Model Evaluation and Attack.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "# %load_ext autoreload\n",
    "# %autoreload 2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import importlib\n",
    "from model.fasterrcnn import *\n",
    "from model.advattack import *\n",
    "from data_utils.utils import *\n",
    "from data_utils.arraytools import *\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import pickle\n",
    "from tqdm import tqdm\n",
    "\n",
    "device = torch.device('cuda') if torch.cuda.is_available() else torch.device('cpu')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<All keys matched successfully>"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = load_Faster_RCNN(backbone='resnet18')\n",
    "\n",
    "model.to(device)\n",
    "\n",
    "checkpoint = torch.load('./saved_models/fasterrcnn_resnet18_2021-04-12_final.pth')\n",
    "model.load_state_dict(checkpoint['model'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_img_list, test_bboxes = wider_read(limit_images=None, train=False)\n",
    "\n",
    "test_dataset = WiderDataset(test_img_list, test_bboxes)\n",
    "\n",
    "test_loader = DataLoader(test_dataset,\n",
    "                         batch_size=1,\n",
    "                         shuffle=True,\n",
    "                         num_workers=0,\n",
    "                         collate_fn=collate_fn\n",
    "                         )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time elapsed in Predicting: 18.62 minutes\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "\n",
    "prediction_info = []\n",
    "target_info = []\n",
    "model.eval()\n",
    "\n",
    "for images, targets in test_loader:\n",
    "    images = list(image.to(device) for image in images)\n",
    "    targets = [{k: v.to(device) for k, v in t.items()} for t in targets]\n",
    "\n",
    "    with torch.no_grad():\n",
    "        predictions = model(images)\n",
    "    \n",
    "    predictions = [{k: v.to('cpu').detach() for k, v in t.items()} for t in predictions]\n",
    "    targets = [{k: v.to('cpu').detach() for k, v in t.items()} for t in targets]\n",
    "    prediction_info.append(predictions)\n",
    "    target_info.append(targets)\n",
    "    images = list(image.detach() for image in images)\n",
    "    images.clear()\n",
    "#     targets.clear()\n",
    "    torch.cuda.empty_cache()\n",
    "\n",
    "end = time.time()\n",
    "print(f\"Time elapsed in Predicting: {(end - start)/60:.2f} minutes\")\n",
    "prediction_info = list(itertools.chain(*prediction_info))\n",
    "target_info = list(itertools.chain(*target_info))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# model.to('cpu')\n",
    "# torch.cuda.empty_cache()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# importlib.reload(data_utils.arraytools)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Processing : 100%|██████████| 3135/3135 [00:15<00:00, 198.05it/s]\n",
      "Processing : 100%|██████████| 3135/3135 [00:15<00:00, 196.37it/s]\n",
      "Processing : 100%|██████████| 3135/3135 [00:15<00:00, 197.32it/s]\n",
      "Processing : 100%|██████████| 3135/3135 [00:15<00:00, 197.77it/s]\n",
      "Processing : 100%|██████████| 3135/3135 [00:15<00:00, 199.38it/s]\n",
      "Processing :  99%|█████████▉| 3119/3135 [00:15<00:00, 191.29it/s]"
     ]
    }
   ],
   "source": [
    "results = {}\n",
    "for i in range(3,11):\n",
    "    r_temp = evaluation(prediction_info, target_info, iou_thresh=i/10, interpolation_method='EveryPoint')\n",
    "    results[i/10] = r_temp\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For the Threshold 0.3: \n",
      "Average Precision: 0.64 Total Recall: 0.70\n",
      "\n",
      "For the Threshold 0.4: \n",
      "Average Precision: 0.63 Total Recall: 0.68\n",
      "\n",
      "For the Threshold 0.5: \n",
      "Average Precision: 0.61 Total Recall: 0.66\n",
      "\n",
      "For the Threshold 0.6: \n",
      "Average Precision: 0.57 Total Recall: 0.62\n",
      "\n",
      "For the Threshold 0.7: \n",
      "Average Precision: 0.49 Total Recall: 0.55\n",
      "\n",
      "For the Threshold 0.8: \n",
      "Average Precision: 0.32 Total Recall: 0.40\n",
      "\n",
      "For the Threshold 0.9: \n",
      "Average Precision: 0.06 Total Recall: 0.16\n",
      "\n",
      "For the Threshold 1.0: \n",
      "Average Precision: 0.00 Total Recall: 0.01\n",
      "\n"
     ]
    }
   ],
   "source": [
    "for thres, result in results.items():\n",
    "    prec = result['total TP']/(result['total TP'] + result['total FP'])\n",
    "    rec = result['total TP']/result['total positives']\n",
    "    print(f\"For the Threshold {thres}: \\nAverage Precision: {result['AP']:.2f} Total Recall: {rec:.2f}\")\n",
    "#     PlotPrecisionRecallCurve(result)\n",
