Donkeycar_Keras.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import sys\n",
    "import time\n",
    "import json\n",
    "import datetime\n",
    "import random\n",
    "import tarfile\n",
    "\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from PIL import Image\n",
    "PATH = \"../tubs/tub5\" #Shadow"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1.8.0\n"
     ]
    }
   ],
   "source": [
    "#make sure tensorflow version is the same as one installed on donkeycar (1.8.0)\n",
    "import tensorflow as tf\n",
    "print(tf.__version__)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['/device:GPU:0']"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from tensorflow.python.client import device_lib\n",
    "\n",
    "def get_available_gpus():\n",
    "    local_device_protos = device_lib.list_local_devices()\n",
    "    return [x.name for x in local_device_protos if x.device_type == 'GPU']\n",
    "get_available_gpus()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "#create df from path\n",
    "def df_from_jsons(PATH, max_items=None):\n",
    "    \n",
    "    #Get json files from path\n",
    "    json_files = [file for file in os.listdir(PATH) if file.endswith('json') ]\n",
    "    \n",
    "    df = pd.DataFrame()\n",
    "    for i, j in enumerate(json_files[1:max_items]):\n",
    "        with open(f'{PATH}/{j}') as f:\n",
    "            data = json.load(f)\n",
    "\n",
    "            userThrottle = data[\"user/throttle\"]\n",
    "            userAngle = data[\"user/angle\"]\n",
    "            camImage_path = data[\"cam/image_array\"]\n",
    "            index = int(camImage_path[:-21])\n",
    "            camImage_path =  PATH + \"/\" + camImage_path\n",
    "            df_row = pd.DataFrame([[userThrottle, userAngle, camImage_path]], index = [index], \n",
    "                                  columns = [\"user/throttle\", \"user/angle\", \"cam/image_array\"])\n",
    "\n",
    "            df = df.append(df_row)\n",
    "    df = df.sort_index()\n",
    "    print(\"Loaded {} json files\".format(len(json_files)))\n",
    "    return df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df_from_jsons(PATH, max_items=None)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Loaded 41262 json files\n"
     ]
    }
   ],
   "source": [
    "#load additional tub\n",
    "PATH2 = \"../tubs/tub6_more_data\"\n",
    "df2 = df_from_jsons(PATH2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user/throttle</th>\n",
       "      <th>user/angle</th>\n",
       "      <th>cam/image_array</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.06</td>\n",
       "      <td>0.02</td>\n",
       "      <td>../tubs/tub6_more_data/0_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.49</td>\n",
       "      <td>0.02</td>\n",
       "      <td>../tubs/tub6_more_data/1_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.49</td>\n",
       "      <td>0.02</td>\n",
       "      <td>../tubs/tub6_more_data/2_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.21</td>\n",
       "      <td>0.01</td>\n",
       "      <td>../tubs/tub6_more_data/3_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.21</td>\n",
       "      <td>0.01</td>\n",
       "      <td>../tubs/tub6_more_data/4_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user/throttle  user/angle                                cam/image_array\n",
       "0           0.06        0.02  ../tubs/tub6_more_data/0_cam-image_array_.jpg\n",
       "1           0.49        0.02  ../tubs/tub6_more_data/1_cam-image_array_.jpg\n",
       "2           0.49        0.02  ../tubs/tub6_more_data/2_cam-image_array_.jpg\n",
       "3           0.21        0.01  ../tubs/tub6_more_data/3_cam-image_array_.jpg\n",
       "4           0.21        0.01  ../tubs/tub6_more_data/4_cam-image_array_.jpg"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user/throttle</th>\n",
       "      <th>user/angle</th>\n",
       "      <th>cam/image_array</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.08</td>\n",
       "      <td>0.01</td>\n",
       "      <td>../tubs/tub5/0_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>../tubs/tub5/1_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>../tubs/tub5/2_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>../tubs/tub5/3_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>../tubs/tub5/4_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user/throttle  user/angle                      cam/image_array\n",
       "0           0.08        0.01  ../tubs/tub5/0_cam-image_array_.jpg\n",
       "1           0.12       -0.03  ../tubs/tub5/1_cam-image_array_.jpg\n",
       "2           0.12       -0.03  ../tubs/tub5/2_cam-image_array_.jpg\n",
       "3           0.14       -0.06  ../tubs/tub5/3_cam-image_array_.jpg\n",
       "4           0.14       -0.06  ../tubs/tub5/4_cam-image_array_.jpg"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [],
   "source": [
    "df = df.append(df2, ignore_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user/throttle</th>\n",
       "      <th>user/angle</th>\n",
       "      <th>cam/image_array</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.08</td>\n",
       "      <td>0.01</td>\n",
       "      <td>../tubs/tub5/0_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>../tubs/tub5/1_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>../tubs/tub5/2_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>../tubs/tub5/3_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>../tubs/tub5/4_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user/throttle  user/angle                      cam/image_array\n",
       "0           0.08        0.01  ../tubs/tub5/0_cam-image_array_.jpg\n",
       "1           0.12       -0.03  ../tubs/tub5/1_cam-image_array_.jpg\n",
       "2           0.12       -0.03  ../tubs/tub5/2_cam-image_array_.jpg\n",
       "3           0.14       -0.06  ../tubs/tub5/3_cam-image_array_.jpg\n",
       "4           0.14       -0.06  ../tubs/tub5/4_cam-image_array_.jpg"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user/throttle</th>\n",
       "      <th>user/angle</th>\n",
       "      <th>cam/image_array</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>79830</th>\n",
       "      <td>0.37</td>\n",
       "      <td>-0.71</td>\n",
       "      <td>../tubs/tub6_more_data/41256_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79831</th>\n",
       "      <td>0.21</td>\n",
       "      <td>-0.67</td>\n",
       "      <td>../tubs/tub6_more_data/41257_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79832</th>\n",
       "      <td>0.21</td>\n",
       "      <td>-0.67</td>\n",
       "      <td>../tubs/tub6_more_data/41258_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79833</th>\n",
       "      <td>0.08</td>\n",
       "      <td>-0.54</td>\n",
       "      <td>../tubs/tub6_more_data/41259_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>79834</th>\n",
       "      <td>0.08</td>\n",
       "      <td>-0.54</td>\n",
