnz5rBfU6lokJ9/Zv+5V0qKzOvyww/Ox2SHqG2hu2no0KG46aabzqh4ycrKQmZmJtLT09GxY8dyW6rjxo3DnXfeidTUVHTs2BFdunQBIKsPderUCcnJyWdNvTt69Gj07dsXTZs2xbJly8ruT0tLw4gRI8r2MWrUKHTq1Clk90owr7/+OsaOHYujR4+iVatWePXVV1FSUoLhw4fj4MGDYGY89NBDqF+/Pv7v//4Py5YtQ9WqVdGuXbuy1ZeUiiXFxTLT4uDBZz/mvTC6YYN9i+M4SafPVRHR90xFu++/l+6Wl18GRo4887H9+4HzzgOmTgV+/3t34gtXqOlztctFKRXXfKfN9deggczrEi+VLprQlVJxLVANuq94Wr0o6hK6W11AKnz6XlVO3313ej7xWJCXByQkAM2aBX48JQX49lugtNTZuOwQVQm9Ro0aKCoq0kQRA5gZRUVFOtiokikqkgT4pz+5HYl5+fmy9FzVqoEfT04Gjh07PT1ALIuqKpfmzZujsLAQe/fudTsUZUKNGjXQvHlzt8NQDlq/Hjh1Cpg9G/jjH6XlG+2C1aB7+Va6BOuWiRVRldATEhLQ0ndsrlIqqhjDKbBrF7BoEXDjje7GY0Z+PnD55cEfb9dOvm/cCAwY4ExMdomqLhelVHTLzZVFlZs3l1Z6tNu/X75CtdDr1pWVjOKh0kUTulLKNI8H6NABuPtuYOnS05NeRatQJYu+vItdxDpN6EopU0pKJOmlpsoAnapVgTlz3I4qNLMJPSUF2LJFrg/EMk3oSilTtm0Djh+XhN6sGdC/PzBvHmAsHxCVvGcQZlroJ0/K7xjLNKErpUzJzZXv7dvL97FjgX37gH/8w72YypOfL33+deuG3i5eFrvQhK6UMsXjAapUOV0Vcs010vKN5ouj5ZUsel16qUyfG+v96JrQlVKmeDxAmzaAdyxZlSrAmDHA8uUy0jIaBZs211+tWrKdttCVUpVCbu7p7havO++UwUUvveROTKGcOgXs2GGuhQ7ER6WLJnSlVLkOH5bWrs9KigCkf3rQIOD114GjR92JLZgdO6Qyx2xCT0kBtm6N7ou85Sk3oRPRRUS0jIg2EdFGInogwDZERDOJaBsReYgoLdC+lFKxydty9U/ogFwcPXgQeOcdZ2Mqj9mSRa/kZPkA2LLFvpjsZqaFXgzgt8zcFkAGgHuJqJ3fNn0BtDa+RgOYZWmUSilX+Ve4+LriCqBt2+i7OFretLn+4qHSpdyEzsy7mHmNcfswgE0A/CeiHADgDRb/A1CfiJpaHq1SyhUez+kh8v6IpJW+ahWwZo3zsQWTlwecc44sYGHGr34lg6ViuR89rD50IkoC0AnASr+HmgHwXWa1EGcnfRDRaCLKIaIcnVFRqdjh8Uh3C1Hgx2+7DahZM7oujubnAy1bSjWOGdWrS1KP6xa6FxHVAfA+gAeZ+ZD/wwGectak5sw8h5nTmTm9cePG4UWqlHIFc+AKF18NGgBDhgDZ2cAh/+zgErM16L5SUipBC52IEiDJPJuZA40LKwRwkc/PzQHsjDw8pZTbCguBAwcCXxD1NXYs8MsvktTdxixdLuHOb56cLB8E0VaxY5aZKhcC8AqATcw8PchmCwHcblS7ZAA4yMy7LIxTKeUS7xzo5SX0yy4DOnUCZs2ShOqmn3+WM4WKtNCZgU2b7InLbmZa6N0A3AagFxGtM776EdFYIhprbLMYQD6AbQDmArjHnnCVUk7zVrh4q0CC8V4czc0F/vc/++MKJdySRa/kZPkeq/3oZqpcVjAzMXMqM3c0vhYz82xmnm1sw8x8LzNfzMztmTnHjmCzs2VtwCpV5Hs0nNopFe88HqluqVev/G2HDpVqGLdLGCua0C+5RCpjYrUfPWZGimZnA6NHA9u3yynR9u3ysyZ1pezlrXAxo25dYPhwGWT088/2xhWK2Wlz/VWrJhN1xW0LPVpMnHj2hYqjR+V+pZQ9TpwANm8OXeHib8wYed7rr9sXV3ny84ELLgBq1w7/ubFc6RIzCX3HjvDuV0pFbvNmGQ5vtoUOyBJ1XbtKt4tbF0crUrLolZwseSVayi/DETMJvUWL8O5XSkXObIWLv7Fjge++A/79b8tDMsXstLmBeC/+RuuUwKHETEKfMkXmLPZVq5bcr5Syh8cjIyhbtw7vebfcIoON3Lg4evIk8MMPkbXQgdjsR4+ZhJ6VJQvSJiZKeVRiovycleV2ZErFr9xcWaGoWrXwnlezJjBihCxPt3u3LaEFtX07UFpa8YTesqXEH4v96DGT0AFJ3gUF8mYVFGgyV8pu4VS4+BszBiguloWknVTRkkUv7zJ72kJXSsWNffuAXbvCq3Dx1aYNcNVVciZdUmJtbKGEO21uILFa6aIJXSkVkHeEaEVb6IBcHC0oAJYutSQkU/LyZN3TJk0qvo/kZPkwc7OWviI0oSulAqpohYuvgQOB88939uJouNPmBhKri11oQldKBeTxyJqhF1xQ8X2ccw4wciSwaJFUnjghkhp0L2+lS6x1u2hCV0oFlJsbWevc6+67ZYDRyy9Hvq/yMEdWg+510UUyjYG20JVSMa+kRFqnViT0li2BPn2AuXOBU6ci318o+/YBhw9H3kInkla6ttCVUjEvLw84dqziFS7+xo6Vi4yLFlmzv2AiLVn05a10cXtu93BoQldKncWKChdf/foBzZvbf3HUyoSenAwUFQF79kS+L6fEXEI/dgx45RVn61qVqmw8ntMDbKxQrZr0pS9denpqWzt4E3rLlpHvKxYrXWIuob/1FjBqFNC9e2wdaKViiccj87fUrGndPkeOBKpWlYFGdsnLA5o2PXvep4qIxUqXmEvod90FzJ8PbN0q6xc+9pjMvWyGrniklDmRDPkPplkzoH9/mQrA7P9suKwoWfRq0gQ477zYajjGXEInkjlcNm0CBg8GJk+WxP7f/4Z+nq54pJQ5R45IYrQ6oQNycXTfPuCDD6zfN2BNyaIXUexNARBzCd2rcWPgzTeBJUuAX36RLph77w0+Kb2ueKSUOd4EZkdCv+Ya6RJ5/33r933iBFBYaF0LHZBul40bY6fSJWYTulefPnLAH3gAmDVLLuJ8/PHZ2+mKR0qZ4x3yb1XJoq8qVYAbbgA+/VTmLbdSQYEkXisTekoKcPAg8OOP1u3TTjGf0AGgTh1gxgzg669lUv3+/aU7xnceZl3xSClzcnNllGRioj37799fBv/85z/W7tfKkkWvWFvsIi4SutfllwOrVwNPPgl8+CHQti3w6qvyqa0rHilljscjrfNIJrcK5eqrpXpm4UJr92vFtLn+Yq3SJa4SOiCTAT36KLB+vbwZd90FXHstkJGhKx4pVR7m0wndLjVrSl/6xx9b2zedlyf7jmQyMX+NGsn+tIXusksvlVO6WbOAVavkD3TnTmDbNl3xSKlgfvwROHDAnguivvr3l0oz74hUK3hLFoms2ycQW5UucZvQATllHDtWVu++9lrgD3+Qbpm1a92OTKnoZMUc6GZcf718D1TAUFFW1qD7Sk6WHFJaav2+rRbXCd2reXPpU3/3XWmBdOkCfPGF21EpFX28Cd077N0uTZsCl11mXUK3atrcQFJSpDR6+3Zr9vfzz/aVQVaKhA7Iadgtt8gnbevWwNCh0gWjlDotN1cqv+rXt/+1+vcHVq4Efvop8n3t2SNJ164WOmBNP/ru3UDnzjLC3Q6VJqF7nXce8N57Mhpu8GD752dWKpbYMeQ/mMxM+f7JJ5Hvy46SRS+rKl2OHQMGDJCk7v3drVbpEjogg4/mzgVWrAAmTHA7GqWiw8mTwObN9la4+EpNlZWBrOh2sTOh16sn3baRtNBLS4Hbb5cCjexs6W6yQ6VM6AAwbBhwzz3AtGn2zSuhVCCffQYcP+52FGfbvBkoLnauhU4k3S5Ll0rrNRJWTpsbSKSVLhMmSM/As88CN95oXVz+yk3oRDSPiPYQUcBfh4h6EtFBIlpnfE2yPkx7TJ8OpKcDI0bYO0ezUl5ffgn07i0jm6ONUxUuvjIzJZlHWqSQlyezOdaoYU1c/pKTZULAiqzD8PLLwDPPSMXdww9bH5svMy301wD0KWebL5m5o/E1OfKwnFG9OvD3v8sczYMGRd5KUKo8s2bJ92ic5dPjkYF5rVs795o9e8rUHZF2u9hVsuiVkiKTf4Xb8PvsM0nk110HPPec9TXy/spN6My8HMDP9obhnqQkmbVx3TrgN78Jva3Op64isWePnHZfeKGcvntbxNHC45HrSwkJzr1m9eqS7CIdNWp3Qq9IpcvGjdJQbNdOSqarVbMnNl9W9aF3JaL1RLSEiJKDbUREo4koh4hy9u7da9FLR+7662Ua3VdekblfAtH51FWkXn1VqqoWLJCzwrfecjuiM+XmOtvd4pWZKSXEa9ZU7PnHjsn4Ejtq0L28S/GZ7UffvVvySq1asjD2uefaF5svKxL6GgCJzNwBwHMAPgy2ITPPYeZ0Zk5v3LixBS9tnSeeAHr1kgul69ef/bjOp64iUVoKvPQS0KMHcMUV0ip9++3oGX24b58kVacqXHz16ydnvRWdrKugQL7b2UKvXVsuuJpJ6EePysXePXvkzMPJGV0jTujMfIiZjxi3FwNIIKJGEUfmMG+LqUEDOU06ePDMx3U+dRWJpUuB77+X/lRAqqx27Ch/pS2neOdUcaOF3rgx0LVrxfvR7SxZ9JWSUn6Xi7c88ZtvJJ+kp9sbk7+IEzoRNSGSrn4i6mLssyjS/brhgguAd96Rf7yRI8/s09P51FUkZs+WxHXTTfLzgAFyOh4t3S5uJnRAul3WrpUVh8Jlx7S5gSQnA1u2hF6Y449/lNWYpk0DBg60N55AzJQtvg3gawBtiKiQiEYS0VgiMtoaGARgAxGtBzATwBDmWFmw6WxXXAE8/bS8Kb6lZTqfujWOHpU+xco0QveHH6T1OXKkVJEAUtkxYIBcLLN65Z6K8HhOT9QBargAABriSURBVBXrhv795XtFWul5edIlYncvbkqK1Olv3Rr48blzgalTgXHjgIcesjeWoJjZla/OnTtztCotZR44kLlaNeYVK07fP38+c2IiM5F8nz/frQhjz4kTzM8/z9y0KTPA/OSTbkfknEmT5G8mP//M+z/+WI7Fxx+7E5evyy5j7tXLvdcvLWW++GLmvn3Df25mJnP79tbH5G/dOnm/Fiw4+7GlS5mrVmXu04f51Cl74wCQw0HyaqUdKRoKkVQkJCbKfC979sj9WVlyAUbnUzevuBh47TWgTRvgvvvktLhHDxlo4btEYLw6dUpabn36nD2K8brrgIYN3e92KSmRvmG3uluA06NGv/hC5lkKh90li15t2sjFW/9+9A0bTpcnvvOOM+WJwWhCD6J+fakZ3rdPEndFRohVZqWl0p2QkgLceackriVLgOXLZaWo48elsijeffwxsGuXnIb7S0iQGUA/+ij8JGal/HzpCnOjwsVXZqYM3vnsM/PPsXPaXH81asigK99Kl59+kvLE2rVlkjGnyhOD0YQeQseOwAsvAJ9/XjmSjxWY5Q+7c2c5u6laVa5HfPONtFKJgF/9Sqo95syR+UPi2ezZMgFVv36BH8/KkmT60UfOxuXLjSH/gXTvLhNhhdOP/tNPUofuRAsdkAuj3ha6tzxx3z6J+aKLnIkhFE3o5Rg5UlqYTz4pLUwV3LJlQLduwA03AIcOAW+8IcnippvOHvI8aZK0ah55xJ1YnbB1q7Q2775bPtgC+fWvpVLKzQFqubnSleAdPOOWhAT54Fu0yPwZsVMli14pKbKM5dGjwG23ATk50mXWubMzr18eTegmPP+8tF6GD9e680BWrpRFf3v1kuMze7a0vG+7LXgia9xYSrwWLpS1X+PRnDny+48aFXybKlWkJn3pUsCtwdMeD3DJJWdXcbkhM1OOw6pV5rZ3OqEnJ0t34uDBwD/+IRP8DRjgzGuboQndhFq1pD/91Cnp84yGMrNo4PHIH3NGhoyunT5dWqVjxpibD+SBB+Q09Xe/i54Rk1Y5flwurA8cKMuthTJsmLRI333Xmdj8ObmoRXn69JEPQbPdLvn5cvaXlGRrWGW8S/MtWiSjyh94wJnXNUsTukmtW8s/6KpVkoAqs+++kyX8OnaU1vWTT8o/1kMPATVrmt9PzZpSx5+TI/ObxJP33gOKigJfDPXXvr0kCjeqXY4ckfcuWhJ6gwbAlVeanwYgL08Wn6he3d64vFq3lgufffsCf/ub/bMnhksTehhuvlmS1nPPyUXS2B0+VXHr10vyWbgQGD9eksGjjwJ161Zsf1lZQKdO0v0SjYs+VNTs2fLPf9VV5rbPygK++kpGKTtp40b5O3a7wsVXZqbEZeZYOFWy6JWQIOsSf/SRu+WJwWhCD9PUqbIgxuOPS9dCcbHbETlr5kz5o/7uO+BPf5I1WiNRpYoMk96xQz4o40FurszRMnas/H5mDB0q399+2764AomWChdf3vU2zXS7OJ3QAVlIw8kphsOhCT1M1aoB8+bJLItz50oFh/8sjPFq/37pFhg+XP6ordKrl9TyTpki3RSxbvZs6QK44w7zz0lMlLK97Gxnz/w8HpmGwKk+aDMuuQRo27b8bpejR6XG34ka9FihCb0CiICnngJefFFqrnv1klrUePfaa9ItYqZfOFxTpwKHD0t/fCw7ckQWTLn1VhlMFY5hw+R03smFL3JzpQvN7JmEUzIz5fqM/6ynvrxdMk630KNZlL2NsWXcOBk0s3691BM73f/ppNJSWT7t17+Wi6FWa9dOyvteeEHqfGPVW2/JB1NFPvRuuUXOAJ26OMocXRUuvvr3l+7MTz8Nvo3TJYuxQBN6hAYOlJGk+/bJnM4VXXUl2n3+uZQk3nuvfa/xxBPSVfHHP9r3GnZilg+91FQp5QxXo0bOLnzx44/SjRaNCT0jQ45HqG4Xp6bNjSWa0C3QrZtcBKteXSaeWrrU7Yis9+KLMhjo5pvte40mTYA//EFK/r76yr7XscuqVbI27bhxFS9ny8qS6XZXrLA2tkDcngM9lKpVZdTo4sXBCw/y8qS6KtyurXimCd0ibdsCX38tp3/XXw/Mn+92RNbZsUMqDkaNsr/e97e/lYE4v/td7JWFzp4tFxgjmYWzf3/nFr7w9tV7B8tEm/795Qwi2Ie7t8Il2mrB3aQJ3UIXXiizCV5xhQx7nzrV2qSUnS3VCFWqyHen5v+YM0d+jzFj7H+t2rXlwujXX8v1iVjx888yOGr48IrX5APy+w8cCPz97/aPSPZ4ZKRugwb2vk5F9e4tC4IE63Zxo2Qx2mlCt1i9ejKJ19ChMvHUAw9YM/VudjYwejSwfbsk1+3b5We7k/rJk1KeecMNUlrnhBEjpNU4fnzsTLPwxhtSAeRdMzQSWVnyARHqgqAVcnOjs7vFq25doGfPwPXopaVShKD952fShG6D6tWly+W3v5XBMkOGRD4KcuLEs+vdjx6V++30/vuywMc999j7Or6qVgWefVb6SGfNcu51K4pZulsyMoAOHSLf37XXygVBOz+sT54ENm2KrhGigfTvL4PYtmw58/5du+R/SlvoZ9KEbhPvCMjp0+UiX+/e0h9YUcFmebR79scXX5RWUO/e9r6Ov+uuk8Q2eXJkx80J//63JByr6vMTEqSOfeFCKYG0w+bNcrExmlvogJwZAme30rVkMTBN6DZ76CHpW125UkYC/vBDxfbTokV491vB45Fqi3HjnB94QiSt9P37ZYqBaDZ7tvRD33KLdfscNkwWbrBr4YtornDxlZgoZz2a0M3RhO6AwYOBf/5T+vxatjw93Wc4p9RTppw9X3WtWnK/XWbNkmW3Royw7zVC6dBBhs/PnClruEajn36SebHvvDO8mSbL8+tf23vh2+ORM4Ff/cqe/VspM1MaFr7TQuTnSyPDqes6sUITukN27pS+Vu8F0u3bpQzQbHljVpZUmyQmygdCYqL8bNdC1YcOyRD2IUPcrfN98knpU58wwb0YQpk3T7ourK4AIpIL6599dnqRcit5PDI6N1onmfKVmSkXQX1XDMvLkwqdc85xL65opAndIRMnnn1h9PhxadlNmWJu2oCsLGmplpbKd7uSOSDJ/JdfnL0YGkjz5sDDD8voyW++cTcWfyUlwEsvAVdfbU9L186FL6K9wsVXeroMOvPtdtGSxcA0oTsk2MXL4mKZT7xVK+ljnz3b/RkHmWVOlcsuky+3PfIIcP750TfY6J//lPfVilLFQFJSJOlaPcioqEiG/Ud7hYtXlSpycXTJktNlrJrQA9OE7pBgFy8TE6W1/ac/yQXAceNkpKR3cMmxY46GCUBmudu0yf3WuVfdujL//PLl4a0Ib7dZs6TlaOeaksOGySAr70VAK8TKBVFfmZlS8bN8uZw57t6tNeiBaEJ3SKiLmomJMiHVhg3A2rXA/ffLvCC33ioJY+RIYNky59bdfPFFqdoYPNiZ1zNj1Cjg0ktlrpdTp9