    "    print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0AAAAFcCAYAAADh6N6LAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuNCwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8QVMy6AAAACXBIWXMAAAsTAAALEwEAmpwYAABccElEQVR4nO3dd3iUddbG8e9JJ4RAaKFXKSKdiICUgEqxd8WCqIjYXVd33X3tbW27dsWyimvDshZ0QV2V2FApikiXTugdQoCQ5Lx/zJANJMBQhsmQ+3NdczHz/H7PM/fkEMjJ08zdERERERERKQ9iIh1ARERERETkUFEDJCIiIiIi5YYaIBERERERKTfUAImIiIiISLmhBkhERERERMoNNUAiIiIiIlJuqAESEREAzOxCM/s8hHnDzez2Q5GpLDGzwWb2XbHXbmZHRDKTiIjsOzVAIiJRwMwWmNkWM8sxsxVm9oqZpRzM93D3N9y9bwjzhrn7vQfzvfeVmd1lZtuDX4/1ZjbOzLpGMpOIiEQHNUAiItHjFHdPAToCRwO37TrBzOIOearIeTv49agOjAXejXCeg87MYiOdQUTkcKMGSEQkyrj7EmAM0BqKDsW6xsx+B34PLjvZzCYX2zvSdsf6ZlbfzN43s1VmtsbMng4uLzrEywIeM7OVZrbBzKaY2Y73G2Fm9xXb3hVmNsfM1prZKDOrU2zMzWyYmf1uZuvM7Bkzs9I+l5mNNrO/F3v9tpm9HMLXIx94A6hrZjWC61Y2s3+a2TIzW2Jm9xVvJoKZZ5jZJjObbmYdg8tvNbO5xZafsdeClP5Zqgb30i0Nfu4Pg8t3OowuuKzoULrg1/a54NdiM/AXM1u+S/YzzGxK8HlMscxrzOwdM6u6P5lFRMoLNUAiIlHGzOoDJwK/FFt8OnAM0Cr4w/zLwJVANeB5YJSZJQZ/kP4EWAg0AuoCI0t5m75AT6A5UAU4D1hTSpY+wN+Ac4Hawe3uur2TCeyxahec1283H+0y4GIz62NmFwbXuWE3c4tnSAAGBfOtCy5+FcgHjgA6BD/PkOD8c4C7guukAqcW+2xzgR5AZeBu4HUzq723DKV4DUgGjgJqAo/tw7oXAPcDlYBHgc1An13G3ww+v55A7XsBdQh8/mf2I6+ISLmhBkhEJHp8aGbrge+Ar4EHio39zd3XuvsW4ArgeXf/yd0L3P1VYBvQBehM4AflW9x9s7tvdffvKGk7gR/AWwLm7jPcfVkp8y4EXnb3n919G/AXoKuZNSo250F3X+/uiwgcqta+tA/n7suBYQSalyeAQe6+aQ9fj3ODX48dn/lsd883s3RgAHBj8DOuJNCAnB9cbwjwsLtP8IA57r4wmOFdd1/q7oXu/jaBPWqd95ChhGDDNAAY5u7r3H27u3+9D5v4yN2/D2bYCrwFDAxuuxKB5vet4Nwrgf9z9+zg1/8u4OxydiikiMg+UQMkIhI9Tnf3Ku7e0N2vDjY7Oywu9rwh8Mfg4W/rg01CfQKNT31gYfCwsd1y96+ApwnsTVhhZi+YWWopU+sQ2OuzY70cAntT6habs7zY81xgTxdv+ASIBWbtpjEr7h13rwKkA1OBTsHlDYF4YFmxz/88gT0xEPgazC1tg2Y2qNihg+sJHGZYfS85dlUfWOvu6/Y6s3SLd3n9JnCmmSUCZwI/72jYCHzWD4rlnQEUEPiaiIhIKdQAiYgcHrzY88XA/cFmaccj2d3fCo41CGUPgbs/6e6dCBzG1Ry4pZRpSwn8EA6AmVUkcNjdkv38HPcT+CG+tpkNDGUFd19NYE/IXcG9L4sJ7PGqXuzzp7r7UcFVFgNNd92OmTUEXgSuBaoFm6upQKnnLO3BYqCqmVUpZWwzgUPjdrxnrdI+0k4v3KcTaDIHsPPhbzvea8AutU4KnicmIiKlUAMkInL4eREYZmbHBC9mUNHMTgoePjUeWAY8GFyeZGbH7roBMzs6uH48gR/atxLYs7CrN4FLzax9cA/FA8BP7r5gX0ObWU/gUgLn5gwCnjKzunteK8DdZwKfAX8KHqr3OfB3M0sNXiigqZn1Ck5/CbjZzDoFvz5HBJufigSaj1XBPJcSvNDEvgi+/xjgWTNLM7P44GcD+BU4Kvj1SiJwyFoo3iRwvk9Pdr7a3XDg/mB+zKyGmZ22r5lFRMoTNUAiIocZd59I4JyYpwmcFD8HGBwcKwBOIXBxgEVANoELHOwqlUAjtY7A3oc1BE7I3/W9vgRuB/5NoLFqyv/OtQlZ8PC6fwHXuvuS4OFv/wRe2d1V40rxCDDUzGoSaKASgOnBz/AegYs04O7vEtjT9CawCfgQqBrc0/J34AdgBdAG+H5fP0vQxQTOo5oJrARuDL73bOAe4AsC5xft7TC/Hd4CMoGvgnu8dngCGAV8bmabgB8JXAxDRER2w9x977NEREREREQOA9oDJCIiIiIi5YYaIBERERERKTfUAImIiIiISLmhBkhERERERMoNNUAiIiIiIlJuqAESEZEDZmZZZrYueC+g4stHmFmemeWY2Voz+6+ZtdyH7XY0s2+C668wsxuKjd1rZr+ZWb6Z3bWX7dxiZlPNbJOZzTezW4qN1TSzt8xsqZltMLPvzeyYYuPtzGyama02sz8UWx5vZj+ZWf1QP4+IiESeGiARETkgZtYI6EHgJqKnljLlYXdPAeoRuCfOiBC3Wx34FHgeqEbg3kWfF5syB/gT8J9QNkfg3kBpQH/gWjPbcb+iFGAC0AmoCrwK/MfMUoLjfwNuBtoBt5lZreDym4B/u/viUD6PiIiUDboPkIiIHBAzuwPoB/wENHf3k4uNjQCy3f224OuTgLeDDdHetvsAUN/dL97LvNeBOe5+1z5kfpLA/4HX7WZ8I9Db3SeZ2QygvbtvM7MfgeuB5QRurnqsu28P9X1FRCTytAdIREQO1CDgjeCjn5mllzYpuEflQuCX4OvuZrZ+D9vtAqw1s3FmttLMPjazBgca1syMwB6rabsZbw8kENjDBDAV6Gtm9YBGwFzgSeBPan5ERKKPGiAREdlvZtYdaAi84+6TCDQHF+wy7eZgozOHwOFmgwHc/Tt3r7KHzdcDLgFuABoA84G3DkLsuwj8//fKrgNmlgq8Btzt7ht25AeuAkYBfwCOBTYB88zsIzP72szOOQi5RETkEIiLdAAREYlqlwCfu/vq4Os3g8seKzbn0R2HwO2jLcAH7j4BwMzuBlabWeVizck+MbNrCeyx6uHu23YZqwB8DPzo7n/bsdzdFwInBuckA+MIHPL3FPA2gXOQpprZl+6+dn9yiYjIoaMGSERE9kuwYTgXiDWz5cHFiUAVM2vn7r8e4FtMIXBhhR12PLf92ZiZXQbcCvR09+xdxhKBD4ElwJV72MwdwEvuvsLM2gC3ufsGM8smcJGG8fuTTUREDh0dAiciIvvrdKAAaAW0Dz6OBL4lsJflQL0CnGFm7c0sHrgd+M7d10PRZaiTCPxfFmdmSWYWW9qGzOxC4AHgBHeft8tYPIELGmwBBrl74W620QrIBJ4LLpoP9Ame89QMWHQgH1ZERA4NNUAiIrK/LgFecfdF7r58xwN4GrjQzPZ4lIGZ9TCznN2Nu/tXwF8JHGK2ksAeluLnF71IoGkZCPxf8PnFu9n2fQQupT0heE+hHDMbHhzrBpwM9AXWFxvvsUukZ4Ab3L0g+PovBK4INw14IPjZRUSkjNNlsEVEREREpNzQHiARERERESk31ACJiIiIiEi5oQZIRERERETKDTVAIiIiIiJSbqgBEhERERGRciPqboRavXp1b9SoUaRjFNm8eTMVK1aMdAwJgWoVPVSr6KFaRQfVKXqoVtFDtSrbJk2atNrda5Q2FnUNUKNGjZg4cWKkYxTJysoiMzMz0jEkBKpV9FCtoodqFR1Up+ihWkUP1apsM7OFuxvTIXAiIiIiIlJuqAESEREREZFyQw2QiIiIiIiUG1F3DpCIiIhItMvLy2Pu3Lnk5ubutLxSpUpMmjQpQqlkX6hWZUNycjJNmzYlISEh5HXUAImIiIgcYnPnzqVKlSq0aNGCmBgdkCOyPwoLC1m+fDnTp0+ndu3apKenh7SevuNEREREDrHc3FzS09PV/IgcgJiYGGrVqkV+fj4jR45k/vz5oa0X5lwiIiIiUgo1PyIHLiYmBjMjOTmZ77//PrR1whXGzF42s5VmNnU342ZmT5rZHDObYmYdw5VFRERERHYWGxtL+/btad26Neecc06J85H2xx133MEXX3yx2/Hhw4fzr3/964DfR2RXCQkJbN26NaS54fzVwwig/x7GBwDNgo+hwHNhzCIiIiIixVSoUIHJkyczdepUEhISGD58+E7jBQUF+7zNe+65h+OPP36348OGDWPQoEH7vF2RgylsF0Fw92/MrNEeppwG/MvdHfjRzKqYWW13XxauTAeTuzPq4x+xDz9i/Oc/7TS2tFtvttSsRUr2QtInjiuxbnbPE9hWtTqVFsyh5uQJJcYX9TmR7amVqTxnJtWn/lJifGHfU8lPrkjazN+oOrPkDrb5J51FYXwC1ab+QpU5M0uMzz3tfDCj+uQJVF4wZ+fB2rVIOesMzKBZXB5xsYEe2YFCB09IIK5SCrjDxo3/W88s8GdCAiQlBcY3by7x3iQkBB6FhVC8S9+xfnw8xMUFxrdvD7y3O+5Q6I7HxVEYE4MXFOL5BTiBMcxwwM3wmBi80KHQcZyk+FiS4mMD23T/33uJiIgIAD169GDKlClkZWVx9913U7t2bSZPnsxvv/3GrbfeSlZWFtu2beOaa67hyiuvBODhhx/mtddeIyYmhgEDBvDggw8yePBgTj75ZM4++2xuvfVWRo0aRVxcHH379uXRRx/lrrvuIiUlhZtvvpnJkyczbNgwcnNzadq0KS+//DJpaWlkZmZyzDHHMHbsWNavX88///lPevToEeGvkBxOInkVuLrA4mKvs4PLSjRAZjaUwF4i0tPTycrKOhT59sjd+eC1rxnx3uMlxgaefz8/NGzHyTO+4elRD5cYvz87nl/rtOC8Xz/joU+fKjH+l5WVmVu9PpdP+JDbv3qpxPj162uxPLU613//Fjd990aJ8cu3NGZTYkVuHfsvho1/v8T4eQWtcIvhvs/+yQmTx+w0tqRSDY5dUh2AWY+eQVxBoAkxIBb4V4eTuKPvVdRNKuD7u08rse0Rx57Nk8dfSqXcTXz9yPklxp/IvJjnjz2PWhtW8tXTl5UYv+f4obyacSpHrFrIZ/+8pui9jcDuylsGXM+7bfvSYclMPnj95hLrX33arYxu2Z3u83/h9Xdu32ksE7j83Lv5tmkn+s4ax9PvP1A0VmAxbI1P5KKL/sb0Wkdw+pQv+esXLwDgBBsmM8675FEWVq3L2ZNGc+O3bxBjOxKCG1w07CnWVqrKwHEfcOG494vW9eC88659jtzEZC79eiSnT/o0OLajITPOuOEl3OGWj56i/co5bElKpjAmloqbNzGrZXteO+0KAM4e8xp1Vi0pWteB5TXq8m7/iwAY+J8RVF2/Gg/mBlhUuyFvdz+LCnHG4NGvkLZmBeubtSBm6xby6tVjVaMmrK3XCNu2jebTJ5EQH0tMfgH5NaoRGxdLbno6eamVidu6lZRVK0hMiMUtBo+NwWKMbZWrEJOcRBKFxOTl/e+9g4/CuDiIiYHCQsy9xHhxOTk5ZeL7XPZOtYoOqlPZU6lSpUhHKJKfn8+YMWPo3z9w4M748eOZOnUqjRs35oUXXqBy5cpMmDCBbdu2ceyxx9K3b19mzpzJhx9+yE8//URycjJr167daZtr167lgw8+YObMmZgZ69evL/G+gwYN4qmnnqJXr17ccccd3H333Tz++ONFmcaPH8/o0aO5++6793hYnQjAokWLKCgoCOnfukg2QKX9Gt5Lm+juLwAvAGRkZHhmZmYYY4XG3TmyVQfGnHYknTpl7DT2WLXqkJgIuV1Z+eDQEus+X70GJCRgm7uwcsMNJcbfqFET4uOxnGNYufHWEuMfpteC2Fhs49GszLm7xPh/a9WGmBjs6gxWbi7ZgP1Quw6YYVd1YmXu//bSbFiQzZoV6xie0YlJC9fy7aZbifGCogbEzFiQVIuT2tQmxgt4/6KbAj+7FtupsvmI1hx3ZF0S8rbxn4tvKlpuBoaR3KoD57ZsROKW6nyxKbg+jmFg0LB9V65s3pSKG6rybeEfCKy1Y33o2KUXjY5oQcVVlfkp8abAXyL3ovfp3TOTjo2bUWlpRSal3MjW7QVsyy8gLiaGDRs3cPQJnWlapyHVGjjfJl29o5jEb92CFRbQp/tRZNSsQ90qnZhlZwT2GAUmAdCvW0s2p1WH3JbMLOhbtNyC89q1bMCWipVI3nQUv+etCYzhReNHt6jL9oQkkte0YlHBeqxo84EnXVvUIW/ZCgrq1WfL9g3E5m+n4aLZrE+tStWVy1mVnwRAtRXLaLBk7k7bL9xeyJqCwHjdxfOpv3IRtiMfsG27M31tIdUqJnDiF+8Sg8MvXxfV/+/dL+TZYwdSZ+NKxj13Z4m/N7cMuIF3255Ax+wZvP/GLSXGrzrtVsa07M6Amd/x3EcPlhgffPk/mNzgKE79+TPu+egfJcZPG/oss2s2ZuDEUfzl8xdwDDdjY3IltiYmc/0Nz7ElJZVTv/uAzB9Hk5AYT9qGNSyq04SCxCTe/fNjmEHnL/5No5m/UGnDWubXakJq1UrExsbw1cBraFg1mbo/jKXChnUUbNlKYt06VKyaSmKtGmxv044YM+KXZkNMLJVrppFSOQVLSNBewz3IysqiLPybLHumOpU9xe8fc/fH05i+dOMeZu+7VnVSufOUo/Y4Z8uWLbRv3x4I7AG6/PLLGTduHJ07d6Zx48YAfP7550yZMoX33nsPgA0bNvD777/zxRdfcOmll5KcnAxA1apVd9p2amoqSUlJDBkyhJNOOomTTz55p/ENGzawfv16evXqBcAll1zCOeecUzR+5plnAtCpUycWLFiwf18EKVcaNGjA9u3bQ/q3LpINUDZQv9jresDSCGXZZ2ZGrfQqVKiXTs0jm5Y+qXIS1K5a+tiO8TrV9jxet/qex6m5l/E9XA+9cq2dXtY8sinNgs/7t64FJ91TYpVexV9cdPTutw1wYec9j593zF7Gu5ZY1K3o2RFwbvcS4/9L1ATOy9xpLCsri2FF3xRHwrCd/zEGKEp0Uiu45cIS40Wf6ORWwHUlxo/d8eT8DsCNJca77DReUtH2bxqw0/LKQEPgqx0Lbs4ssW5d4MsdL/74U4nxfsCOi0P6X/PZunkLeVu2smHBEijI54wqVTm1ajXWrcvht35f4Hl55CxaQkzwkMXjm7WiW+062Op6jGn0D5JjDQoKMS+EwgJ6te9CWkJVGjbpwXtJuVRMiAk0x4FWhqO6tKFBjVrUSu3GVxXzMPdg8xaYk9n9KDpXrU69Kj34krWkVqpETE4OsZs2YgUFNKpdmc3xFdgUm8jWKlVJXb2UtUmVSFi5gpjYWL6evYpCd46aOoO240aTHxNL/d8mkJSfR3ZqTZ6rHfjN5jtvPExG9vSdvjY/1W/NeRcEmrbvnruUehtX7TT+UdvjuePMW/DCQkY/PghPTKIwJoYKWzazLr0un2X0Z84p55Gcu4lLnrwVA+LytrEpNY2C1m1Y3usENrQ4ioQVy6k/8TsSE+Px2FiwGIiLZcvRXUloUJf49WupMH8uBTGxJCcnERsfR2ysQfMWJFasgG3bhm3bRkxCApXSKhETqytIicj+23EO0K4qVqxY9Nzdeeqpp+jXr99Ocz799FNsD78ciouLY/z48Xz55ZeMHDmSp59+mq+++mq383eVmJgIBC7UkJ+fH/J6IqGIZAM0CrjWzEYS+LlzQ7Sc/yMS7SwmhqRKFUmqVJHUmrs04emVoGXtPaxdD05oV+pI4KDHNjD0xBJjfYuetQbOKjHec8eTk1qR1ak23Xb5DU7RftZBO+9x3eHHHU/+L3Dy7Y5/3AoLCqm+cRM/xSZSUOj44E/IztnM5mUrWL1hK5afR35SMk+0bI07LEm7hyUzppNDHDFeSEx+PlsbHMEZHeqyYMUGsjt0JTZvG9UW/M7y2o3Yum07ufmFTMleT6XNGyhct55GKxdScVsu+RZD3Pgv+cOKGD5o7Ry7YDJvvH1bieyDz76LrKYZu997Fhy/6JfR3Pf5s0XL82LjSCjI55S/f0VMYgJ9Pn6VbluWsaVGOp6YRH6VNJIaNWBej77Uq1KBlGm/kpK/DWJj8dgYsBi2p1SisHkLaqYmUSPOSUhOChyqKCKHzN721ERSv379eO655+jTpw/x8fHMnj2bunXr0rdvX+655x4uuOCCokPgiu8FysnJITc3lxNPPJEuXbpwxBFH7LTdypUrk5aWxrfffkuPHj147bXXivYGiYRb2BogM3uLwGkX1c0sG7gTiAdw9+HAaOBEYA6QC1wariwiUn7FxMaQlFaZpB0LqjQOPmlNi9JW6DBszxu84qMSizoDRQer3nX6/9670Fm7aQt/LXRujY0ld31n5t90KlZYgBcWQn4hXpDPVVVrcXlyReJWNmTKye3YkLOVRBzyt1O4aDHH9z+VPmlpVG5pjG9eheSl2WyKr0DCvDnEbM+jRtUUfp+zlHbzfuWoBb8SW1hAYkHgN6bTazbm9jV1APh4xA0cuWLuTtl/rN+a8y94kIT87cz++xlFy7fHxFJoMXzStg+3n3wjedvymPnIGSyvXIPCpCQKYuPJj43j94bteefSRNLycjntzSfYUgip8caGth2xxCRiMnuR26ARtQq3krRwHsnJiSTXr0uVRvWIjY/k7+BEJBRDhgxhwYIFdOzYEXenRo0afPjhh/Tv35/JkyeTkZFBQkICJ554Ig888L9zazdt2sRpp53G1q1bcXcee+yxEtt+9dVXiy6C0KRJE1555ZVD+dGkHDP3Uk+7KbMyMjJ84sSJkY5RRMdVRw/VKnqoVgfG3dm2NY+Ni5eRl5fP+rQarNmcR+rUycRt2oQVFgQuRlFQwLrEiqxtfzTLFy6n4TuvkkQBMfkFWGEhMYX5LGt8JFO692fDwiWc/eZjmDtxhQXE5W+nyoolvNu+L5/3Pocaa5fxj2dvpO4uhw9ef8rNjGqVSe+5E3jlvZ3PWdwal8Cfzv0/vm/ZlfN//4Yr3n+K+Q1bUjFnA+uatmRrfCJfXnIjNdKrccTimaQtXURypWQSqqWR0q4NKTXSAs1tfOyh/PJGJX1PlT2TJk2iU6dOkY4hcliYNGkSU6dOZfv27QwZMgQAM5vk7qUeNqJfv4mIHGbMjKQKiSQ1bwQETrAEoPkJu1+pXR04tfT7UQcOWGwN1/crMdYlK4tbd/xg/bcLANi+ZSubspexcWMuFyekcF6FirCyEd90bUjKyqVsXbyEzXkF1Foyny3tOpJRO43q0zYzv05T6mTPJT8untRJ31Nxaw7D2p3Plpg13PXfFxjw8yc7vfe22Hgyh77AstQaPPXxI/T5/SfyY2LZlFKFWIPZjVrx8MW3c0q7OtT/9guqpFYgtVY1kqtXg4QECtPSiE+vSZ0qFUiI02F/IiLlhRogERE5qOIrJFG1WWOqAo12LDyiOnQreZ7DizueXFz6uV0z3MndXsDyS47i95W3U7hqNbnZy9i4aCmpa1dxfN8MKleIx/L7M6FJQ5K25FBx5TISV61gbe52pi3dyLQlG5j01J+otmXnq2yNbNuXWwdcT1xBPnMePZ0lqTXYWqkKG2MTiK+QyPhuJ7L2vIvIW7qM4yd8Rk5qVZKbNKBavXQqpSYT26gRVK5MpcRYkhL036mISLTQv9giIlJ2mZGcEEeTIxvBkY1KDBddT7FfyVsGNAdOLigkJzePTad8ydJlq9myfAXx27cTk7eNKlVr88emzdk6Zx5fHHcOKdtyiVu5kiZLZlN18wbebHIsb46dQ7uls/jrayUvTnHFmbfx32ZdOG7OTzz+8aNsTahAYsF2VlepwYZ2nZh61S2k1K9NjfWrqJW3iSoN6pDauAEJSQkH8yskIiL7SA2QiIgctuJjY0irlERa95J7mNoB/QGOawZXljy874HgY+u2vqx9aBDr5ixk/ex5bFy7kfj8PAa06UTLhMpUqJnLb1vPIG3uLAq35VFl01qO+vwDrm00gCWVV/Knr0fQ/cf3ira7LTaOzYnJHHvT26Qlx3PLrM9ovGgWBX37UbNTG2p160R8xeRwfUlERMo9NUAiIiJ7kJQYT1Lt6lStXR16lHLSet8WcNM5Oy0qKChk9NZ85q7ZjPdKY8oPndm0eCmx27aSk7OFgu35NKiaTCIFtBjzHq1Wzoexo4DAFfg+bdaFq874C6e0rUOPrA9oVSMZ79CByjXSSGlUnyr10jFdrlxEZL+oARIRETnIYmNjqFwxgY4VE6BBNxjQrcScHfuc/IY5bFy6iuxfpjF//G+kTJ3MrMIUmlSvyIRvf+XJF+4tse4jPS7mmW7n0SF/Hf8YcSvJOJvjEthWrTokVWBc95PYevpZdEzcRuWZU9m8JYdtGUeTmFKxxLZERMobNUAiIiIRZDExpNZLp1W9dFqd0geAXsANwfGVf+zDyl+ms23OPHJytpK3YgUpTVpzYoNabF+Qx5qGzchbsYSCxCTqLPydKps38HCTPoz9bBaDJ47iri9foBXAtUPITq1Jcv423njzK5LSqnDEjElUnDWN+GZHYNWqktakAXXatSRe92gqF1JSUsjJydnjnMcff5yhQ4eSnBzewzJHjBjBxIkTefrpp3c7Jysri4SEBLp1K/kLhT1p1KgREydOpHr16gcac6+WLl3K9ddfz3vvvbfbOd26dWPcuHEH5f1uvPFGzjzzTHr27Ln3yWXIxIkT+de//sWTTz652znr16/nzTff5OqrrwZg1apVXHzxxXz66acH/P76F05ERKQMq9m0ATWbNthp2f8uaN4J/u+MXVdheH4BC1bnsmpKDSadm0n2z7+QsHwtbb77lK2xMTz+fTYFMct49Z0H6Tz/553WzUmoQI87RtGwekWu/tf9JKxfS35qFUitRExKCtur12Tb9TdQL60CdQu3UrlODR2Odxh7/PHHueiii/apASooKCA29uDfnysrK4uUlJR9boAORH5+PnFxof+4XKdOnT02P8BBa37Wrl3Ljz/+yOOPP35Qtleaff38ocrIyCAjo/Srf+6wfv16nn322aIGqEaNGtSuXZvvv/+eY4899oDeX/9iiYiIHGYS42JpUasS3ft2ptPQgVQ+/0QGfPgS9VZnU2ftMmb97WR+vv0Emo55n9+/+pHJ737Gb8+9xq9X3MRXp19Gi9qpOLBuWwFp61bRZPpE2nwzmmPff5nCL77gytcmcdKT32FHNKUwLp71FSoxo1ZTFlSty8cX38TdH/7GLSPG8dMpFzH5xtuZ/swIsn/8hYINGyHKbsBeHuy4Ue7ZZ59Ny5YtufDCC3F3nnzySZYuXUrv3r3p3bs3AJ9//jldu3alY8eOnHPOOUV7kBo1asQ999xD9+7deffdd8nMzOTGG2+kW7dutG7dmvHjxwOBH9pPP/102rZtS5cuXZgyZUqJPB9//DHHHHMMHTp04Pjjj2fFihUsWLCA4cOH89hjj9G+fXu+/fZbVq1axVlnncXRRx/N0Ucfzffffw/AmjVr6Nu3Lx06dODKK6/Ed/N3LiUlhT/+8Y907NiR4447jlWrAjdyzszM5K9//Su9evXiiSeeYNKkSfTq1YtOnTrRr18/li1bBsCcOXM4/vjjadeuHR07dmTu3LksWLCA1q1bAzBt2jQ6d+5M+/btadu2Lb///nvR+0LgptW33HILrVu3pk2bNrz99tt7rMeu3nvvPfr371/0urScM2bMoHPnzkVzFixYQNu2bXc7f9fPf//999O4cWO2b98OwMaNG2nUqFHR6x0GDx7MsGHD6NGjB82bN+eTTwL3bdu6dSuXXnopbdq0oUOHDowdO7boM5588skA3HXXXVx22WVkZmbSpEmTor1Ct956K3PnzqV9+/bccsstAJx++um88cYbpdZzn7h7VD06derkZcnYsWMjHUFCpFpFD9UqeqhW0eFg1Sk/v8AXL1juo6cs9Q9+zvZx1/zVvzjrCh/b/RSfk97YV1Sp6Y/3u8Ib/WmUn3HRI+6BdmenxxNdz/NO937uf7pthL/X/xKfcPWtPvO5f/nGFasPSsZoMXHixEhH8IoVK7p74O9HamqqL1682AsKCrxLly7+7bffurt7w4YNfdWqVe7uvmrVKu/Ro4fn5OS4u/uDDz7od999d9G8hx56qGjbvXr18iFDhri7+9dff+1HHXWUu7tfe+21ftddd7m7+5dffunt2rVzd/dXXnnFr7nmGnd3X7t2rRcWFrq7+4svvug33XSTu7vfeeed/sgjjxS9x8CBA4tyLly40Fu2bOnu7tddd11Rrk8++cSBos9QHOCvv/66u7vffffdRe/fq1cvv+qqq9zdPS8vz7t27eorV650d/eRI0f6pZde6u7unTt39vfff9/d3bds2eKbN2/2+fPn7/RZd2x/27Ztnpubu9PX/b333vPjjz/e8/Pzffny5V6/fn1funTpHutR3KBBg3zUqFF7zdmuXTufO3duUc3uvffePc4v/vnd3QcPHuwffPCBu7s///zzRfUo7pJLLvF+/fp5QUGBz5492+vWretbtmzxRx991AcPHuzu7jNmzPD69ev7li1bfOzYsX7SSScV1bVr166+detWX7VqlVetWtXz8vJ2+lrukJ2d7a1bty7x/hMnTvQRI0b4iy++WLy+E303/YQOgRMREZGQxMbGUK9hOvV2LHj6/hJzbgg+CgtPZv3fL2PjrLmsnTWXZTPmkzLjNxZ1P4Nm1SuR/9GvnPXpq/9b8SpYnVKVM/70BhnNa3HC8mlUXLuKpEYNqNu2BXU7HHl4H2qXmVly2bnnwtVXQ24unHhiyfHBgwOP1avh7LN3HsvK2qe379y5M/XqBSrbvn17FixYQPfu3Xea8+OPPzJ9+vSiw4/y8vLo2rVr0fh555230/yBAwcC0LNnTzZu3Mj69ev57rvv+Pe//w1Anz59WLNmDRs2bNhpvezsbM477zyWLVtGXl4ejRs3LjXzF198wfTp04teb9y4kU2bNvHNN9/w/vvvA3DSSSeRlpZW6voxMTFFmS+66CLOPPPMEp9l1qxZTJ06lRNOCBx4WlBQQO3atdm0aRNLlizhjDMCh6AmJSWV2H7Xrl25//77yc7O5swzz6RZs2Y7jX/33XcMHDiQ2NhY0tPT6dWrFxMmTCA1NTWkeixbtowaNWrsMSfAueeeyzvvvMOtt97K22+/zdtvv73H+cU/P8CQIUN4+OGHOf3003nllVd48cWiW1jv5NxzzyUmJoZmzZrRpEkTZs6cyXfffcd1110HQMuWLWnYsCGzZ88use5JJ51EYmIiiYmJ1KxZkxUrVpT6HjVr1mTp0qWlju0LNUAiIiJy0MXEGFVqVqVKzao06HE07YPLe+yYMLQLm9beydopM1j79Ti2Tp1BwaLFpKWl8MWMFfR/82l6/f5j0faWp1Tjp16nMPq8a2hbrwrt5v5Cw5aNqdG8EYk1qmFmh/gTHl4SExOLnsfGxpKfn19ijrtzwgkn8NZbb5W6jYoVd77K4K41MbNSD+Xadd51113HTTfdxKmnnkpWVhZ33XVXqe9XWFjIDz/8QIUKFfa6zVAUX2fHZ3F3jjrqKH744Yed5m7cuHGv27vgggs45phj+M9//kO/fv146aWX6NOnT9F4aV+LHUKpR4UKFdi6desec0KgmTnnnHM488wzMTOaNWvGb7/9ttv5sHMtjz32WBYsWMDXX39NQUFB0SF+uwq13qUJ5fNC4JC60uq9rw7jX6WIiIhIWVapamUaZnahw5030fXdF+n+06eMurY7k+/oS/vP/83vX4xjwt+e5adzr2BL5SqsWbGWsTNX8djoaXS66mLq9ziapPQaFMTFsaRmfb48dTCvZP3O7Pkr2b4tL9Ifb99kZZV8BE/+Jjm59PHBgwPj1auXHDtIKlWqxKZNmwDo0qUL33//PXPmzAEgNze31N/m77DjnJbvvvuOypUrU7lyZXr27Fl0DkdWVhbVq1cnNTV1p/U2bNhA3bp1AXj11f/tJSyeBaBv3747XTVu8uTJADu9x5gxY1i3bl2p+QoLC4suWPDmm2+W2MMC0KJFC1atWlXUKGzfvp1p06aRmppKvXr1+PDDDwHYtm0bubm5O607b948mjRpwvXXX8+pp55a4nynnj178vbbb1NQUMCqVav45ptvdjpfZ2+OPPLIolrsLidA06ZNiY2N5d577y3as7On+aUZNGgQAwcO5NJLL93tnHfffZfCwkLmzp3LvHnzaNGixU61mD17NosWLaJFixYhfb5d671jG7trwPaF9gCJiIhImRIbY6Q3qkN6ozpw3P8OsWoMXAZsytnCr+3eZdP02aQsX0LaL+PZtGEzz6S25ucxszh76pc8NOZJVlaqyvq0GuRUrsq2Bo2ZNHAo1Y9oSIvVC6nXqBbVWzY9vA+rOwiGDh3KgAEDqF27NmPHjmXEiBEMHDiQbdu2AXDffffRvHnzUtdNS0ujW7dubNy4kZdffhkInPB+6aWX0rZtW5KTk3dqcHa46667OOecc6hbty5dunRh/vz5AJxyyimcffbZfPTRRzz11FM8+eSTXHPNNbRt25b8/Hx69uzJ8OHDufPOOxk4cCAdO3akV69eNGjQoMR7QGAvx7Rp0+jUqROVK1cuatiKS0hI4L333uP6669nw4YN5Ofnc+ONN3LUUUfx2muvceWVV3LHHXcQHx/Pu+++S0yxv09vv/02r7/+OvHx8dSqVYs77rhjp22fccYZ/PDDD7Rr1w4z4+GHH6ZWrVrMnDkzhMoEDht7/vnnGTJkyB5zQmAv0C233FL0tdzb/F1deOGF3HbbbUWHNZamRYsW9OrVixUrVjB8+HCSkpK4+uqrGTZsGG3atCEuLo4RI0bstLdnT6pVq8axxx5L69atGTBgAI888ghjx47lpJNOCmn9PbFQd02VFRkZGT5x4sRIxyiy40odUvapVtFDtYoeqlV0KC91Wp+bx4/z1rBm/GSajHmfmBXLqLZwLrWXL6Ri3hZa/eFdtsfG8ff/PMapM75ha1wCm5MqkhufxPcderPiL3fRPD2FTikF1Gxcb+9veAAmTZpEp06dwvoekZSZmcmjjz6610sdR1Io90Eq67p3784nn3xClSpVwvo+7733Hh999BGvvfZaqeODBw/m5JNP5uxdz0U7yHr27MlHH31U4ryuSZMmMXXqVLZv386QIUMAMLNJ7l7qX0DtARIREZHDQpXkBPq3rg2ta8NlA3YaK8wv4MvNeSzbsJW8drF8+m134rIXkz5nGpazibTs+dz6xWxiCguY+vi5rI1PYl7LDmzocDS1unemZp9jqVI3nfhY7TGSsuPvf/87ixYtCmsDdN111zFmzBhGjx4dtvcIxapVq7jpppt2e1GLfaEGSERERA57MXGx1K5cgdqVK8BFpwQexbQGvl2by8+L1jE5+1KSFsyj0ZQJVP85C/4Jf+l3LW+17895LKf3z19QMTmRen/9I7XbNCepQmiH9JQnWQfxPKRwifa9PwDHHHNM2N/jqaee2uucESNGhD1HjRo1OP300w/KttQAiYiIiAD1qyZTv2oyvPkcAF5YyPKps1ny9U+0rNaAEz2V/Pe+of9/RwZW+OhVCjHGHdGJX4a/Qe8ja3Fk7Uq6Ip1IGacGSERERKQUFhNDrbYtqdW2JZ2ASwAu7AQ8xJyvfmDZR59S7eP3qbB5E498PptHP5/FN89fwfbUKixtdzQJLVuQdko/6nc8iqSU5BLbLyws3OmkeRHZd4WFhfu8jhogERERkX10RJ+uHNGnKzxxN4WFzujlm5gxZxmrxnUidf5sun/yOnwCPHo7r3U4kXsHXMvxlfM5a+F40ju3J7F9M5YvX06tWrXUBInsp8LCQpYvX8727dv3aT01QCIiIiIHICbGaFUnlVZ1UuHb/wCwddNm5vz3e9b9/Bvz01tSL68CG8d/x3Hv3AfPQl7Nmvz88sssTU/XIXMiB2D79u0sWrSIvLy8kG+SqgZIRERE5CBLqlSR1mf2hTP70gO4A9i+vTuLrjqetVNns+bX6dS84