       "      <td>../tubs/tub6_more_data/41260_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       user/throttle  user/angle  \\\n",
       "79830           0.37       -0.71   \n",
       "79831           0.21       -0.67   \n",
       "79832           0.21       -0.67   \n",
       "79833           0.08       -0.54   \n",
       "79834           0.08       -0.54   \n",
       "\n",
       "                                         cam/image_array  \n",
       "79830  ../tubs/tub6_more_data/41256_cam-image_array_.jpg  \n",
       "79831  ../tubs/tub6_more_data/41257_cam-image_array_.jpg  \n",
       "79832  ../tubs/tub6_more_data/41258_cam-image_array_.jpg  \n",
       "79833  ../tubs/tub6_more_data/41259_cam-image_array_.jpg  \n",
       "79834  ../tubs/tub6_more_data/41260_cam-image_array_.jpg  "
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [],
   "source": [
    "df.to_feather('tmp/df_wPath')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\Jason\\Anaconda3\\envs\\tf-gpu\\lib\\site-packages\\pandas\\io\\feather_format.py:112: FutureWarning: `nthreads` argument is deprecated, pass `use_threads` instead\n",
      "  return feather.read_dataframe(path, nthreads=nthreads)\n"
     ]
    }
   ],
   "source": [
    "df = pd.read_feather('tmp/df')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(38574, 3)"
      ]
     },
     "execution_count": 62,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user/throttle</th>\n",
       "      <th>user/angle</th>\n",
       "      <th>cam/image_array</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.08</td>\n",
       "      <td>0.01</td>\n",
       "      <td>0_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>1_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.12</td>\n",
       "      <td>-0.03</td>\n",
       "      <td>2_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>3_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.14</td>\n",
       "      <td>-0.06</td>\n",
       "      <td>4_cam-image_array_.jpg</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user/throttle  user/angle         cam/image_array\n",
       "0           0.08        0.01  0_cam-image_array_.jpg\n",
       "1           0.12       -0.03  1_cam-image_array_.jpg\n",
       "2           0.12       -0.03  2_cam-image_array_.jpg\n",
       "3           0.14       -0.06  3_cam-image_array_.jpg\n",
       "4           0.14       -0.06  4_cam-image_array_.jpg"
      ]
     },
     "execution_count": 67,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "def to_bin(a):\n",
    "    arr = np.zeros(15)\n",
    "    a = a + 1\n",
    "    b = round(a/(2/14))\n",
    "    arr[int(b)] = 1\n",
    "    return arr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'../tubs/tub5'"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_keys = ['cam/image_array']\n",
    "y_keys = ['user/angle', 'user/throttle']\n",
    "PATH"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "def read_record(record_dict, path = PATH):\n",
    "    data = {}\n",
    "    for key, val in record_dict.items():\n",
    "        if key == 'cam/image_array':\n",
    "            img = Image.open(f'{val}')\n",
    "            val = np.array(img)\n",
    "        \n",
    "        data[key] = val\n",
    "    return data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [
    {
     "ename": "FileNotFoundError",
     "evalue": "[Errno 2] No such file or directory: '32540_cam-image_array_.jpg'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mFileNotFoundError\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-66-482145bbc900>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[1;31m#Testing\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      2\u001b[0m \u001b[0mdf_t\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mdf\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msample\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mn\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mto_dict\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0morient\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'record'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m0\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 3\u001b[1;33m \u001b[0mrecord\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mread_record\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdf_t\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      4\u001b[0m \u001b[0mdata\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mrecord\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;34m'cam/image_array'\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      5\u001b[0m \u001b[0mpic\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mImage\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mfromarray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mdata\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'RGB'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m<ipython-input-65-8b54ee502073>\u001b[0m in \u001b[0;36mread_record\u001b[1;34m(record_dict, path)\u001b[0m\n\u001b[0;32m      3\u001b[0m     \u001b[1;32mfor\u001b[0m \u001b[0mkey\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mval\u001b[0m \u001b[1;32min\u001b[0m \u001b[0mrecord_dict\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mitems\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mkey\u001b[0m \u001b[1;33m==\u001b[0m \u001b[1;34m'cam/image_array'\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 5\u001b[1;33m             \u001b[0mimg\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mImage\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mopen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34mf'{val}'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      6\u001b[0m             \u001b[0mval\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimg\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      7\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\Anaconda3\\envs\\tf-gpu\\lib\\site-packages\\PIL\\Image.py\u001b[0m in \u001b[0;36mopen\u001b[1;34m(fp, mode)\u001b[0m\n\u001b[0;32m   2578\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2579\u001b[0m     \u001b[1;32mif\u001b[0m \u001b[0mfilename\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 2580\u001b[1;33m         \u001b[0mfp\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mbuiltins\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mopen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mfilename\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m\"rb\"\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   2581\u001b[0m         \u001b[0mexclusive_fp\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;32mTrue\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2582\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mFileNotFoundError\u001b[0m: [Errno 2] No such file or directory: '32540_cam-image_array_.jpg'"
     ]
    }
   ],
   "source": [
    "#Testing\n",
    "df_t = df.sample(n=1).to_dict(orient='record')[0]\n",
    "record = read_record(df_t)\n",