yORq6BLF4scdnZD23HwhfRuKhFea65Ri7Qf/yxVriEogndIWYuahLJ1LTTpkl54+efywIaf/+7zLmemCgJzdvCssPOncAHHwB33WVt1UakEhKAZ56Reu+5c92ORmIgAu6+297XadFCppKwcuGL3Fy50N2kiTX7c0KtWpLUFy6UC6KAJvSAmNmVr86dO7My55dfmBcsYM7MZK5WjRlgfuUVe17riSdk/1u32rP/SJSWMvfowdy4MfOPP7oXx4kTzBdcIO+HE2bPlvdk7Vpr9telC/NVV1mzLye99JIch7vuku9FRW5H5A4AORwkr2oLPQbUqiXdHwsXypDn3r2lTO5f/7L2dU6dkqqN666TZcCiDRHwl78ABw5I6+zee92pT//oI+nDtetiqL9Bg6xb+KKoSLr2Yqm7xcs7ajQ7W+ZMitZJxdykCT3GNGokZWxt2sjc5N9+a92+Fy6ULhc7F7GIVOfOsiL8bbdJt8cllwC3327tcSjPrFky6Oe665x5vYYNgb59I1v4Yvt2mSiuRQuZA6hPH2tjdMKFF0oJ44kTOm1uMOUmdCKaR0R7iGhDkMeJiGYS0TYi8hBRmvVhKl/16slapjVqyNzrVg08efFF+Yfv18+a/dmldWtJ5vn5cgH5/feB5GTgxhvlYrId8vOlDz89XS5Qjx4tA56cMmwYUFgIfPlleM/LzZUPv4svlvd30CBpocdiQgdksi5A+8+DCtYX4/0CcCWANAAbgjzeD8ASAAQgA8DK8vbJ2oduiVWrmGvWZM7IYD56NLJ9ffut9Ev+6U/WxOakvXuZJ01ibtBAfoerr2b+/HPpc4/Etm3Mf/4zc1qa7Bdgvuwy5mnTpB/dSUeOMNeuzTx6dPnblpYyL1/O3K+fxFy7NvODDzJv325/nHZbu1Z+pz/8we1I3IMQfeimLmACSAqR0F8CMNTn5y0Ampa3T03o1nj/fWYi5ltvZS4pqfh+7r+fOSGBefdu62Jz2qFDzM8+y9ykifxld+nC/MEH4R2X775jnjKFuWPH00n88ssliX//vW2hmzJ8uHxoBfswKSlh/vBD5q5dJe5GjZgnT46vi4elpcyPPca8YYPbkbjH7oS+CEB3n5//BSA9yLajAeQAyGnRooVDv378mzpV3skJEyr2/CNHmM89l3nYMGvjcsuxY1IZ0rKlHJd27ZjffJP51KnA22/ZwvzUU8wdOpxO4hkZzH/5S3S1ahcvltg++ujM+0+cYJ43j7ltW3k8KYn5+eelOkrFH7sT+icBEnrn8vapLXTrlJbKqTgg/9jh8paD/fe/1sfmplOnmLOzmVNSTie6F1+UhL9pk7Re27c/ncS7dmWePp15xw63Iw/s5Ekp2bz1Vvn50CH50GnWTOLv0IH5rbeCf3Cp+KBdLpXAyZPM114rder/+pf555WWMqemSjKItM85WpWUSKs2I0P+4mvVOp3Eu3VjnjGD+Ycf3I7SnHvvZa5Rg/mPf2SuX19+h549mZcsid/3T50pVEK3omxxIYDbjWqXDAAHmXmXBftVYUhIkBGl3nLGzZvNPe/rr2Uo+D33xG8ZWJUqUh3x1VdSoTJkCPC3v0nVyIoVUs7XvLnbUZqTlSWzdD79tIweXrlSfqc+feL3/VPmkST8EBsQvQ2gJ4BGAHYDeAxAAgAw82wiIgDPA+gD4CiAO5k5p7wXTk9P55yccjdTYSooAC6/XFaPX7kSaNw49PbDh8v8GD/+CNSp40iIKgLMsqRhaqp8eKvKh4hWM3N6wMfKS+h20YRun1WrgB49gE6dgC++kHr1QPbskUUCxoyRVYGUUtEvVELXkaJxqEsXWQnp669l+bhgowtfeUXml46WaXKVUpHRhB6nbr5ZRja+8w4wadLZj5eUyGIavXrJtLRKqdinCT2O/f73Ml/3lCnAa6/JfdnZMg9JtWqy2k5KipsRKqWsVM3tAJR9iGT+joICmXtk61ZgxgyZnMlr7lzporFzfVKllDO0hR7nvOWMrVtLqZtvMgdkibuJE92JTSllLU3olUD9+jI7Y7CLo8EWsFZKxRZN6JVEUlLwJceCLWCtlIotmtArkWnTgHPOOfM+70LVSqnYpwm9EsnKAubNkxZ5sIWqlVKxS6tcKpmsLE3gSsUrbaErpVSc0ISulFJxQhO6UkrFCU3oSikVJzShK6VUnNCErpRScUITulJKxQlN6EopFSc0oSulVJzQhK5iineBjipV5Ht2ttsRKRU9NKGrsLiZULOzZaGO7dsBZvk+erQmdaW8NKEr09xOqBMnnr1Ax9GjukCHUl6a0JVpbifUYAtx6AIdSglN6Mo0KxJqJF02wRbi0AU6lBKa0JVpkSbUSLtspkyRBTl86QIdSp2mCV2ZFmlCjbTLJitLFuRITNQFOpQKhJjZlRdOT0/nnJwcV15bVVx2tiTgHTukZT5livmEWqWKtMz9EQVfwFopdSYiWs3M6YEe0xWLVFgiWfGoRQvpZgl0v1IqctrlohyjfeBK2UsTunKM9oErZS/tclGO0kWqlbKPqRY6EfUhoi1EtI2Ixgd4fAQR7SWidcbXKOtDVUrFOp2Lx17lttCJqCqAFwBcC6AQwDdEtJCZv/Xb9B1mvs+GGJVSccA7DsFbuuodhwDoWZtVzLTQuwDYxsz5zHwSwAIAA+wNSykVb9yeOqIyMJPQmwH4wefnQuM+fzcTkYeI3iOiiwLtiIhGE1EOEeXs3bu3AuEqpWKVzsVjPzMJnQLc5z885GMAScycCuBzAK8H2hEzz2HmdGZOb9y4cXiRKqVims7FYz8zCb0QgG+LuzmAnb4bMHMRM58wfpwLoLM14SmlfMXyRUUdh2A/Mwn9GwCtiaglEZ0DYAiAhb4bEFFTnx/7A9hkXYhKKcD9+egjpeMQ7FduQmfmYgD3AfgUkqjfZeaNRDSZiPobm91PRBuJaD2A+wGMsCtgpdzkZgs5Hi4qZmUBBQUyd09BgSZzq+nkXEqZ5F92B0iXgVOtTJ3cTAGhJ+fSof9KmeR2C1kvKqryaEJXlUokXSZul93pRUVVHk3oqtKI9KKi2y1kvaioyqN96KrSSEoKPB97YqJcoCuP233oSgHah64UgMi7TLSFrKKdTp+rKg0rVkzS6X9VNNMWuqo09KKiinea0FWloV0mKt5pl4uqVLTLRMUzbaErpVSc0ISulINiebZEFf20y0Uph+gSbMpu2kJXyiFuzwWj4p8mdKUc4vZcMCr+aUJXyiFuzwWj4p8mdKUcogOblN00oSvlEB3YpOymCV0pB+kSbJHRss/QtGxRKRUTtOyzfNpCV0rFhGgo+4z2MwRtoSulYoLbZZ+xcIagLXSlVExwu+wzGs4QyqMJXalKJNq7DEJxu+zT7TMEMzShK1VJRLpItncfbn0guF326fYZghm6SLRSlYQukh2ZaPn9dZFopVTEXQax0IdcnkjOMNw+QzBDW+hKVRKRttCrVJGuGn9EMlAq2kVLCztS2kJXSkV8UTEW+pBDiYczjPJoQleqkoi0y8DtKpNIRUOVit0XlXVgkVKVSCSLZHufN3GiJMEWLSSZx0p3RYsWgbucnDrDcGJgkqkWOhH1IaItRLSNiMYHeLw6Eb1jPL6SiJKsCU8pFU1ieXIxt88wnOjyKTehE1FVAC8A6AugHYChRNTOb7ORAPYz8yUA/grgGetCVEqpyLldpeJEl4+ZFnoXANuYOZ+ZTwJYAGCA3zYDALxu3H4PwNVERNaFqZRSkXPzDMOJi8pmEnozAD/4/Fxo3BdwG2YuBnAQQEP/HRHRaCLKIaKcvXv3VixipZSKQU50+ZhJ6IFa2v7VqGa2ATPPYeZ0Zk5v3LixmfiUUiouONHlY6bKpRDART4/NwewM8g2hURUDUA9AD9bEqFSSsWJSKqMzDDTQv8GQGsiaklE5wAYAmCh3zYLAdxh3B4E4At2awiqUkpVUuW20Jm5mIjuA/ApgKoA5jHzRiKaDCCHmRcCeAXAm0S0DdIyH2Jn0Eoppc5mamARMy8GsNjvvkk+t48DuMXa0JRSSoVDh/4rpVSc0ISulFJxwrXpc4loL4AAMyuY0gjAPgvDsVq0xwdEf4waX2Q0vshEc3yJzByw7tu1hB4JIsoJNh9wNIj2+IDoj1Hji4zGF5lojy8Y7XJRSqk4oQldKaXiRKwm9DluB1COaI8PiP4YNb7IaHyRifb4AorJPnSllFJni9UWulJKKT+a0JVSKk5EdUKP5qXviOgiIlpGRJuIaCMRPRBgm55EdJCI1hlfkwLty8YYC4go13jtnACPExHNNI6fh4jSHIytjc9xWUdEh4joQb9tHD9+RDSPiPYQ0Qaf+84jos+IaKvxvUGQ595hbLOViO4ItI1N8T1LRJuN9/ADIqof5Lkh/x5sjO9xIvrR533sF+S5If/fbYzvHZ/YCohoXZDn2n78IsbMUfkFmQgsD0ArAOcAWA+gnd829wCYbdweAuAdB+NrCiDNuF0XwHcB4usJYJGLx7AAQKMQj/cDsAQyn30GgJUuvtc/QQZMuHr8AFwJIA3ABp/7pgIYb9weD+CZAM87D0C+8b2BcbuBQ/H1BlDNuP1MoPjM/D3YGN/jAH5n4m8g5P+7XfH5Pf4XAJPcOn6RfkVzCz2ql75j5l3MvMa4fRjAJpy9klO0GwDgDRb/A1CfiJq6EMfVAPKYuaIjhy3DzMtx9lz+vn9nrwMYGOCp1wH4jJl/Zub9AD4D0MeJ+Jh5KctKYQDwP8iaBa4IcvzMMPP/HrFQ8Rm541YAb1v9uk6J5oRu2dJ3djO6ejoBWBng4a5EtJ6IlhBRsqOByapRS4loNRGNDvC4mWPshCEI/k/k5vHzuoCZdwHyQQ7g/ADbRMuxvAty1hVIeX8PdrrP6BKaF6TLKhqO3xUAdjPz1iCPu3n8TInmhG7Z0nd2IqI6AN4H8CAzH/J7eA2kG6EDgOcAfOhkbAC6MXMagL4A7iWiK/0ej4bjdw6A/gD+HuBht49fOKLhWE4EUAwgO8gm5f092GUWgIsBdASwC9Kt4c/14wdgKEK3zt06fqZFc0IPZ+k7kAtL3xFRAiSZZzPzP/wfZ+ZDzHzEuL0YQAIRNXIqPmbeaXzfA+ADyGmtLzPH2G59Aaxh5t3+D7h9/Hzs9nZFGd/3BNjG1WNpXIS9AUAWGx2+/kz8PdiCmXczcwkzlwKYG+R13T5+1QDcBOCdYNu4dfzCEc0JPaqXvjP6214BsImZpwfZpom3T5+IukCOd5FD8dUmorre25ALZxv8NlsI4Haj2iUDwEFv14KDgraK3Dx+fnz/zu4A8FGAbT4F0JuIGhhdCr2N+2xHRH0APAKgPzMfDbKNmb8Hu+LzvS5zY5DXNfP/bqdrAGxm5sJAD7p5/MLi9lXZUF+QKozvIFe/Jxr3TYb84QJADcip+jYAqwC0cjC27pBTQg+AdcZXPwBjAYw1trkPwEbIFfv/Afi1g/G1Ml53vRGD9/j5xkcAXjCOby6AdIff31qQBF3P5z5Xjx/kw2UXgFOQVuNIyHWZfwHYanw/z9g2HcDLPs+9y/hb3AbgTgfj2wbpf/b+HXorvy4EsDjU34ND8b1p/H15IEm6qX98xs9n/b87EZ9x/2vevzufbR0/fpF+6dB/pZSKE9Hc5aKUUioMmtCVUipOaEJXSqk4oQldKaXihCZ0pZSKE5rQlVIqTmhCV0qpOPH/ZtYwMVXUbN8AAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "acc = history.history['accuracy']\n",
    "val_acc = history.history['val_accuracy']\n",
    "loss = history.history['loss']\n",
    "val_loss = history.history['val_loss']\n",
    "\n",
    "epochs = range(len(acc))\n",
    "\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "Y_pred = model.predict(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_pred = np.argmax(Y_pred, axis=1)\n",
    "y_test = np.argmax(y_test, axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion Matrix\n",
      "[[ 6 12  1  0  6  0]\n",
      " [ 1 12  1  0  9  0]\n",
      " [ 0  0 22  0  0  1]\n",