irG1kjjw1a9yItN4NiFk0nbupHtcQlsrZBCbFIC22vXJbVFU5LiYyP9kWQXO/bgSeSZGWbG8ccfH9J8NUAiIiIih0B8fBwNMrvQILNLcMmDpMyZT+rMbBav20yNsQWkzJ1F0qaNNJk7mfjCfH6u3YKHkztSOSmWq75/G+vYkcanHk9q7ZoR/SwCkydPpn379pGOIQQaoBo1apCenh7S/LA2QGbWH3gCiAVecvcHdxlPA14GmgJbgcvcfWo4M4mIiIiUFY2PaEzjIxoHXlzcf6eb1m7J2cKKH2ZwZUwlJv22gPN/+pL4Hz6HZx5kWUo1fq/bjM/OuZK6x3Xngs4NqJKcELkPUg6tWbOGNm3aRDqG7IewNUBmFgs8A5wAZAMTzGyUu08vNu2vwGR3P8PMWgbnHxeuTCIiIiLRokJKBY4/oSPHAxzXjMJrtzH9rY+YN+Zr0qZMIn1lNr8tWscbY2Yy/ekRXP7jv9l24slUPetUmh3fDYvVYXMipQnnHqDOwBx3nwdgZiOB04DiDVAr4G8A7j7TzBqZWbq7rwhjLhEREZGoExMXS6uLz6TVxWcWLXu/oJDx81az5bwn6LBkJrw4E158lHVJlZjY/1zW/OUOTu9QV+cQiRRj7h6eDZudDfR39yHB1xcDx7j7tcXmPAAkuftNZtYZGBecM2mXbQ0FhgKkp6d3GjlyZFgy74+cnBxSUlIiHUNCoFpFD9UqeqhW0UF1ih4HUit3Z+PMhWz84Tea/zyO2QlVub3v1SRu38brnz/Cgl7HkdavG3GVSt6TSPadvq/Ktt69e09y94zSxsK5B6i0azru2m09CDxhZpOB34BfgPwSK7m/ALwAkJGR4TuOjS0Lih+rK2WbahU9VKvooVpFB9UpehxwrXoDVwWediwoJOmXJax96VWOnvojR0/9kdwXEllapwmb+p1ItZuuo0GLhgcjdrmk76voFc4GKBuoX+x1PWBp8QnuvhG4FMACF8GfH3yIiIiIyAGIj43hnIz6kHEbuY/8gbHPvk3j156n1bTxrHp9Gf0SO1JYfRZDYpbRceVcWt88jNQGdSIdWyTswnnr4QlAMzNrbGYJwPnAqOITzKxKcAxgCPBNsCkSERERkYMkuVJFTvrzZbSa+hMFedtZ+9PPXHxKBiek5HHt7YPp9tS9pDasy+Q6LRh95f+xat7iSEcWCZuw7QFy93wzuxb4jMBlsF9292lmNiw4Phw4EviXmRUQuDjC5eHKIyIiIiIQGx9Hi9aNadEaOKE5W6/IYdEHn/LbGx/RccKXtH/hAe6au5q5Ay+jf3WjX8vqVG/RJNKxRQ6asN4HyN1HA6N3WTa82PMfgGbhzCAiIiIiu5dUqSLNB51F80Fn4YWF/PzPd6lAVb79fTXnPvoQ1Wd+y8KaDVjS+EhSr7uKZmcPIDFR9xyS6BXWBkhEREREoofFxNDxivPoCNxc6EzvnMCPL7xEvZ++odPEr0i86DNG39ONJ6/6G50bV+XUwhVknKFbOEp0UQMkIiIiIiXExhhtzjwBzjwBgHXLVvP90yPYVBhHXn4h33/yHfe8dBXbY+OY3utECntlUueic0hvUi/CyUX2TA2QiIiIiOxVWu3q9Ln/ZgDOAzasaMt30z8mY/wXtPtqFHw1Cu68iYyb36VT20YMzltAs8zOVG/aILLBRXahBkhERERE9lnl9Kp0//4/AKxbspKp7/yH9eMmkFitKuOnL+H5B88B4Oc23Ui49hpaD70gknFFiqgBEhEREZEDkla3Jj3+cCn84VJOATZvy2dy1WfJ/9frZPw2Dq4cx9y772TqP9+mT2YbKiXFRzqylGPhvA+QiIiIiJRDFRPjaH/LVWT89j0bFy7hh/7nU3HTOm7+chFt7vqcZ17LYlvulkjHlHJKDZCIiIiIhE1qgzp0HfMW1dcs56GBGRy38Gcuu6w/29OqMX7A+fz+7zGRjijljBogEREREQm7uPg4zuxYj+de+TNf3/EY69Jq0vnTt2l29olkV6/LkyO+ZNWmbZGOKeWAGiAREREROWQSkhLof/tV1F++gFW/LyDr8ptZVqk6j83I5ej7v+DT10aDe6RjymFMDZCIiIiIRESNIxqS+dIjHD3/V569KIM+a3+n/6CTWFyrEYu+Hh/peHKYUgMkIiIiIhE3oE1tnnn0Cr645A/UX7mIBpnH8GubbqyZsyDS0eQwowZIRERERMqECqkVOX7EP1gxeTrZabVoN/UHlnU/npve/oUFqzdHOp4cJnQfIBEREREpU9LbHYmvWcrYF97j40W5vP/LUqb890duT8jmmEduI6lihUhHlCimBkhEREREyhwzo/eV59AbGPz7cip1v4bGKxfCcw8woUEb/JJL6HzPHyMdU6KQDoETERERkTKtbbNaNFo2j7H3PM3nPc/g6EW/0fnem/n7OX/k+zmrIx1PoowaIBEREREp8ywmht63X0Pfr99nffYKRvU5j4/TWnDhSz/x/JX3MuebiZGOKFFCDZCIiIiIRJUqdWty6pcjeefRQZxfN5aLX7mfI3odzc/te7JmwZJIx5MyTg2QiIiIiESlmqlJPHhdf1aM/5Vvjz6Bjr9+y7Z2HVg+Y26ko0kZpgZIRERERKJa4/Yt6DH+c757/m3qbFzFlNMu4s0fFkQ6lpRRugqciIiIiBwWug89l++2bOXZSauZ/NE05r/9EVdefQrVmzeJdDQpQ7QHSEREREQOG91vGMTLz11L6+pJDH75fqq3aMpPw9+KdCwpQ9QAiYiIiMhhpWrFBD65+TjW3P0AAMdcdQHfD7yKDct0yWxRAyQiIiIih6m2fxjCwnGTWFitHseOHM6mlq34YtrySMeSCAvrOUBm1h94AogFXnL3B3cZrwy8DjQIZnnU3V8JZyYRERERKT8adu0Iqxfz1Yvv8fqPC/nqtUlcXmEt53drQrMTjo10PImAsO0BMrNY4BlgANAKGGhmrXaZdg0w3d3bAZnA380sIVyZRERERKR86nPF2TzxzPU0qpZMxpvDada3O1MaHMWKb36IdDQ5xMJ5CFxnYI67z3P3PGAkcNoucxyoZGYGpABrgfwwZhIRERGRcqpSUjxZt/Sm6esv8kOPk2m7eDqcfApLp86JdDQ5hMzdw7Nhs7OB/u4+JPj6YuAYd7+22JxKwCigJVAJOM/d/1PKtoYCQwHS09M7jRw5MiyZ90dOTg4pKSmRjiEhUK2ih2oVPVSr6KA6RQ/V6tBa/98fOP2Bv/Jb7WZMeeZp6qaFfiCSalW29e7de5K7Z5Q2Fs5zgKyUZbt2W/2AyUAfoCnwXzP71t037rSS+wvACwAZGRmemZl50MPur6ysLMpSHtk91Sp6qFbRQ7WKDqpT9FCtDrHMTH7cuJkffpnPCz9u4e1rj6Zt/bSQVlWtolc4D4HLBuoXe10PWLrLnEuB9z1gDjCfwN4gEREREZGw6/LUfXR59UmqbFrHxPOvJHfFqkhHkjALZwM0AWhmZo2DFzY4n8DhbsUtAo4DMLN0oAUwL4yZRERERER20rVpNR5IXcll495lygVX4oWFkY4kYRS2Bsjd84Frgc+AGcA77j7NzIaZ2bDgtHuBbmb2G/Al8Gd31x2qREREROSQ6v3gnxh3TD+6fPUB05p3ZOVM/U7+cBXW+wC5+2hg9C7Lhhd7vhToG84MIiIiIiKhOOa7//DNKRfR89ORTD75LGr8/jOBixXL4SSch8CJiIiIiESN2LhYeo55iynNO/Jho84MeOJbstflRjqWHGRqgEREREREijlq6k9knzmQxQtX8MrNj7Fx8bJIR5KDSA2QiIiIiEgxsfFxvHR1Ji/6VG5/+TYqNqzH9/c9HelYcpCoARIRERERKUW3x+9m8svvsKxSdY69/Tom3P14pCPJQaAGSERERERkN9pfeg4VZ0wFoP7f72PSD1MjnEgOlBogEREREZE9SKtTg7kff8nKimlcPHoxefm6T1A0UwMkIiIiIrIXTU/uwyfP/5vc7YU8cPUjFOTo6nDRKqz3ARIREREROVz89dS2VJn3O1c/dCtTvmwDJ58Y6UiyH7QHSEREREQkRFffeBa/129B23m/Me7p1yMdR/aDGiARERERkX2Q9PkY1lZIpdt1F/PTaRezPb8g0pFkH6gBEhERERHZB/VbNub7h//Btth4cqfOoMXtn7I+Ny/SsSREaoBERERERPZRpdZNid+2ldf/8BCFDq+99mWkI0mI1ACJiIiIiOyHmNgY/nntcZyRM4+h15zGlycPinQkCYEaIBERERGRA/DgQ5ezNrUax/3nNcYddyZeqPsElWVqgEREREREDkBixWTSsuczs04zun31AT/d80SkI8keqAESERERETlASclJNJnzGxuTUmjx8J0sX7Ak0pFkN9QAiYiIiIgcBAkVEpn//hiOv/w5rvxkPu4e6UhSCjVAIiIiIiIHSbsB3cno3IJFsxfx1d9fiXQcKUVcpAOIiIiIiBxOnji/A9P/MoiOS2Yws1l9Wp52QqQjSTHaAyQiIiIichAlxcdS8+1/AZD/l/9j2+bcCCeS4tQAiYiIiIgcZPWOzWB8r1NpPWMCv1x5S6TjSDFqgEREREREwiDjyw9YUakaHUa+yOLZCyMdR4J0DpCIiIiISBjExMaw5q33OHnMYhp8uYQ3GtcjKT420rHKvbDuATKz/mY2y8zmmNmtpYzfYmaTg4+pZlZgZlXDmUlERERE5FBpdVImA07owNQ5y/n29EsjHUcIYwNkZrHAM8AAoBUw0MxaFZ/j7o+4e3t3bw/8Bfja3deGK5OIiIiIyKF2z2mteWTl9xw3+nWyf5ke6TjlXjj3AHUG5rj7PHfPA0YCp+1h/kDgrTDmERERERGJiIxrBxGDs27goEhHKffC2QDVBRYXe50dXFaCmSUD/YF/hzGPiIiIiEhE1DmmHdlV69Bm1iTmff5tpOOUa+bu4dmw2TlAP3cfEnx9MdDZ3a8rZe55wEXufsputjUUGAqQnp7eaeTIkWHJvD9ycnJISUmJdAwJgWoVPVSr6KFaRQfVKXqoVtFjf2q1afYiTrnyEuZUb8Dvr/+TCom6Hlm49O7de5K7Z5Q2Fs6vejZQv9jresDS3cw9nz0c/ubuLwAvAGRkZHhmZuZBinjgsrKyKEt5ZPdUq+ihWkUP1So6qE7RQ7WKHvtVq0z4Knstr09fx9rfk3htyDFUSooPRzzZg3AeAjcBaGZmjc0sgUCTM2rXSWZWGegFfBTGLCIiIiIiEdfnnhtpPPg8JmdvYOQ/3ox0nHIpbHuA3D3fzK4FPgNigZfdfZqZDQuODw9OPQP43N03hyuLiIiIiEhZcfvJrci4+yYGTPyMKdXiaHvlhZGOVK6E9T5A7j7a3Zu7e1N3vz+4bHix5gd3H+Hu54czh4iIiIhIWXLkP+4DoNm1l5OXuzXCacqXsDZAIiIiIiJSUqMeGYz/0/1UyN/GpCv+GOk45YoaIBERERGRCOh0358AOPqt4axdr7NBDpWQzwEys7pAw+LruPs34QglIiIiInK4i42P44e7HuPBuYXwxmQ+vLobZhbpWIe9kBogM3sIOA+YDhQEFzugBkhEREREZD91vfNGjnjnVz4ZP4+Xv5jB5Se0inSkw16oe4BOB1q4+7YwZhERERERKXcePrstNw7tR83H17Bu3iLSGtSOdKTDWqjnAM0DdJcmEREREZGDLDbGWHHTrSQW5LO8T/9IxznshdoA5QKTzex5M3tyxyOcwUREREREyotOt13P9phYjpw7hfF/fzHScQ5roR4CNyr4EBERERGRg8xiYtiycjXx1dNIeOZpNl19CZUqJEQ61mEppAbI3V81swSgeXDRLHffHr5YIiIiIiLlS2q1Knx31+O8O3ExtcbO4S8n6oII4RDSIXBmlgn8DjwDPAvMNrOe4YslIiIiIlL+dL/zBv7Tpg8jvpzJ2vnZkY5zWAr1ELi/A33dfRaAmTUH3gI6hSuYiIiIiEh59PIlGSQd35tFH8YT/82nVKpbK9KRDiuhXgQhfkfzA+Dus9FV4UREREREDrqeLWqypc/xtJ/3K9tatiJ/W16kIx1WQm2AJprZP80sM/h4EZgUzmAiIiIiIuVVr1efYEaTNlTPWcdPV9wc6TiHlVAboKuAacD1wA3AdGBYuEKJiIiIiJR3zWf9AkDrd14mb5uuP3awhHoVuG3AP4IPEREREREJs9i4WMZedycvr4hlzfAfGX1Dj0hHOizscQ+Qmb0T/PM3M5uy6+PQRBQRERERKZ96P3kX3zbuyIyl65k0fkak4xwW9rYH6IbgnyeHO4iIiIiIiJT0xY09qNisCevfrQ3zfot0nKi3xz1A7r4s+HQ1sNjdFwKJQDtgaZiziYiIiIiUe0fUSmVRx24cOX8q4+/UGSkHKtSLIHwDJJlZXeBL4FJgRLhCiYiIiIjI/7Qe9SYAbR/4C7nrNkQ4TXQLtQEyd88FzgSecvczgFbhiyUiIiIiIjtUrJzCuME3kJSfx7grb410nKgWcgNkZl2BC4H/BJeFdAU5ERERERE5cN1eeZw32/VjzOYKjJ21MtJxolaoDdCNwF+AD9x9mpk1AcaGLZWIiIiIiJTQ6dN3+Xeb47jmjZ8jHSVqhXofoK+Br4u9nkfgpqgiIiIiInKItKhVibO3LaTPB2+w5LT61O10VKQjRZ293Qfo8eCfH5vZqF0fe9u4mfU3s1lmNsfMSj1Y0cwyzWyymU0zs69LmyMiIiIiIgHD+rfmxNnjWHvpFeAe6ThRZ297gF4L/vnovm7YzGKBZ4ATgGxggpmNcvfpxeZUAZ4F+rv7IjOrua/vIyIiIiJSnhzRryfTO/WkzaRvGH/3Y3S+66ZIR4oqe7sP0KTg04nAt+7+dfBwuO+ACXvZdmdgjrvPc/c8YCRw2i5zLgDed/dFwffT2VwiIiIiInvR5MtPAKjw7tsRThJ9Qr0IwpdAcrHXFYAv9rJOXWBxsdfZwWXFNQfSzCzLzCaZ2aAQ84iIiIiIlFtJlSvxw+mDaTN9PN9+MWnvK0iRUC9lneTuOTteuHuOmSXvaQXASlm260GKcUAn4DgCTdUPZvaju8/eaUNmQ4GhAOnp6WRlZYUYO/xycnLKVB7ZPdUqeqhW0UO1ig6qU/RQraJHWajVpvNO5bjqx5A/fiV3x0U2SzQJtQHabGYd3f1nADPrBGzZyzrZQP1ir+sBS0uZs9rdNwff4xugHbBTA+TuLwAvAGRkZHhmZmaIscMvKyuLspRHdk+1ih6qVfRQraKD6hQ9VKvoUVZqdd+8L2DJEmz8Nnr9aWik40SFUBugG4F3zWxHA1MbOG8v60wAmplZY2AJcD6Bc36K+wh42szigATgGOCxEDOJiIiIiJRrWX/MpEJSPGuSK+M3D8FiQj3DpfwK6Svk7hOAlsBVwNXAkcUukLC7dfKBa4HPgBnAO8GbqA4zs2HBOTOAT4EpwHjgJXefur8fRkRERESkPKmQGMdPl95AtdwNTHn135GOExVCaoCC5/v8GbjB3X8DGpnZyXtbz91Hu3tzd2/q7vcHlw139+HF5jzi7q3cvbW7P75/H0NEREREpHw68p6/AGAPPxLhJNEh1H1krwB5QNfg62zgvrAkEhERERGRkKXWS2fKUcfQduYEtqzfFOk4ZV6oDVBTd38Y2A7g7lso/SpvIiIiIiJyiOUOvpwPW/Xik/e/iXSUMi/UiyDkmVkFgpexNrOmwLawpRIRERERkZAd88chtFxenbxZhRy3OY+qFRMiHanMCnUP0J0ELlZQ38zeIHBj1D+FLZWIiIiIiITMzLjjtNZc+dP7/HjjnZGOU6btdQ+QmcUAacCZQBcCh77d4O6rw5xNRERERERCdMGxTbGvR7DhpxQ23nEDqfVrRTpSmbTXPUDuXghc6+5r3P0/7v6Jmh8RERERkbLFYmKY+MjzVN6aw+Qrb4p0nDIr1EPg/mtmN5tZfTOruuMR1mQiIiIiIrJPMm4eypa4RLp9+jYF2/MjHadMCrUBuozADVC/BiYWe4iIiIiISBky8cKriPNCvvrb8L1PLodCvQpcKwINUHcCV4L7FtBXVERERESkjDn2pUe5ckMyK+KbcEKkw5RBoe4BehU4EngSeCr4/NVwhRIRERERkf0TExdLwelnMHmDs3RdbqTjlDmh7gFq4e7tir0ea2a/hiOQiIiIiIgcmJt7NqDv3//C7FVjqfPPRyMdp0wJdQ/QL2bWZccLMzsG+D48kURERERE5EC0bFyTs3/7koaj3490lDIn1AboGGCcmS0wswXAD0AvM/vNzKaELZ2IiIiIiOyXX9t2o/Hy+WxZvynSUcqUUA+B6x/WFCIiIiIiclBtad8RpnzPjFfeoeMfLo90nDIjpD1A7r5wT49whxQRERERkX3T5pE7ASh8860IJylbQj0ETkREREREokilmtUY07wbT3c9B3ePdJwyQw2QiIiIiMhh6vdnR5CVXI8p2RsiHaXMUAMkIiIiInKYOqN9Hc6Z8jkz7nwo0lHKDDVAIiIiIiKHqfrVKvLImCdp983oSEcpM9QAiYiIiIgcxsb3OpUj506hYHt+pKOUCWqAREREREQOYyvadALgh1vui3CSskENkIiIiIjIYazvgzcDkPqZDoOD0G+EKiIiIiIiUSixYjKfHH0iU5Nr0jbSYcqAsO4BMrP+ZjbLzOaY2a2ljGea2QYzmxx83BHOPCIiIiIi5dHTF/yZ4V3OZt3mvEhHibiwNUBmFgs8AwwAWgEDzaxVKVO/dff2wcc94cojIiIiIlJe/d9JR9J47RI+u+7OSEeJuHAeAtcZmOPu8wDMbCRwGjA9jO8pIiIiIiK76NGsBmNfvBKADff9gcp1akY4UeSE8xC4usDiYq+zg8t21dXMfjWzMWZ2VBjziIiIiIiUW+MGXQfA9L89FeEkkWXuHp4Nm50D9HP3IcHXFwOd3f26YnNSgUJ3zzGzE4En3L1ZKdsaCgwFSE9P7zRy5MiwZN4fOTk5pKSkRDqGhEC1ih6qVfRQraKD6hQ9VKvoEY21KiwooM/xx/Nzk7Zs/OcTkY4TVr17957k7hmljYXzELhsoH6x1/WApcUnuPvGYs9Hm9mzZlbd3VfvMu8F4AWAjIwMz8zMDFvofZWVlUVZyiO7p1pFD9UqeqhW0UF1ih6qVfSI1lqNOfZUBnw/iuykytTr0iHScSIinIfATQCamVljM0sAzgdGFZ9gZrXMzILPOwfzrAljJhERERGRcqv+NZcztkkn3piTG+koERO2PUDunm9m1wKfAbHAy+4+zcyGBceHA2cDV5lZPrAFON/DdUyeiIiIiEg513rgqXSYnci6qeu5Pq+ACgmxkY50yIX1RqjuPhoYvcuy4cWePw08Hc4MIiIiIiLyP7d1q83sBx7j13cK6XLRKZGOc8iFtQESEREREZGypWeDFM7KGsHkldOhHDZA4TwHSEREREREypgaLZoAcOSsX/DCwginOfTUAImIiIiIlDPjBt9IYsF2Jjz3ZqSjHHJqgEREREREypk2D90GQOzw5yKc5NDTOUAiIiIiIuVMpZrVeKv72Xxb7Qg6uhO8M025oD1AIiIiIiLl0Iyb72R0y+7MWZkT6SiHlBogEREREZFy6MJjGtJ39g/8eu/jkY5ySKkBEhEREREph1rUqsTDnz5JxjsvRTrKIaUGSERERESknBqbeQaNVi3ijRc/jnSUQ0YNkIiIiIhIOVX5hmsBqPXs45ENcgipARIRERERKaf69DuadUmV6DhzQrm5KaoaIBERERGRcuzLQTcwuuWxrF2/OdJRDgndB0hEREREpBzbdtkQ/i+tC03X51GtaqTThJ/2AImIiIiIlGM9m9Wg5qY1zJwyN9JRDgk1QCIiIiIi5Vj9qsl89dIwWj39UKSjHBJqgEREREREyrkVNepR9+cfIh3jkFADJCIiIiJSzq1tdAR11y1j+dTfIx0l7NQAiYiIiIiUc5X//EcA5v/hLxFOEn5qgEREREREyrnmpxwHQMWF8yKcJPx0GWwREREREeHim0cwKakm0yMdJMy0B0hERERERGjVuzO5eQVkr8uNdJSwUgMkIiIiIiKckL+Cz/95DdnPvRLpKGGlBkhERERERDiic2uar1lEzPvvRzpKWIW1ATKz/mY2y8zmmNmte5h3tJkVmNnZ4cwjIiIiIiKlq1KrOmsrVqbzpK/Iy90a6ThhE7YGyMxigWeAAUArYKCZtdrNvIeAz8KVRURERERE9m7miecAMPnBZyKcJHzCuQeoMzDH3ee5ex4wEjitlHnXAf8GVoYxi4iIiIiI7MVRj90HQNK/341wkvAJ52Ww6wKLi73OBo4pPsHM6gJnAH2Ao8OYRURERERE9qJy3XSuP/s2ptdrweeFTkyMRTrSQRfOBqi0r5bv8vpx4M/uXmC2+y+umQ0FhgKkp6eTlZV1kCIeuJycnDKVR3ZPtYoeqlX0UK2ig+oUPVSr6HE416rCiT2ZMyuPlz/6iiPSYiMd56ALZwOUDdQv9roesHSXORnAyGDzUx040czy