    "data = record['cam/image_array']\n",
    "pic = Image.fromarray(data, 'RGB')\n",
    "pic"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=160x120 at 0x1E35C3773C8>"
      ]
     },
     "execution_count": 45,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pic_gray=pic.convert(\"L\")\n",
    "pic_gray"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [],
   "source": [
    "pic_gray_arr = np.array(pic_gray)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 120, 160)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test = np.array([pic_gray_arr]) #channel first\n",
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=160x120 at 0x1E35C365E10>"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image.fromarray(test[0], \"L\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "def rgb2gray(rgb):\n",
    "    \"\"\"\n",
    "    take a numpy rgb image return a new single channel image converted to greyscale\n",
    "    \"\"\"\n",
    "    return np.dot(rgb[...,:3], [0.299, 0.587, 0.114])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "data = rgb2gray(data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(120, 160)"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "data.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.Image.Image image mode=L size=160x120 at 0x1E35C3617F0>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image.fromarray(data, \"L\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_record_gen(record_transform=None, shuffle=True, df=None):\n",
    "    while True:\n",
    "        for _ in df.iterrows():\n",
    "            if shuffle:\n",
    "                record_dict = df.sample(n=1).to_dict(orient='record')[0]\n",
    "            \n",
    "            record_dict = read_record(record_dict)\n",
    "            \n",
    "            if record_transform:\n",
    "                record_dict = record_transform(record_dict)\n",
    "            \n",
    "            yield record_dict"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [],
   "source": [
    "# record_gen = get_record_gen(record_transform=None, shuffle=True, df=df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 66,
   "metadata": {},
   "outputs": [],
   "source": [
    "# next(record_gen)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 67,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_batch_gen(keys=None, batch_size=128, record_transform=None, shuffle=True, df=None):\n",
    "    \n",
    "    record_gen = get_record_gen(record_transform=record_transform, shuffle=shuffle, df=df)\n",
    "    \n",
    "    while True:\n",
    "        record_list = [next(record_gen) for _ in range(batch_size)]\n",
    "        \n",
    "        batch_arrays = {}\n",
    "        for i, k in enumerate(keys):\n",
    "            arr = np.array([r[k] for r in record_list])\n",
    "            batch_arrays[k] = arr\n",
    "        yield batch_arrays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_train_gen(X_keys, Y_keys, batch_size=128, record_transform=None, df=None): #\n",
    "    batch_gen = get_batch_gen(X_keys + Y_keys, batch_size=batch_size,\n",
    "                                  record_transform=record_transform, df=df)\n",
    "    while True:\n",
    "        batch = next(batch_gen)\n",
    "        X = [batch[k] for k in X_keys]\n",
    "        Y = [batch[k] for k in Y_keys] \n",
    "        yield X, Y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_train_val_gen(df, X_keys, Y_keys, batch_size = 128, train_frac=0.8, #\n",
    "                      train_record_transform=None, val_record_transform=None): #\n",
    "    train_df = df.sample(frac=train_frac, random_state=200) #\n",
    "    val_df = df.drop(train_df.index) #\n",
    "    \n",
    "    train_gen = get_train_gen(X_keys=X_keys, Y_keys=Y_keys, batch_size=batch_size,\n",
    "                             record_transform=train_record_transform, df=train_df)\n",
    "    val_gen = get_train_gen(X_keys=X_keys, Y_keys=Y_keys, batch_size=batch_size,\n",
    "                           record_transform=val_record_transform, df=val_df)\n",
    "    return train_gen, val_gen\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [],
   "source": [
    "def record_transform(record):\n",
    "    record['user/angle'] = to_bin(record['user/angle'])\n",
    "    return record"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_gen, val_gen = get_train_val_gen(df, X_keys, y_keys,\n",
    "                                               train_record_transform=record_transform, #\n",
    "                                               val_record_transform=record_transform, #\n",
    "                                               batch_size=bs, #\n",
    "                                               train_frac=train_frac) #"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.python.keras.layers import Input\n",
    "from tensorflow.python.keras.models import Model, load_model\n",
    "from tensorflow.python.keras.layers import Convolution2D\n",
    "from tensorflow.python.keras.layers import Dropout, Flatten, Dense, Cropping2D, Lambda\n",
    "from tensorflow.python.keras.callbacks import ModelCheckpoint, EarlyStopping"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "def default_categorical():\n",
    "    img_in = Input(shape=(120, 160, 3),\n",
    "                   name='img_in')  # First layer, input layer, Shape comes from camera.py resolution, RGB\n",
    "    x = img_in\n",
    "    x = Convolution2D(24, (5, 5), strides=(2, 2), activation='relu')(\n",
    "        x)  # 24 features, 5 pixel x 5 pixel kernel (convolution, feauture) window, 2wx2h stride, relu activation\n",
    "    x = Convolution2D(32, (5, 5), strides=(2, 2), activation='relu')(\n",
    "        x)  # 32 features, 5px5p kernel window, 2wx2h stride, relu activatiion\n",
    "    x = Convolution2D(64, (5, 5), strides=(2, 2), activation='relu')(\n",
    "        x)  # 64 features, 5px5p kernal window, 2wx2h stride, relu\n",
    "    x = Convolution2D(64, (3, 3), strides=(2, 2), activation='relu')(\n",
    "        x)  # 64 features, 3px3p kernal window, 2wx2h stride, relu\n",
    "    x = Convolution2D(64, (3, 3), strides=(1, 1), activation='relu')(\n",
    "        x)  # 64 features, 3px3p kernal window, 1wx1h stride, relu\n",
    "\n",
    "    # Possibly add MaxPooling (will make it less sensitive to position in image).  Camera angle fixed, so may not to be needed\n",
    "\n",
    "    x = Flatten(name='flattened')(x)  # Flatten to 1D (Fully connected)\n",
    "    x = Dense(100, activation='relu')(x)  # Classify the data into 100 features, make all negatives 0\n",
    "    x = Dropout(.3)(x)  # Randomly drop out (turn off) 10% of the neurons (Prevent overfitting)\n",
    "    x = Dense(50, activation='relu')(x)  # Classify the data into 50 features, make all negatives 0\n",
    "    x = Dropout(.5)(x)  # Randomly drop out 10% of the neurons (Prevent overfitting)\n",
    "    # categorical output of the angle\n",
    "    angle_out = Dense(15, activation='softmax', name='angle_out')(\n",