      " [ 0  0  0 16  3  0]\n",
      " [ 0  3  1  1 25  0]\n",
      " [ 0  0  2  0  0 15]]\n",
      "Classification Report\n",
      "                    precision    recall  f1-score   support\n",
      "\n",
      "    CricketBowling       0.86      0.24      0.38        25\n",
      "       CricketShot       0.44      0.52      0.48        23\n",
      "FieldHockeyPenalty       0.81      0.96      0.88        23\n",
      "  HandstandPushups       0.94      0.84      0.89        19\n",
      "  HandstandWalking       0.58      0.83      0.68        30\n",
      "     SoccerPenalty       0.94      0.88      0.91        17\n",
      "\n",
      "          accuracy                           0.70       137\n",
      "         macro avg       0.76      0.71      0.70       137\n",
      "      weighted avg       0.74      0.70      0.68       137\n",
      "\n"
     ]
    }
   ],
   "source": [
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "\n",
    "print('Confusion Matrix')\n",
    "print(confusion_matrix(y_test, y_pred))\n",
    "print('Classification Report')\n",
    "print(classification_report(y_test, y_pred, target_names=classes))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\">منابع</div>\n",
    "\n",
    "<div style=\"direction:rtl;text-align:right;font-family:Tahoma\">\n",
    "برای این فایل آموزشی از مخزن زیر کمک گرفته شده است\n",
    "    </div>\n",
    "    \n",
    "https://github.com/harvitronix/five-video-classification-methods"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<div class=\"alert alert-block alert-info\">\n",
    "<div style=\"direction:rtl;text-align:right;font-family:B Lotus, B Nazanin, Tahoma\"> دوره پیشرفته یادگیری عمیق<br>علیرضا اخوان پور<br>  آبان و آذر 1399<br>\n",
    "</div>\n",
    "<a href=\"http://class.vision\">Class.Vision</a> - <a href=\"http://AkhavanPour.ir\">AkhavanPour.ir</a> - <a href=\"https://github.com/Alireza-Akhavan/\">GitHub</a>\n",
    "\n",
    "</div>"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "tf2-GPU",
   "language": "python",
   "name": "tf2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}