3f3D4pPc/QXgBYCMjAzPzMwMU+R9l5WVRVnKI7unWkUP1Sp6qFbRQXWKHqpV9Dica9W64xY6dzuJeb82J3PM8EjHOejC2QBNAJqZWWNgCXA+cEHxCe7eeMdzMxsBfLJr8yMiIiIiIodO9dQKtFi9iKOXzAAOvwYobBdBcPd84FoCV3ebAbzj7tPMbJiZDQvX+4qIiIiIyIHJTatOg/XL2b7l8LscdljvA+Tuo929ubs3dff7g8uGu3uJVtLdB7v7e+HMIyIiIiIie7epZ28Apr40MsJJDr6wNkAiIiIiIhJ9Mu68CYDCEa9GOMnBpwZIRERERER2UrlODT46shefte4V6SgHnRogEREREREpYcSQO3ipztHk5RdGOspBpQZIRERERERKGHRsI87/ZQz/ffL1SEc5qNQAiYiIiIhICad2bMADnz1D7TdfiXSUg0oNkIiIiIiIlBAbH8eWuETiczdHOspBpQZIRERERERKNfuoDKovXYi7RzrKQaMGSERERERESpXbqTO1N6xk+tufRDrKQaMGSEREREREStXh8XuZmt6UGR9/FekoB01cpAOIiIiIiEjZlFSpIrf++SWmrtjMWe6YWaQjHTDtARIRERERkd1Kr5xEWu4GFi1bF+koB4UaIBERERER2a0rWcIvT13Iug/+E+koB4UaIBERERER2a1aRzYGIHdsVmSDHCRqgEREREREZLcadO3IxsSK1M/6LNJRDgo1QCIiIiIiskebkyqSvm5FpGMcFGqARERERERkjxZ260NCYT6rl6yKdJQDpstgi4iIiIjIHlUYdgVnVu1Er+nruKFujUjHOSDaAyQiIiIiInvU9pTe/FzvSD78/Be8sDDScQ6IGiAREREREdkjM+NPzRN459lhzBnzdaTjHBA1QCIiIiIisleZyVupkbueVaPGRDrKAVEDJCIiIiIie9XkxN4AxC5YENkgB0gNkIiIiIiI7FVSagpz6jXniO8+j3SUA6IGSEREREREQrIhvS7VcjdE9YUQwtoAmVl/M5tlZnPM7NZSxk8zsylmNtnMJppZ93DmERERERGR/be87ync2/tyZi1cHeko+y1s9wEys1jgGeAEIBuYYGaj3H16sWlfAqPc3c2sLfAO0DJcmUREREREZP9VvvRi/lnwI6mz19Cycc1Ix9kv4dwD1BmY4+7z3D0PGAmcVnyCu+e4uwdfVgQcEREREREpk449ohqZ8yZS4flnIx1lv4VtDxBQF1hc7HU2cMyuk8zsDOBvQE3gpDDmERERERGRA2Bm3DrxPVoumEZh/uPExMVGOtI+s//tgDnIGzY7B+jn7kOCry8GOrv7dbuZ3xO4w92PL2VsKDAUID09vdPIkSPDknl/5OTkkJKSEukYEgLVKnqoVtFDtYoOqlP0UK2iR3muVe7jIzjxo1f58Nb7qNLv2EjHKVXv3r0nuXtGaWPhbIC6Ane5e7/g678AuPvf9rDOfOBod9/tWVUZGRk+ceLEgx13v2VlZZGZmRnpGBIC1Sp6qFbRQ7WKDqpT9FCtokd5rtXGFatJrVWDnzv0pOPPX0c6TqnMbLcNUDjPAZoANDOzxmaWAJwPjNol2BFmZsHnHYEEYE0YM4mIiIiIyAFITa8OQP3ZU6LycthhOwfI3fPN7FrgMyAWeNndp5nZsOD4cOAsYJCZbQe2AOd5uHZJiYiIiIjIQfHmacOoMmMKtacuokPbRpGOs0/CeREE3H00MHqXZcOLPX8IeCicGURERERE5OBq9+yDnPTkd/xh5XY6RDrMPgrrjVBFREREROTw06p2Kgn524mb/HOko+yzsO4BEhERERGRw4+ZMWH4YNan1YCbz4t0nH2iPUAiIiIiIrLPFjVrQ+2lCyIdY5+pARIRERERkX22oXZ9EgrzyV23IdJR9okaIBERERER2WcF7dsDsOLzsnkvoN1RAyQiIiIiIvus0Qk9AJg7bV6Ek+wbNUAiIiIiIrLP6vbozL/bHMdnbXpHOso+UQMkIiIiIiL7LC4ulqcH3ca/f86OdJR9ogZIRERERET2S6ufv+GRj//O1k2bIx0lZGqARERERERkv1xcyzlr2liW/BA9N0RVAyQiIiIiIvslsUkjAFaN/iKyQfaBGiAREREREdkvR15yNgAxkydHNsg+UAMkIiIiIiL7JalSRfJi4qgxc0qko4QsLtIBREREREQken3VNpOV8RVpHOkgIVIDJCIiIiIi+23mbQ/w5E/L6L9pKzUrJUU6zl7pEDgREREREdlvXdo3psbmdUyaMCvSUUKiBkhERERERPbbMY2rMu65y6hx258jHSUkOgRORERERET2m8XE8HLGqWzt05+MSIcJgRogERERERE5IN0/epXKFeIjHSMkaoBEREREROSAHFk7NdIRQqZzgEREREREpNxQAyQiIiIiIuWGGiARERERESk3wtoAmVl/M5tlZnPM7NZSxi80synBxzgzaxfOPCIiIiIiUr6FrQEys1jgGWAA0AoYaGatdpk2H+jl7m2Be4EXwpVHREREREQknHuAOgNz3H2eu+cBI4HTik9w93Huvi748kegXhjziIiIiIhIORfOBqgusLjY6+zgst25HBgTxjwiIiIiIlLOmbuHZ8Nm5wD93H1I8PXFQGd3v66Uub2BZ4Hu7r6mlPGhwFCA9PT0TiNHjgxL5v2Rk5NDSkpKpGNICFSr6KFaRQ/VKjqoTtFDtYoeqlXZ1rt370nunlHaWDhvhJoN1C/2uh6wdNdJZtYWeAkYUFrzA+DuLxA8PygjI8MzMzMPetj9lZWVRVnKI7unWkUP1Sp6qFbRQXWKHqpV9FCtolc4D4GbADQzs8ZmlgCcD4wqPsHMGgDvAxe7++wwZhEREREREQnfIXAAZnYi8DgQC7zs7veb2TAAdx9uZi8BZwELg6vk725XVbFtrio2vyyoDqyOdAgJiWoVPVSr6KFaRQfVKXqoVtFDtSrbGrp7jdIGwtoAlQdmNnFvTZuUDapV9FCtoodqFR1Up+ihWkUP1Sp6hfVGqCIiIiIiImWJGiARERERESk31AAduBciHUBCplpFD9UqeqhW0UF1ih6qVfRQraKUzgESEREREZFyQ3uARERERESk3FADFAIz629ms8xsjpndWsq4mdmTwfEpZtYxEjklpFq1NLMfzGybmd0ciYwSEEKtLgx+P00xs3Fm1i4SOSWkWp0WrNNkM5toZt0jkVP2Xqti8442swIzO/tQ5pP/CeH7KtPMNgS/ryab2R2RyCmhfV8F6zXZzKaZ2deHOqPsGx0CtxdmFgvMBk4Asgnc4HWgu08vNudE4DrgROAY4Al3PyYCccu1EGtVE2gInA6sc/dHIxC13AuxVt2AGe6+zswGAHfp++rQC7FWKcBmd3czawu84+4tIxK4HAulVsXm/RfYSuAefe8d6qzlXYjfV5nAze5+ciQySkCItaoCjAP6u/siM6vp7isjkVdCoz1Ae9cZmOPu89w9DxgJnLbLnNOAf3nAj0AVM6t9qIPK3mvl7ivdfQKwPRIBpUgotRrn7uuCL38E6h3ijBIQSq1y/H+/TasI6DdrkRHK/1cQ+IXdvwH9gBY5odZKIi+UWl0AvO/uiyDws8Yhzij7SA3Q3tUFFhd7nR1ctq9zJPxUh+ixr7W6HBgT1kSyOyHVyszOMLOZwH+Ayw5RNtnZXmtlZnWBM4DhhzCXlBTqv4FdzexXMxtjZkcdmmiyi1Bq1RxIM7MsM5tkZoMOWTrZL3GRDhAFrJRlu/52M5Q5En6qQ/QIuVZm1ptAA6TzSiIjpFq5+wfAB2bWE7gXOD7cwaSEUGr1OPBndy8wK226HCKh1OpnoKG75wQPtf8QaBbuYFJCKLWKAzoBxwEVgB/M7Ed3nx3ucLJ/1ADtXTZQv9jresDS/Zgj4ac6RI+QahU8n+QlYIC7rzlE2WRn+/R95e7fmFlTM6vu7qvDnk6KC6VWGcDIYPNTHTjRzPLd/cNDklB22Gut3H1jseejzexZfV9FRKg/B652983AZjP7BmhH4NwhKYN0CNzeTQCamVljM0sAzgdG7TJnFDAoeDW4LsAGd192qINKSLWSsmGvtTKzBsD7wMX6LVpEhVKrIyz4E3XwKpgJgBrWQ2+vtXL3xu7eyN0bAe8BV6v5iYhQvq9qFfu+6kzgZzZ9Xx16ofxs8RHQw8zizCyZwAWxZhzinLIPtAdoL9w938yuBT4DYglcMWeamQ0Ljg8HRhO4AtwcIBe4NFJ5y7NQamVmtYCJQCpQaGY3Aq2K/6ZNwi/E76s7gGrAs8GfAfLdPSNSmcurEGt1FoFfAm0HtgDnFbsoghwiIdZKyoAQa3U2cJWZ5RP4vjpf31eHXii1cvcZZvYpMAUoBF5y96mRSy17o8tgi4iIiIhIuaFD4EREREREpNxQAyQiIiIiIuWGGiARERERESk31ACJiIiIiEi5oQZIRERERETKDTVAIiJy2DOzwWb2dPD5XWZ2c6QziYhIZKgBEhGRMit4g2n9XyUiIgeN/lMREZEyxcwamdkMM3sW+Bm43cwmmNkUM7u72LxBwWW/mtlrwWWnmNlPZvaLmX1hZumR+hwiIlI2xUU6gIiISClaAJcCHwJnA50BA0aZWU9gDfB/wLHuvtrMqgbX+w7o4u5uZkOAPwF/PNThRUSk7FIDJCIiZdFCd//RzB4F+gK/BJenAM2AdsB77r4awN3XBsfrAW+bWW0gAZh/aGOLiEhZp0PgRESkLNoc/NOAv7l7++DjCHf/Z3C5l7LeU8DT7t4GuBJIOjRxRUQkWqgBEhGRsuwz4DIzSwEws7pmVhP4EjjXzKoFl+84BK4ysCT4/JJDHVZERMo+HQInIiJllrt/bmZHAj+YGUAOcJG7TzOz+4GvzayAwCFyg4G7gHfNbAnwI9A4IsFFRKTMMvfSjiAQERERERE5/OgQOBERERERKTfUAImIiIiISLmhBkhERERERMoNNUAiIiIiIlJuqAESEREREZFyQw2QiIiIiIiUG2qARERERESk3FADJCIiIiIi5cb/A0+sTv8YYRXLAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1008x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "PlotPrecisionRecallCurve(results[0.5])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Adversarial Attack"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "detector = PyTorchFasterRCNN(model=model, clip_values=(0, 1), preprocessing=None)\n",