    "        x)  # Connect every input with every output and output 15 hidden units. Use Softmax to give percentage. 15 categories and find best one based off percentage 0.0-1.0\n",
    "\n",
    "    # continous output of throttle\n",
    "    throttle_out = Dense(1, activation='relu', name='throttle_out')(x)  # Reduce to 1 number, Positive number only\n",
    "\n",
    "    model = Model(inputs=[img_in], outputs=[angle_out, throttle_out])\n",
    "    model.compile(optimizer='adam',\n",
    "                  loss={'angle_out': 'categorical_crossentropy',\n",
    "                        'throttle_out': 'mean_absolute_error'},\n",
    "                  loss_weights={'angle_out': 0.9, 'throttle_out': .01})\n",
    "\n",
    "    return model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "img_in (InputLayer)             (None, 120, 160, 3)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1 (Conv2D)               (None, 58, 78, 24)   1824        img_in[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_2 (Conv2D)               (None, 27, 37, 32)   19232       conv2d_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_3 (Conv2D)               (None, 12, 17, 64)   51264       conv2d_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_4 (Conv2D)               (None, 5, 8, 64)     36928       conv2d_3[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_5 (Conv2D)               (None, 3, 6, 64)     36928       conv2d_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "flattened (Flatten)             (None, 1152)         0           conv2d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 100)          115300      flattened[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_1 (Dropout)             (None, 100)          0           dense_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 50)           5050        dropout_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "dropout_2 (Dropout)             (None, 50)           0           dense_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "angle_out (Dense)               (None, 15)           765         dropout_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "throttle_out (Dense)            (None, 1)            51          dropout_2[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 267,342\n",
      "Trainable params: 267,342\n",
      "Non-trainable params: 0\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = default_categorical()\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 80,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train(train_gen, val_gen, saved_model_path, epochs=100, steps=100, train_split=0.8,\n",
    "         verbose=1, min_delta=0.0005, patience=5, use_early_stop=True):\n",
    "    \n",
    "#     checkpoint to save model\n",
    "    save_best = ModelCheckpoint(saved_model_path, monitor='val_loss', verbose=verbose, save_best_only=True, mode='min')\n",
    "    \n",
    "    #stop training if stop improving\n",
    "    early_stop = EarlyStopping(monitor='val_loss', min_delta=min_delta, patience=patience, verbose=verbose, mode='auto')\n",
    "    \n",
    "    callbacks_list = [save_best]\n",
    "    \n",
    "    if use_early_stop:\n",
    "        callbacks_list.append(early_stop)\n",
    "        \n",
    "    hist = model.fit_generator(train_gen, steps_per_epoch=steps, epochs=epochs, verbose=1, validation_data=val_gen,\n",
    "                                callbacks=callbacks_list, validation_steps=steps * (1.0 - train_frac)//train_frac)\n",
    "    return hist"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "new_model_path = 'models/pilot_jn'\n",
    "bs = 128\n",
    "train_frac = 0.8"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "498"
      ]
     },
     "execution_count": 81,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "total_records = len(df)\n",
    "total_train = int(total_records * train_frac)\n",
    "total_val = total_records - total_train\n",
    "steps_per_epoch = total_train // bs\n",
    "steps_per_epoch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 82,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.7191 - angle_out_loss: 1.9067 - throttle_out_loss: 0.3097\n",
      "Epoch 00001: val_loss improved from inf to 1.36563, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 224s 449ms/step - loss: 1.7188 - angle_out_loss: 1.9063 - throttle_out_loss: 0.3096 - val_loss: 1.3656 - val_angle_out_loss: 1.5149 - val_throttle_out_loss: 0.2251\n",
      "Epoch 2/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.3901 - angle_out_loss: 1.5431 - throttle_out_loss: 0.1301\n",
      "Epoch 00002: val_loss improved from 1.36563 to 1.26580, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 222s 445ms/step - loss: 1.3899 - angle_out_loss: 1.5429 - throttle_out_loss: 0.1300 - val_loss: 1.2658 - val_angle_out_loss: 1.4059 - val_throttle_out_loss: 0.0504\n",
      "Epoch 3/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.2863 - angle_out_loss: 1.4287 - throttle_out_loss: 0.0500\n",
      "Epoch 00003: val_loss improved from 1.26580 to 1.22300, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 224s 449ms/step - loss: 1.2860 - angle_out_loss: 1.4283 - throttle_out_loss: 0.0500 - val_loss: 1.2230 - val_angle_out_loss: 1.3583 - val_throttle_out_loss: 0.0500\n",
      "Epoch 4/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.2105 - angle_out_loss: 1.3445 - throttle_out_loss: 0.0492\n",
      "Epoch 00004: val_loss improved from 1.22300 to 1.17472, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 219s 441ms/step - loss: 1.2105 - angle_out_loss: 1.3444 - throttle_out_loss: 0.0492 - val_loss: 1.1747 - val_angle_out_loss: 1.3047 - val_throttle_out_loss: 0.0488\n",
      "Epoch 5/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.1493 - angle_out_loss: 1.2765 - throttle_out_loss: 0.0497\n",
      "Epoch 00005: val_loss improved from 1.17472 to 1.14794, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 219s 440ms/step - loss: 1.1495 - angle_out_loss: 1.2767 - throttle_out_loss: 0.0497 - val_loss: 1.1479 - val_angle_out_loss: 1.2749 - val_throttle_out_loss: 0.0505\n",
      "Epoch 6/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.0942 - angle_out_loss: 1.2152 - throttle_out_loss: 0.0501\n",
      "Epoch 00006: val_loss improved from 1.14794 to 1.11081, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 222s 446ms/step - loss: 1.0943 - angle_out_loss: 1.2154 - throttle_out_loss: 0.0501 - val_loss: 1.1108 - val_angle_out_loss: 1.2336 - val_throttle_out_loss: 0.0524\n",
      "Epoch 7/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 1.0402 - angle_out_loss: 1.1552 - throttle_out_loss: 0.0502\n",
      "Epoch 00007: val_loss improved from 1.11081 to 1.07355, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 242s 487ms/step - loss: 1.0398 - angle_out_loss: 1.1548 - throttle_out_loss: 0.0502 - val_loss: 1.0735 - val_angle_out_loss: 1.1923 - val_throttle_out_loss: 0.0500\n",
      "Epoch 8/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.9968 - angle_out_loss: 1.1070 - throttle_out_loss: 0.0505\n",
      "Epoch 00008: val_loss improved from 1.07355 to 1.05807, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 241s 483ms/step - loss: 0.9965 - angle_out_loss: 1.1067 - throttle_out_loss: 0.0505 - val_loss: 1.0581 - val_angle_out_loss: 1.1751 - val_throttle_out_loss: 0.0504\n",
      "Epoch 9/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.9421 - angle_out_loss: 1.0462 - throttle_out_loss: 0.0508\n",
      "Epoch 00009: val_loss improved from 1.05807 to 1.01785, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 239s 480ms/step - loss: 0.9422 - angle_out_loss: 1.0463 - throttle_out_loss: 0.0508 - val_loss: 1.0178 - val_angle_out_loss: 1.1304 - val_throttle_out_loss: 0.0516\n",
      "Epoch 10/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.9035 - angle_out_loss: 1.0034 - throttle_out_loss: 0.0518\n",
      "Epoch 00010: val_loss did not improve\n",
      "498/498 [==============================] - 242s 487ms/step - loss: 0.9035 - angle_out_loss: 1.0033 - throttle_out_loss: 0.0518 - val_loss: 1.0342 - val_angle_out_loss: 1.1486 - val_throttle_out_loss: 0.0515\n",
      "Epoch 11/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.8677 - angle_out_loss: 0.9635 - throttle_out_loss: 0.0517\n",
      "Epoch 00011: val_loss did not improve\n",
      "498/498 [==============================] - 247s 497ms/step - loss: 0.8676 - angle_out_loss: 0.9634 - throttle_out_loss: 0.0517 - val_loss: 1.0376 - val_angle_out_loss: 1.1524 - val_throttle_out_loss: 0.0506\n",
      "Epoch 12/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.8306 - angle_out_loss: 0.9223 - throttle_out_loss: 0.0519\n",
      "Epoch 00012: val_loss improved from 1.01785 to 0.98209, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 245s 492ms/step - loss: 0.8308 - angle_out_loss: 0.9225 - throttle_out_loss: 0.0519 - val_loss: 0.9821 - val_angle_out_loss: 1.0907 - val_throttle_out_loss: 0.0499\n",
      "Epoch 13/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.7984 - angle_out_loss: 0.8865 - throttle_out_loss: 0.0529\n",
      "Epoch 00013: val_loss did not improve\n",
      "498/498 [==============================] - 245s 492ms/step - loss: 0.7983 - angle_out_loss: 0.8864 - throttle_out_loss: 0.0529 - val_loss: 1.0114 - val_angle_out_loss: 1.1233 - val_throttle_out_loss: 0.0500\n",
      "Epoch 14/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.7719 - angle_out_loss: 0.8570 - throttle_out_loss: 0.0539\n",
      "Epoch 00014: val_loss improved from 0.98209 to 0.97576, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 243s 487ms/step - loss: 0.7718 - angle_out_loss: 0.8570 - throttle_out_loss: 0.0539 - val_loss: 0.9758 - val_angle_out_loss: 1.0836 - val_throttle_out_loss: 0.0508\n",
      "Epoch 15/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.7346 - angle_out_loss: 0.8157 - throttle_out_loss: 0.0531\n",
      "Epoch 00015: val_loss did not improve\n",
      "498/498 [==============================] - 242s 485ms/step - loss: 0.7344 - angle_out_loss: 0.8154 - throttle_out_loss: 0.0532 - val_loss: 0.9839 - val_angle_out_loss: 1.0927 - val_throttle_out_loss: 0.0537\n",
      "Epoch 16/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.7186 - angle_out_loss: 0.7978 - throttle_out_loss: 0.0535\n",
      "Epoch 00016: val_loss improved from 0.97576 to 0.96886, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 243s 487ms/step - loss: 0.7185 - angle_out_loss: 0.7977 - throttle_out_loss: 0.0535 - val_loss: 0.9689 - val_angle_out_loss: 1.0760 - val_throttle_out_loss: 0.0502\n",
      "Epoch 17/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.6981 - angle_out_loss: 0.7751 - throttle_out_loss: 0.0534\n",
      "Epoch 00017: val_loss did not improve\n",
      "498/498 [==============================] - 239s 479ms/step - loss: 0.6981 - angle_out_loss: 0.7751 - throttle_out_loss: 0.0534 - val_loss: 0.9765 - val_angle_out_loss: 1.0845 - val_throttle_out_loss: 0.0492\n",
      "Epoch 18/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.6791 - angle_out_loss: 0.7539 - throttle_out_loss: 0.0554\n",
      "Epoch 00018: val_loss improved from 0.96886 to 0.96627, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 242s 485ms/step - loss: 0.6791 - angle_out_loss: 0.7540 - throttle_out_loss: 0.0554 - val_loss: 0.9663 - val_angle_out_loss: 1.0730 - val_throttle_out_loss: 0.0549\n",
      "Epoch 19/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.6645 - angle_out_loss: 0.7378 - throttle_out_loss: 0.0551\n",
      "Epoch 00019: val_loss did not improve\n",
      "498/498 [==============================] - 242s 487ms/step - loss: 0.6643 - angle_out_loss: 0.7375 - throttle_out_loss: 0.0551 - val_loss: 0.9704 - val_angle_out_loss: 1.0775 - val_throttle_out_loss: 0.0594\n",
      "Epoch 20/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.6481 - angle_out_loss: 0.7195 - throttle_out_loss: 0.0545\n",
      "Epoch 00020: val_loss did not improve\n",
      "498/498 [==============================] - 242s 486ms/step - loss: 0.6484 - angle_out_loss: 0.7198 - throttle_out_loss: 0.0545 - val_loss: 0.9810 - val_angle_out_loss: 1.0894 - val_throttle_out_loss: 0.0523\n",
      "Epoch 21/100\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "497/498 [============================>.] - ETA: 0s - loss: 0.6247 - angle_out_loss: 0.6934 - throttle_out_loss: 0.0549\n",
      "Epoch 00021: val_loss improved from 0.96627 to 0.92355, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 234s 470ms/step - loss: 0.6245 - angle_out_loss: 0.6933 - throttle_out_loss: 0.0549 - val_loss: 0.9235 - val_angle_out_loss: 1.0256 - val_throttle_out_loss: 0.0516\n",
      "Epoch 22/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.6134 - angle_out_loss: 0.6809 - throttle_out_loss: 0.0556\n",
      "Epoch 00022: val_loss improved from 0.92355 to 0.91956, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 234s 470ms/step - loss: 0.6133 - angle_out_loss: 0.6809 - throttle_out_loss: 0.0556 - val_loss: 0.9196 - val_angle_out_loss: 1.0212 - val_throttle_out_loss: 0.0520\n",
      "Epoch 23/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5955 - angle_out_loss: 0.6611 - throttle_out_loss: 0.0568\n",
      "Epoch 00023: val_loss improved from 0.91956 to 0.91934, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 233s 469ms/step - loss: 0.5955 - angle_out_loss: 0.6610 - throttle_out_loss: 0.0568 - val_loss: 0.9193 - val_angle_out_loss: 1.0209 - val_throttle_out_loss: 0.0542\n",
      "Epoch 24/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5807 - angle_out_loss: 0.6446 - throttle_out_loss: 0.0564\n",
      "Epoch 00024: val_loss did not improve\n",
      "498/498 [==============================] - 232s 465ms/step - loss: 0.5807 - angle_out_loss: 0.6446 - throttle_out_loss: 0.0563 - val_loss: 0.9565 - val_angle_out_loss: 1.0622 - val_throttle_out_loss: 0.0559\n",
      "Epoch 25/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5813 - angle_out_loss: 0.6453 - throttle_out_loss: 0.0571\n",
      "Epoch 00025: val_loss improved from 0.91934 to 0.91325, saving model to models/pilot_79k\n",