    "attack = ProjectedGradientDescent(detector, eps=0.01, eps_step=0.01, max_iter=10, verbose=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "attack_img_list, attack_bboxes = wider_read(limit_images=100, train=False)\n",
    "\n",
    "attack_dataset = WiderDataset(attack_img_list, attack_bboxes)\n",
    "\n",
    "attack_loader = DataLoader(attack_dataset,\n",
    "                         batch_size=1,\n",
    "                         shuffle=True,\n",
    "                         num_workers=0,\n",
    "                         collate_fn=collate_fn\n",
    "                         )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████| 100/100 [18:44<00:00, 11.24s/image]"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Time elapsed in Predicting: 18.74 minutes\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "start = time.time()\n",
    "\n",
    "adv_image_list = []\n",
    "prediction_info = []\n",
    "attack_info = []\n",
    "target_info = []\n",
    "\n",
    "\n",
    "with tqdm(attack_loader, unit=\"image\") as t_loader:\n",
    "    for images, targets in t_loader:\n",
    "\n",
    "        org_images = list(image.clone().to(device) for image in images)\n",
    "        attack_images = np.array([tonumpy(image.clone()) for image in images])\n",
    "        targets = [{k: v.to(device) for k, v in t.items()} for t in targets]\n",
    "\n",
    "        adv_image = attack.generate(x=attack_images, y=None)\n",
    "        adv_image_list.append(adv_image[0].copy())\n",
    "        adv_image = [totensor(adv_image[0].copy()).to(device)]\n",
    "    #     print('Image Generated')\n",
    "\n",
    "        model.eval()\n",
    "\n",
    "        with torch.no_grad():\n",
    "            org_predictions = model(org_images)\n",
    "            adv_predictions = model(adv_image)\n",
    "\n",
    "\n",
    "        org_predictions = [{k: v.to('cpu').detach() for k, v in t.items()} for t in org_predictions]\n",
    "        adv_predictions = [{k: v.to('cpu').detach() for k, v in t.items()} for t in adv_predictions]\n",
    "        targets = [{k: v.to('cpu').detach() for k, v in t.items()} for t in targets]\n",
    "\n",
    "        prediction_info.append(org_predictions)\n",
    "        attack_info.append(adv_predictions)\n",
    "        target_info.append(targets)\n",
    "\n",
    "        org_images = list(image.detach() for image in org_images)\n",
    "        adv_image = list(image.detach() for image in adv_image)\n",
    "        org_images.clear()\n",
    "        adv_image.clear()\n",
    "        torch.cuda.empty_cache()\n",
    "\n",
    "pickle.dump(attack_img_list, open('./data/Adver_images/adv_attack_images_1p_40i_100img.p', \"wb\" ))\n",
    "\n",
    "end = time.time()\n",
    "print(f\"Time elapsed in Predicting: {(end - start)/60:.2f} minutes\")\n",
    "# adv_image_list = pickle.load( open( \"./data/Adver_images/test.p\", \"rb\" ))\n",
    "prediction_info = list(itertools.chain(*prediction_info))\n",
    "attack_info = list(itertools.chain(*attack_info))\n",
    "target_info = list(itertools.chain(*target_info))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Processing : 100%|██████████| 100/100 [00:01<00:00, 66.39it/s]\n",
      "Processing : 100%|██████████| 100/100 [00:01<00:00, 58.77it/s]\n"
     ]
    }
   ],
   "source": [
    "r_original = evaluation(prediction_info, target_info, iou_thresh=0.5, interpolation_method='EveryPoint')\n",
    "r_adversarial_samples = evaluation(attack_info, target_info, iou_thresh=0.5, interpolation_method='EveryPoint')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "For the original samples : \n",
      "Average Precision: 0.51 Total Recall: 0.56\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "For the adversarial samples : \n",
      "Average Precision: 0.07 Total Recall: 0.19\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1008x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    }
   ],
   "source": [
    "prec = r_original['total TP']/(r_original['total TP'] + r_original['total FP'])\n",
    "rec = r_original['total TP']/r_original['total positives']\n",
    "print(f\"For the original samples : \\nAverage Precision: {r_original['AP']:.2f} Total Recall: {rec:.2f}\")\n",
    "PlotPrecisionRecallCurve(r_original)\n",
    "print()\n",
    "prec = r_adversarial_samples['total TP']/(r_adversarial_samples['total TP'] + r_adversarial_samples['total FP'])\n",
    "rec = r_adversarial_samples['total TP']/r_adversarial_samples['total positives']\n",
    "print(f\"For the adversarial samples : \\nAverage Precision: {r_adversarial_samples['AP']:.2f} Total Recall: {rec:.2f}\")\n",
    "PlotPrecisionRecallCurve(r_adversarial_samples)\n",
    "print()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "environment": {
   "name": "pytorch-gpu.1-8.m65",
   "type": "gcloud",
   "uri": "gcr.io/deeplearning-platform-release/pytorch-gpu.1-8:m65"
  },
  "kernelspec": {
   "display_name": "Python [conda env:visionad]",
   "language": "python",
   "name": "conda-env-visionad-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}