      "498/498 [==============================] - 231s 463ms/step - loss: 0.5813 - angle_out_loss: 0.6453 - throttle_out_loss: 0.0571 - val_loss: 0.9132 - val_angle_out_loss: 1.0141 - val_throttle_out_loss: 0.0546\n",
      "Epoch 26/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5761 - angle_out_loss: 0.6394 - throttle_out_loss: 0.0574\n",
      "Epoch 00026: val_loss did not improve\n",
      "498/498 [==============================] - 339s 681ms/step - loss: 0.5759 - angle_out_loss: 0.6393 - throttle_out_loss: 0.0575 - val_loss: 0.9507 - val_angle_out_loss: 1.0558 - val_throttle_out_loss: 0.0520\n",
      "Epoch 27/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5659 - angle_out_loss: 0.6281 - throttle_out_loss: 0.0567\n",
      "Epoch 00027: val_loss did not improve\n",
      "498/498 [==============================] - 302s 607ms/step - loss: 0.5659 - angle_out_loss: 0.6281 - throttle_out_loss: 0.0567 - val_loss: 0.9321 - val_angle_out_loss: 1.0351 - val_throttle_out_loss: 0.0532\n",
      "Epoch 28/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5502 - angle_out_loss: 0.6106 - throttle_out_loss: 0.0590\n",
      "Epoch 00028: val_loss did not improve\n",
      "498/498 [==============================] - 262s 526ms/step - loss: 0.5502 - angle_out_loss: 0.6106 - throttle_out_loss: 0.0590 - val_loss: 0.9140 - val_angle_out_loss: 1.0150 - val_throttle_out_loss: 0.0526\n",
      "Epoch 29/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5445 - angle_out_loss: 0.6043 - throttle_out_loss: 0.0586\n",
      "Epoch 00029: val_loss did not improve\n",
      "498/498 [==============================] - 251s 505ms/step - loss: 0.5443 - angle_out_loss: 0.6041 - throttle_out_loss: 0.0586 - val_loss: 0.9343 - val_angle_out_loss: 1.0375 - val_throttle_out_loss: 0.0593\n",
      "Epoch 30/100\n",
      "497/498 [============================>.] - ETA: 0s - loss: 0.5323 - angle_out_loss: 0.5908 - throttle_out_loss: 0.0585\n",
      "Epoch 00030: val_loss did not improve\n",
      "498/498 [==============================] - 239s 480ms/step - loss: 0.5324 - angle_out_loss: 0.5910 - throttle_out_loss: 0.0585 - val_loss: 0.9268 - val_angle_out_loss: 1.0292 - val_throttle_out_loss: 0.0530\n",
      "Epoch 00030: early stopping\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras._impl.keras.callbacks.History at 0x17df96f6908>"
      ]
     },
     "execution_count": 82,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train(train_gen, val_gen, saved_model_path=new_model_path, steps=steps_per_epoch, train_split=0.8) #change train_split to 0.9 for large data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'\\nRecords\\n\\nOriginal architecture with 79k samples\\n498/498 [==============================] - 215s 432ms/step - loss: 0.4393 - angle_out_loss: 0.4875 \\n- throttle_out_loss: 0.0551 - val_loss: 0.8894 - val_angle_out_loss: 0.9876 - val_throttle_out_loss: 0.0586\\nEpoch 00028: early stopping\\n\\nOriginal architecture with 79k samples and 0.2 and 0.3 dropout at the end\\n498/498 [==============================] - 239s 480ms/step - loss: 0.5324 - angle_out_loss: 0.5910 - \\nthrottle_out_loss: 0.0585 - val_loss: 0.9268 - val_angle_out_loss: 1.0292 - val_throttle_out_loss: 0.0530\\nEpoch 00030: early stopping\\n\\nOriginal architecture with 79k samples and 0.3 and 0.5 dropout at the end\\n\\n'"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\"\"\"\n",
    "Records\n",
    "\n",
    "Original architecture with 79k samples\n",
    "498/498 [==============================] - 215s 432ms/step - loss: 0.4393 - angle_out_loss: 0.4875 \n",
    "- throttle_out_loss: 0.0551 - val_loss: 0.8894 - val_angle_out_loss: 0.9876 - val_throttle_out_loss: 0.0586\n",
    "Epoch 00028: early stopping\n",
    "\n",
    "Original architecture with 79k samples and 0.2 and 0.3 dropout at the end\n",
    "498/498 [==============================] - 239s 480ms/step - loss: 0.5324 - angle_out_loss: 0.5910 - \n",
    "throttle_out_loss: 0.0585 - val_loss: 0.9268 - val_angle_out_loss: 1.0292 - val_throttle_out_loss: 0.0530\n",
    "Epoch 00030: early stopping\n",
    "\n",
    "Original architecture with 79k samples and 0.3 and 0.5 dropout at the end\n",
    "\n",
    "\"\"\"\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [],
   "source": [
    "#check against original\n",
    "def unbin(arr):\n",
    "    index = np.argmax(arr)\n",
    "    return index/14*2-1\n",
    "#     return index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.0"
      ]
     },
     "execution_count": 174,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "unbin([0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run(img_arr):\n",
    "    img_arr = img_arr.reshape((1,) + img_arr.shape)\n",
    "    angle_binned, throttle = model.predict(img_arr)\n",
    "    angle_unbinned = unbin(angle_binned[0])\n",
    "    return angle_unbinned, throttle[0][0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 194,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAKAAAAB4CAIAAAD6wG44AABYiklEQVR4nM29abMsx3Ugds7JzFq6++737Q8P7+EBBEmAIEhQBEmRGi2kKIkeSTHWhD9MzJeJ8AeH/Q8ctv/BxHyx/WEiNP42YY2scEgaUcORSHAXsXEDiR0glre/d5feasnMc/zhVFVX970PGynPZNzoqFtdXVWZZ18T/7f//jQRkbPGGCQrhAAoAMwwGgxDFfI8f+Dyh1DkR8/+8KWXXhqU3hjQPyJCRIEIAERERMagMQYRAUCERWTLWkEIYDxQDRTJ3v/hhz71uc8N1teKuooxOmeMwVAXRTnzVW0kAwAACAze+zoERCSb+Bh2d07unNhN09zHyAzMLCLikizLLGFRFNPxYTGbV8XMe3/j+q3Nzc1sOESyReVv3rx9cDi21u5s7Z45c2ZjYyPGuL9/+Is337h27VqMEcEIAgAgEhhCRJ2dgcXA3iAiY0yWZc65qqpijLu7u2fPnl3b2tnc3ATgvf073vutrY31rXVjzPr6OrkUmENg61yMXJY1IubpAK0BQAAGYAACCb6aj/f35sX0ypW3X3/99Tu3b9Z1WVVVURQ8PcUQIpd1OPRxLKZOUrGphOA9Rx8kBKgK3j+Y3bk9G4/BGmMAgJkR0ZDCjATAGVPXdajD2mhU1/Xk8HAymVhriTwiIIJOUldERBSoRwczAyEgICIIOOeyLEvTlJkbCIkwC7fDIACAntTbImJ30N1WRPSzQTIRiMzMMUa9p1PEQWRmiTHGGEIAgKIoRGQ4HFpriezewf7e3l5ZloRGEEQEkaT3rBhjH7SKx3ogIlVVhRD0cYPBYG1t7YHL9ycbWwCcXk1u3LgWQmDmPM8REYCA0BgBJIRm+pgJgIAwRA8AYAggOsIktbW3eZqNhvlsmtV1qZMNNGEOgt4lQpgKYOCqnoVsMBIfZuPi8KDYuzO7ebO6eRPKAqy1NsbYW0xERBApyxIR10frp06dyhL71t7+nTt3ynkxICBqAKwzh3Y5WsKVPoADRwSLFkVEBLIsW1tbS5KkDIGjsDAzEkCMLBFEhIUVYDr/DpzdiktvdA9V0HKI3a9EJMbovRfGEEIH+A5CzFzXZTkviqKo69oaUQA3KEvNUriGhJWxCYuAEAKCUAgBER04IiLjBGJZzff29ka+jDHeunXr4OAgy7I0dfrEfOQAqGMGIJGDBF9ZsWAMGGqeIwJElowlcolZGwwPnT0MMdQliiR5WZbzqi6Y2VgCROHEe3dwpTzYL65fH9+6UYwnUJcQGQDAtiClFjwQY4gxVlU9GAxGawMiunnz5pUrb80mM5HuDZfG3chXcaVbcWZJ0zQbDBARWLB3XslXQUVEHX3rnWOMxtkOgVZHZBGOMeodOtYiIiEEYVQqJERjjCXiEA4PD6uqunXr1t7e7bquEZElAJM0iCskSIgMYIxZPCfGGKO+hj7CWqu8OoSwt7dXFMWVazcRMYRQVaVzZnt3yzlTVvO4vW2tdUkCDQqRMUYklPNZmqYuz4GUUUfl1QYBJToyaeqcJY4+1N4YU8thJI8UAaH2XJVyOCnn0/jmG7cmk3iwx2UNAA4gdTB0JrPMIGQIAIlEJIZmOJesDYeWzNtvvPnqy6/cvL5vDIwGCdS1CIiAriYRIS4obIWwEJGQIgsCKASzLMvz3HvfXWYARRhYAICE+qspImStUrNt+URL3EuQ7ji84i0KWFUFIocoIQTvvfeeRebz6Ww2cc7VIcQYicgZS9aEEJAAhIAaPkxkiEgk9BlGH3c7qaGPrut6MpkkLmcEEUaULE9mU5dmDnAU6holAqoAFEAyxsQYfVUK18K1cw6FQQQIgAg4GoLEmTxxgyzP0qQqLSJOiojorEu4ksnh9Mq1vatvH+zvwXgMICBACBlAKmA9kI/Bigg0KI8xchBu5LExg8FgOBwe7u2PxwchwCCjHs+Ebomxne0KO20+DYUYnACDELrBYJTnua8DIwAACYqggDAzskirDXQE3a2gUkz3COaG+iUyE0NklbIhhODrEILSpQAGhjrEWHvFKqVva8mYtM5z51yMvqoqoJYJoWkBTIho3WLWfaBaa/UF9EHd8L7O8zxNB5F9XZWHh/vDYb6ztYUI1lpAA8jK8AkARQiCL2M5m7LE6AMAJInNB6lERuHU2WGerY+GW+sboSqrqsqyk7NZcef24fVrd25c37t1MxzsK3XoYIKYOEkT65wjspYBCQiBoooxBGuttTYxiTN2bTDc2dre2d62sO+MLcuajE4VWuGLS/IYZAXGCEbERxAEQ86ORqPBYBRiZCQWBSEIS6NzRGYG1V8U0p087la2T0+LTxGO0XtflmVdFd77yWQSY3RJapMsz4fWJMP19SRJdre2t7a2kiQpy5I5WCLnHFG0NulksNJxA0g1KxYwBsVkiYGIQIBFAMAYY6wlEI4cao8kIdTz+dSUZr6+Vtc7ofYcAjkGDkAEYEyj7nAMvqqKqqpms1ldFaqcD7PUOQcoAJA6O8hSgzSfzsb18Pr14tVXrr791uF4DKArDmAQnIPBIB2OBmvr+Wh9OBwO0tTZPpWEENAaa+0wy0ejdSKq61pEHBlnbOqcxEigUJGeVgkKhj4HWzBtQhYkESGy1iZZlqZpKEslUaSGwcagonShVfVvuHLzlZMkEJVL++DruizLuq7rus6yLLFuY2NjfXNbgJglHeRW0Fo7nc9ms1ld12nqzpw65dI0xghAUbhl+g1Bl9W0EwGdOACAGGOSJKqo63nV0tfXN6uqqsazGGMItaR2Pp8e7t/JU1fVRW6tREZHYACADUoVvXBNKIQSQz0ej0NVGmO2trbWRsMkSaIPupjR++l4/LOXy9u37ly9ejidAQBYAgCIAhtbtLU9On1m++SpjeHICpZ1Pan93IogAFXexxidSxlAgiAaX5XnTp00SK/eunHnzq26Eg41CgC1Wllrnyi37LPQ/pjP5+vrm9OynBfzC6fPjUbrRVG4JJtNxtY5Q05iiLVvFN3ILsvquq6qKkSx1uqqWWuNMWmaAoByWtWQkyQxgJ36s8ADFmesEhYilmXJgt6Hg4MD7FgAg3NuZ2eHmQEISNXMvgwmRBwOc1gI/sWYTqf6DopM3nu1iKbjCREQIhBQ6vI8NUjz+fTGjWve+xMnTgzXNlyM1iYIYskgNOhS12WMHmKYTCaz2ezWzRtZlmVZZowJdTk+OLhx48abb7751lvpwcF8NgNDkCSQZbi9vbl7Yuvsud0klSSLZGuyU4FSaBpgYvvKoXOuY4N5nocQpvNicnBYlQIR0AK1grYvdVomsUq7zb9AQThGQTTGOuec2s0da23kOItKEhWjMUaRJYbcl/F98u20LGzXH1hERO1dBAoseVbbNHNJmiTJ2mjY082AoWEViIYbDdySNaobI+J8OmllBESOwo2FvrWxuVAkewgxOTysvJ8X08PDg9lsFmo/L6aAfPrUqSxLiAiYgaMEHwS8961/Y6FwAACgHE7Gk8mkMf98mE6n81lJaMnOh2swWoe1tWx9uD4YDDY317e210IsrBMyFUKBUpPxYAMkYpmBRTiCS8g5B61pPxgMvPeTw3Fd15kDSiB1LsbYqlR3Y5iCrV+iQ4UY2ceAxqZpmiQJEsUYhVEEO6YngiKIzFVVKTUAWViWuN2/3DvTGVfcGlv61dra2mg0Wt/YHI3WjU0YiQVE5PaNmwDArbgVBlnMhaBToI1TMG9uDDsQGmP6ZrSqdR0F1z4y8+7utoiEsF2eOFH7EgzkeZ7naVlXvq7nswkz5/kwS4kASFiwv3qg9jcI+jr4ui6KoqpKfYpyrJOnKnUnbKxvDQYjSy5JKM+59tEgs3jmKMwoYCR1CDbEaIwxxliTgFCItYgQIDBPJtP9/X2OYTDILBJEjrVHq57IVdfS0dG8NGGIkZlNYvI8T/MMEWMISmpq2EhkiUqCqCZNjNEmCwD3NSyFYd+hIbGzkljJF0X2b9+ZTCZ7e/uAxgcual9XPnBcy9OGXrt3F+xgjGhkQZGWiGo/J9UekiRJEuec2r7qpMyyTB1Yw+FwMBg45/J8CAAQ/XQ2PjzcH4/HRV1Mpz7LMomhqipEdMaKS0BQlUdBYBBpnYNCyNh4kAARkUSAWcAYm6bn7xEAsDbJUkMUY135GmKgxDlGjDFwFAAAa0UwerDM4Jw1BlXgFWVhrXXOFUVx69atG1euzqYTh4TGSvTCR0kW+iy6s4mlp0gzszA659J84GwKALG1LTmiROnoD1pdj5lNBz8RROpr1H1hCCyCgLyQkfrQra2teVXu7+9PpvOy8ojGuMQ5Nwl1i5fErf4PgkCqJxIvm0NJ2uihy1IJ0jTtKFtVBKXv3c3NNE0Hg0GepzYxg0G2sbFmLJW+xtZJ1LlUg6+BUFq9AAjRWCADQEKGkQBIgFhQgIxN8oyGawkzIxCiiNRAggAE4KuSiJARGBETiMgxxhpsy3wsIoYQfB0VWxFxNpvt7Y/9HPIsUhINUpoaEAXhEoBFRH3afei2F5ByHuNSVTsBSKen3FVYmQqDWsSqVMfG6cgxIjbwPZaC+9imnwaREauq8mXlvSei4XBobYLGAsAgcz0uQx0dExpu0YkbZEUAyBK7mGQrUQCgnM2hFcAddIlodnhIRM4Z50ySJFme5MM8SZK1tTUwZMgmCSfWhZAQWQAAIWEUIAECNK0Gi2QNlOSj1IFDYBFM09wObQWlAMcYo/plmUEXMQQiC4goaESFCEAkq4oVMyM2moVyHom+tUobXRwRE2O9932jSEQAe3R9ZAgAAhCRMjdEjA1sqLdijCIsiCAAoLLNhBBCYAAiMrTkp1yBbjNY+qQWYzTGbGxspNkgywYCNJ0X8/mcOfTQInYwjsILOxhIWvZTlsVRCkZEa10fxZmjCMcIFpABqioiirGYJEkySZxzh4eHaZpmg+FwOHTOpUlOzhhjAiMjRGhfngjQgBCCafV5KxgCR4PojJPCiYjEIKGMsWb2Ej1zsIYYIqkDCiOSRRBCsc65xgJGtJZU0lhr5+U8xmgMmBSShIDZ+6iaagu4bpmOg2tf70UiAsXxPikANG5eEHXtKXdYUHCMkTVkZJeItWPF3dI3x+1ViGitraqqrmsEY4wDNN77qqo4VAvUBBDBhogF22gSqiRuAEk9p03PQ6ccq4O3afwWSNYRQUIWLZIBIgJufKXWWkVNxWBEFEFo45EtLyEAEAQNUqGx1rIy17quQwih3jRISI7QWHKAVRBSbYYkRuWIiM4IiCAEW3m21qZ5DgCBORtsba5vWDMcDZyxVwkRUULgxIIxIJFjVC+HagACLYuLMSIuaIgIRQgAMKSzqsZBmm/uwGhtSjFyKKQE4BQSFAm+jrEO6BnFsyeb2ST4CD5GE8E5Z4wTwdQO6jIah8iGIqCgMYYiUY51Xc/rqpLoCT2hN5aRphzAmdHmepqmZVl6P49QB5nP6qXwHwAgLFyPCCgiGuFW+z7Uq/xJf6UWh7S+T2OMKl/RR+dcmtrEpGmapGmapJasXd8YNS5SZ6OEOs6TfJRnWTHzQhQRfQixriRUBqKB6CwEzwh1iPPIhUAFKAIs7lbshTJj9AEDGwCESCgiEZiIahNFxHO06pPqXr0TJ76RgtheoIEk7s0TmtVpGXbHtpUypK8lwUKdkVZVYxB1U+iIKACgeCqtK0OR2pCNMTqH1lpCE0JgBpUmnQGjS9xZeh/+8IerqtKTIQSdVFVV27tnWxRcGK/9FVihV/VyK/UrS1AlX43UzkzSaHyMcTQYqcfUzO1gkG9sbOSDNB8MqqpK09TaJE1T55yIFEVRVRUBKbpkWaZPrKqqLMvxeKz+gJWw6cpxZz50PkFpPbt60ChZ0AbPdY2MMToNgM7nTMrVGFt/LBIAtzQgse+tBWFhEkIAat9Pl1hV5ihAjELCIqHTi0EYZDAYKI/qnIVoTZIkqlpDLxpIRECocYg0TY0xo9GoL5tns5lOVQGsfPv27dsA0C1Eh16qWxwF/CDNOtVkOBxubGwopXZa9JL2x3zrxq26rouimJdFjEFRVhDX1oeI6L2HEhExSVye51mWTSZTEYjCgWMdvP75GIBQFezuIDILQ8czZXnAMnSxDdhoFK516hKpACaisiyb8AuZztxdMZPURFOy7hYWBRDVyRwRUXNrbCulmDlwVIOVGUGUz8SIzCIsUFTlrJjP53NDLssyIquYce7cOSCDjQaOgX1dlVLKrCqVvIqiKMtSUyyY+c6dO1VVKWXoVIuiODg4sMmgE/M6dF2Gw2EH4O4TEX1ZdRxYR2caWWs1+pnnufpwiOjC+QuIqKugaJFmziaJdcTM3nsfaxGx1hIZERiNRt57Zu74TZqmeZ4rP6jrutNd9G2ts339pmM2yhcbam5xLsbYUTCKiL6QMU5E5vOyrjyzoEVEFBYGYQZRZk0IiEAKVew4s0K6nSAgQowxCgMQGkNEzBw8MzMKMANzVCspsjCygNRVFWNMXDYajTY3N5N8gIggdDCeFEUxm86LoihrX5blfD6vSr8/HetCKP+s61phlmWZiKyvr6vYUU9ZCFyHeV8zUsVe6Qx6zsLuYJjlzKxE2UcIpeDuJjqMMU/Rk0mSDDTUOhwOhtlgMLBJcur0iSzLhsPhcG3b2oXffj4vYxQimyQZM4TAxswVNxCb19Z4Qef46wsRjbwBgGbCdKDtOEoXTeoCpZaIQgiayMIRQKj5CQMzMAAiMAti3wOklvGCvjstzCu9Ehrj0BhFK2ZGBkYG5hg5RmFgRmaQNM0ILRENBgMwdjqdTifzsiyv/+DJsqhms1kdVAPiugp1XddtqKOjM9dOQa8pi1q/FUZrEpYAIIhNcKHzF61tbcCyPa2f3nskNITWOcSkW9yqqpiZOdbeV7V0RG+wtegAALlRFKzNB+lgMNja2tre3dna2trYWBuNRmmahsBKo3meawqA5pB0caq+AwfaeHk3FMCICK3c7d5Qr7d9Sm8WQiSExifMrCEBEmFhjnEhg5EAO2NJgNQT1EFc6VrANI6BBgbM0ObWYAQhwdj6GJlQgOsqlGU5r8ry6pXJZHZwcDCbFoEFEdWSUQ+iIafqwtC6/hS6T82bVP6p6oVGZqw7PurVjT50AUA5QYeX3Lq7rbUrV7ao3shjlUQKKiAajwGQX3/9dTSUpuloNFpbWxsMBvddfiBNUw0cOeeSNF/f2FKnhw88m5dUeYFK/VnMIiEo26A2fIdtCpECuAO5DittHENkoTrGGH0dY1TYkAgCY6MFdevDnSat8wQAlC63AJWKRRhFBKF5IRaJICxIuHB0iAgjiTAjMUIdw+Hh4Y0bt65fvz4+nKHBwWC0sb7psiRNc0SMQSKIc845B2g7Ptl9qmIlIrqIAE0WQ13Xo7UcjwsetBr78SBXRbpNZwoiMp1OoZV5vh0xRuAmr09tWWUksUEIJR5fVdV4PNbX+OlzP8+ybH19fWdn5+TJk9vb26o3bG5uxhjn83lZlkrTnYNzBbQ6OgD3GTgiWk0YM8ao50zhPx6PDw4OjDFJYmNkEoIoHIEIQqtnKQdGBH1ciAKqWbffKpcKHKNImqYuSesQCEkE67pG61hsGyyQCIwGALEoiqIoqsozs7FJNojOpXmer29upGma50OX5qYZzhjDAhoq7tQclZRlWSoGqNDpWGiarAZIdEVsmhyrZEVZsMS+TdU9SEPCHSzLeVHXdVmWla+991Xlq6oq6xpaTduIQ2gAxhGC97NZcePGrRdffLnT2pIk0eA3M8coAGSMQ/TS03LbPNEmKZ05UpPFjH1hbAFAgAEWfhkRibEJ4UFne4kILwV+lTN1SdEijNio2djy7tZl3WTAtCqDKgUiCBJVgwAAiIDAnGWDqvIqHtWttr6+ubG5tb297Vzq0tS5VNGxIVnnOsIqy0JtUwVwqzOaPlLPpgfd/Ptr0WkrHdXqwWC00K775hO2/g3FIbXTiIgAm5C2LJxWPsre3p73viiKWTFXC1g5QVlVIoBACODrUJXj/b0D5e19G0yBYMiGUClBY5NJ0ijPHUPsIy4iLgRJ5/FQrGxKBxiBUBg0309aT4eqyiDAwoQEAgIIIgvvR+P70HqBRf56EI7MUdC1vujGxQ/CLCwSfF17H2JENEmSWJsMR6O1tbXNzU0gi2iYua5D5Fqn0So73AkeJY66rtU1qDDuALOzs3MspSIuPDP9z7KuOi6t6SX6leba9bFE/93e2tCTxibKbIkogmxtbelv6+CLoppOp7PZrKqq2wcHCnjNNKqqKnIM3DhSuql1uJUljT+Hen5fRFRp2v27YNEKf0Lp+HgMrPgVgyrmwIISRRgUVn0KBpbG3dU4sBb5WXohM4NBIqPxuBhjo7gBiGCQKCKsCR3MIcbJpKyqyhg3HA6TJNGAeFVVV67dUPsTEQWo1S3BJWSM6YK1AKAUfGKwo2GxztvVwFJWtVA8kqbSh/oy21qw9E7zWpHNBwf7zewAuiTyCJK4Jn6c5oO1tbWNjUZpP5jO67qezWbT6VQl48HBwXw+T5KsLMvZbDafz+vac8ubN9ebaWqUqHv5LmNpRfWzIlGB2E0sxti5CzStlYWFRRgIoeOoAqymESKgprwCIEEHXT0TmbWMCdFEhgiR24hFp18p344cQwhElCSZ2uKa0gyA1lqyzrk0TdMsHyo/VNt9tJZ2i64GsRp4IdSaChNCvcS7gu/DqQ+wFVasqxFA+jZYp5dpkhMROWc0F0evv/fCeRUTVV1XVeV9DCEE4bKovffzeenjbVWtnU2stcPNzcFgsLGxoTbSZDK5ffv2eDx+/fXXteikdSlqKjjM53WShEacG9NlP6rbEQBUGKOG4RsZ3M6/85io1sAxonq4I3BUkAJQhyMgjVa1WD2JHZg7CgYS7HQ8tXo7ztx3FnKUEEKSDBC9tRYI26IEJGd3tk8gGjRkSOObokQznd3p8tq71GiVwUrBne9J32G8tw/HjRXp1X0mg3zljB6MRqO+DO4EwdpoqO4EJJNl2WBAxji0hqPyeamDL4tKdcm6rvffutoxYYVxURR1FdIk93W0pjDkrGFhHzkKQ12DCAOUCmCirgSwcWB1vuvG8wVHGHeLNSKsCa/QJKYxCAEDAqJIVDmLsog39FLGGhrv6LgnHRdLGTV02aZSR+EYI6kqQGiMUZssRo5Bbt68qasTgzBzCA0sBUppHbCdHQ8AXUqNwriL9pw5cbIDUh9sfcbbP5hVzf37rk1mvn379hIfaq931jjnkiRxSapqfJrmJnHWJGmajkbrg9FwY30TWlECSTafzyeTiVpEs9nMx1DWFRpCQ2QNWYOBICJH8BEy16iuypMRG5cImUV6WsfPQggLR0d/kq0KrpLViAQRWBZGpPrbQtws9Oqlgw7AHRRFpXnnIOuFR7iJZqJ1NnGZLuu8qmKo9/b2YhQGUR91luWDwUAI14ZNdZDOs2M/WZZ1ruP+BJuw4BElq++46Iux9e2tPn13B11VZseHdH0n48NGzNW+KIrxeMwMjGDIKX6TNWmSDYdD9WRRNrDW5nm+tbUlIgcHB9evX9eDRkoSde+M6BMH/Rpd6GXR6Lt1AG5dlXUiwTM4xiSgrQUq7+uiND44QQsIIQAAEEUjtYgL2FApQ2zKk4CIjGkDf40jkAgJEaP44IO1GUeQCJlJZvN5Yi36CByjr0JZlLHyEKtQlb5ymthmgKAG4NSJQSsiu5tnoFUDj6q+iAzAMUaIMYFoMFJgg86ZxIDpBJAxhlgQkZBQkJC6+KdwY0RSR9mEAMBxBkcENgCom1avAaPnEQDWBtu9Kxca3GQyU+Qri7osy/n+9cObPoRgcnA2TZLMplmW5kTWhHrgcGuYU6hjaoldIWHm61iH6CEgIIpzlKaptVYhy8wsAXrMuFMerQhqLR0IARAAMUOTICXI3EyJUQRRVnBZmkBDH4Pa2LB02hY28DB9hsbC2M/K6AWGOynSkZq07pu+3gjLgvMokd3lGlbtbKWsuaUDaJxwrWuvfxNczgh+h/PtceNgAABVm5sVYOx+e3t8w9dxNium0+n+3kEIPJ+V03khDJoBr+ETACACaxfO8xaKDe6GyCuyRvmfXZl/E9uJMYoYFlJDCUEAWISbzA3lANLzPStC8ELbagFsCAV6/KQLnapjs0uJFdAsrdDztWKvELQDcH8OR1f2HWC/JA+O4MGxQ0SEFqh27E9WziMyAHSZOFq4AEISAwAgGgQtsG7Q6MSJE8IYAs+ruiyqyWR2+9ZeHeL4cKLhgNZ1D0S9mOxC22/fE5poLPMRADe8uzVbQwh1DCIQBUiL+BAYgfWAAZrs9vZ50pHvYqptgAGBDLYBZ+kyEBCjtNmiPegogImoc1ysuFiPAuBuB3BE/WmO2uK+lv90t8XWiIeOSKCnePQIuv+5el5Bq7FwABDRoBUjIgipZ5F7FFxw4WxqbTJySZ4NkiTjCAxYV957XwcPCwTt/zWcA1sOasC068b9+Vp1UYEgALIAs8TAvg4cQUTZWePAEoAogNxocfqk1iOtQZSFekXUgFOar5rV1UpRRmTBSASC0theSpom+EC9TON2Gsew5RX2uALIo3jAvCib65TtDnV6S4ZL55eKSvQrgGVe0j9PCwYDCAQQBQDRsDBAVFGlWov+PEQnLEVRVCFyFO8jEXVV1JVvnNjklZNBK0ClH9uHtgoL2op1aLUt20qEZiGUglufXCdkQQQiQjeN7mHYlFM2ukpLqS1RIDCy0n0QJmmsdWGIKE22BzN3PnTCwJE4Bo5GuHVxLgCgi9kc6/tA60dD6EoEuj9NUGjSJQEAhERlMDB3YZk+9iz8WXqsnqyOWhdKVv+XvfOGG3sDEQE054lAAseodSkAARG57e5iMgPqwrYSA4vUHWy60XpaQATUn7HMm3R20o+FdMhqGVAY9ROAWDAKxNhCXXViWWJprY+yqTsWBoML60ikiykJAsQoDdUqNIkiCIFospnCNwoziRASkQ8hkqhrhZQ/ACAi94qcEVdiuh2GAoiyDhFEYWzSypW1IApgJEEAAaHWvdpicBto6zHuBWG2S/auYli12b7+rDPIsgwANHjaMQ8AODw8dDZ1LkVqIqrqmu5uqHBSfiYSO59li1qtFyGErvZgCcCLSxsCbfQsAABSFARGEASEhX91gT7KwNugYSuVQUNPSBhEiCRowRg38rp7ORFsvhLUPiF9wcm9wv7+e8sRVbbPoo8dR3/S5/Ar91l60JGvjn10d779eez0Zx3z+RwREQwuj42NDWH0PpZlyVFijEmSjEajsqjwSAiLCImk06L7928zJLGjfn0TG4MkWVYUlcsZjWMGZ1NrEs8lMwQBYmEERtDab22p0U5pIRWcA6RG01voX8CGIMlSddSJSIycuCx67xwCYJt42RT2OZvETOq6jnNGQ8M0AXUmMC+kJrSMAgAAjDF6QbOy1HDnJjExoGm7qDALgVhYzL+/EMeaYQAQ40IGY+9r0yVDLafU5CZtqar7IQJI55hr0UvpRTxGhToRcYydiOxHQTp1OcbW8mzzq4xpm1nV0l4DHYIioi19oCQVMghGBIMIABHZJhKiOoYyZ2n4swAq6UpL99BYl13ieDfQtlFIFmAQjhCFfYwE6LruLYYARQAiiDGOiBVvRNAYo7HjbmKLW7eLTr1QLreVqFpOZ61T3TLGqOWlNiX9TQesxmfe2sFKhkqNKoMRERp1oNM5IATfLT1RAyEACKGtQ5fu5hHRGA2v9lh3S3ARAXW1sZdDD+9hSGu5dPwAejxJ/7WaEATRk7OgkYYYYozc04oFG4VLAEQ0BbNfowtIncIlnXatqACOQBZtkbiVu4ws6gYiJCFGRjCIkQgBIISgEfskSaSNnBw7yb5OAa2Zp25ejcZ0PkVdEd+Y16vlr8aYjlUjtuIIBKghxD4RQ48hNzmE7Xlq9Gh1ibXFZEKt8GR1+DSPFogQqedZbJSSjgzfBbqL45UX617VgmAIsapqm/nISqQYVdFVXVVAACK0xtLivt0qtOTciuE+i2ZGz9Er0jBAjwqjILDEIFEErFaQ0GCQIaK2YcNeNjWvaBftoDYJHlsPu7piNH0H2g49nefEEnbX9wFszSrRNKy7VxbVH8Ys5St2ByFIwyAACanR2LTRTftTvbT5hwwCKYCgpx4f+9Cjb7iiUvVptwHweDy2ZTGrSjY4WltLnFMfvVYmogBK4+iIjXzV91it7xNpOy12dC8ggiGIeO9902sOiYgsUWQQZBYN8ElARLBARGVV1yH4GH2MaAwa45xDoq5bUad0UNuKrA9g1Ho1xK7SqxE5zqlt7XDhx+4vylHQ6mCz0Hj7CnCoj8nVQmjUVkEUIS2SVYmrWjQsCuSWRv9WCwfVe4PxO19gfQwEtntLXnSUaWkRGsVGGlLWEyqL2ozM1p5C5efNWwOo50uT0zhGEYMmCgOhxIaajTFASKkRC8TMDMY4a5MYo3NJkmSDwcAYs7m53ZqDS4H3FRbdpVhoOA8RteRkOBzmee6ckxD70D12rfv/Rlz4kvqjc5j3Q9oAwF1ChAhzExPkKEimRY4lhT+E2pATwcDSJea+A9j6b9cndfUmdefbb8WOx9Mtm7g0S5Mc0YTAPrI69NQj0fyGFu0sut+3wF0ejaOn84kLADKzKtKhrZ90xiXWOJMiAjjEhNhIiHH91CYieu/n87mI5HmujUO1Jr3PfDpILBEcs3Muxri3twcARNQlXGoRgxwxmXSseM26T8ZFeLEfxdICsqMAnvX6ogEAN+mLUteLjnm6FCp0QmBVbylEjyHG452yR8eKDO4A3GczImJff33Ph7C+s9VFs4koSZKle+HqfXt3p85xs6hD7AGdqMG65sFdSUVq0zQZZIlzllKDCUWEOlQ2LtLJiqLo1KXZbCbH1RTBsqEibdMgDabKcvfRGKNF02eq3XuuoE73Lbnm+hXOodl0TdO4JOliz8kwbwsskl56emviM3Kvf4GIeJ+AUBdsICrqKlR+UaX+HiHdX4H+eYsA0/HMmQS2QlJDkmSIVV3UghQtxs5LxNqgUkIjtDqPXeOg0Nyadq2QkFgQGSkBQqorwApySDmyj2yzvKiLaTULe9o/ptIMIRGZl3Ntv6z5GOqV1fbDXWkrACChtp04sbkdQtDaTo29ODKYoC63JtO4VgAjYkA5CsX+uvQpGAA6wutXNkSR2wfjbln7GEaxQ3dDRMZZYwyh3d3d1TLJNM2yPM+yzLnEWpuMNnRG6zFqxdDGweTw8DAdbGyMp3t7e4eHh/P5HM088iTyPPIcjQVyQE7QMDQtu43NAUDNH30BBo5CFg10hVPYeqtXKVjNQ3Uk9LpUSo9Wk8QtkIgFpOmoUNdRswB1NWOMRVlGkMnksOsrQ1qZQ4SIl09c7mhUy+60ZgsROyrpy7DD23vdEnecnIhijNSWWmHfx7TM/+4m7To2rgnosjwA4NSpU0fvICKZce31akoDIoLQ/v5+BM29jb415Jg5EiVJoj20Nc0oeNZmPPr+yiqcS7QldYizXr53w+pZuKqa9F7mQG2TL0S0iE33TRGpvQcAbtL1MIpQlw0LjVmUOLMy1RZ4mhndYkNrTXYSihGIKLUpIwRhstixPmPQtdyv8pVOTItCvfeKH2matp6aRTGBiNRlSb2y5m5iZVlSL5WnB4l3kXAdaGXZDdknd2X4sEzu+tlV/jepos3684nTp9prek0+AcRaZq6qaj6fz+fzvb29w4PJZDIBgOl0Ph6P5/N5CMH7oLnTabboZGztYmplNYemjVzrb9acrDyHwWCQ5pltG/eCIeectCSrqNhMm9CX9cK/ge0nopaudOf1JAB4AK79dDY7PDwcra/Z3DGzgHRd9bA18/VTE9b1ZbAnvDv22HWr0J9oUd4KcffjaL23Wj1zlHz7oG0Iok1wOJal90GuBxVXehmRlabTPTL1ap9wKXMheK88RuvB19fX19c2Z7PZ22+/3SGrdBUS3ld14ZzTirpUe0lrQnw6AOQQgvemSYXW+uDNne3Nzc3RaJQNcm1zRETpINeXidpyoaljUKtjqaqum7M6mo5Ksho8UlOND4astRJC5NhxISJSP57eajQadcut+nAIYT6fX7t2rU9hnXAdDAYdRxKRLgedW/d1HxJ9GPeheCxoO4F1LCBt2054BVf0uYgo2MMJabRuQWhLORr1kJzTR3SCSal5bW0NgLQGn4hEoClwC6j8Q1NfvPcLJbetXiFqfNREZPM8d1mK1kAnMNTxrRFA1vCpTg5AI0utVJbG8aEGaM907hFGhZBlEIR1UaxzEUHq2LWkULO+69Gsk+xLAVU9dnd3u/UlIs2Cd86Vkxm0DKAvueE9jD6mHgvjdtLHAHLlYHETMsyiKYmKcxqYq+u6aYuExhhDtqHOybzoCqhU4vo6xhhv3ryp/UAasC78r4kx1hjXMwRQhGOMyMjSYHaMFEKoKm+lzXQvfe1abNKXZQTAxvzFNstoWmpuESBqPmWTYaQ/oaV8METEQYqqTFbBz+ZzIaljKOsqz9PWEAzUtuchAZM5aW2bButF1ITt+JVyci2KNG2iNbcWtv5kpWKlB4wlRt2BB3tRSOkZypr7AEd4gCpf3ejOJ6ajbAIABrCIAGRbJZSjRGFutfM8HxC13R28L8tyfDgdj8cdxlCvgYRzzoemWX5XfqeXxaida5C56jhBWZaWiEQT6kSEENtGq0swFiABJkCEjRMjbLtPaIa3Cr/RqDnfvYo+fjBMZ8XccxwMBsO1UZInhqNxtgoVIhpAIjIExhiLhIhzX1Kbh6WdN0REQw6doO3AEGM02MQSOh+QtA4s7I0FRa4mCxzjtYcjo7sPtVFFFTH98/pDVf0aAGvNvzFAWJZlgwdI+nr6b2hzqlVvUCLW+obOoO/IV10CnV9dpKlY6TMhQw57oRR7MBkzwtb6RoxRAIwxnv3uyRMflY8ioiFyzmUu6ZxBkDR+4H7JHvTc932KAQCKHhNb1zUgeu9NomUYcfFCxhhasMeOWAFAbR4lVmoL1zuvhc7ZS+M6kLbHSOd2OJbCsPWi9CEHLZM/SqlobHemP68uGb1/cYNqqim3XkUGgS5zQcPVqr4hgKbC90LLse030tQrEFlrdaYN6jD2ivAB0bRNcBrmpCIfNFGH2ZZlmSRJPRwG4RBC7b01Zn1rc2trS7u4SjtzpbCZ1IREAAxomrgqICIaaoleuigTIo7yjIJFIs1K8jHE1nRDxEBNoMoANv29mlgLdoxauZCaPd3ModWuE1zyNHUV312GQ3e3BpDvFonrM20ACN5Dj4K7e3a8cQXG1lrse5cQtPiHiNqNt1D1mPaHzUNX3HN9qu1OYhvIJrRdcgiAUfUDUXPNekVKjDbNs8BRq2A1JOxDEOaqbfTLrHmXzSSz3VFnX/btzsGgaU/UMUkdSWAyJghXZfBVFdgLIjMb20ZOkIxB09rSHnmFE2CbhtKRSAdRRHRoOnnRZ+AqO/HI8Eu93BaU1wGsz7EBgHCp2+xRAK/cp6chRyCEqC9sujJUQen/yhi3cgd9UN871oJWp9yzAKXLDe23nYOOwETEDkbDqijrumYQY8y8Kvdu3R6Pxwf7+0qSBK0xQ4SI1ye3sNcQTwcAlM02DE3gttNuHDey02Xp5tbWqbOnTp48ORgNDw73tDGMhu+MJkUAMjZlmdRGc9UmVsuvh8jNMAwdwnV2kbTdb/u4oiPyktL0zkONhZXVh56j4+hYBCHaADAiSlddjlomv9AJOkalPL8lxCUfQDdxESFaQiZgFNAqtEUpKSJ2tpPVZDCbOGPMvCz27+y9/fbbh3v7s5lHAFpueoUItTT6c/epw1rdKkXvCx2F2cRprLAMHhCTPEmyNArneW6QrLWpdc6So8ZyY4cd/HA5B69b3D4lWTnGwIU2T3gFwAAQlgmozxiOgS5AlCW/RHef/jX9QbYNC7bZyIKgjQnaLsXUQh0AINSx00zVtK1Kb9uxouggol7bsU9Q8kXNKqEOxkoFiGT39w92drbX1tbImslkcjgZM3Oe59Te2Da5J81jjD3GWS8iqiV2a9cB+M54v9EXjEnyTL2MVVVF9gbJGBOM7QCMiCsA7tax48z9kyKS0monqIZSl/ccXLyqXUqu6y3cor6m/xWDgeMATHfp6bGIqEqjVSGiVuV0AO6EMQB4DE0cU40CY4JvumNqqmXL8AUa+oZW1XDGtLJ28QLU1M60w9YBVClHRGBJkmR3dzfGGKraEBljEtPii8I4lCtksbIu/eUDALvxsIjU3td1LYTZIA0hHB4ehlgryVokQ2CRCBAAPC0qffsIJD09oL+sSZv3tGL13u1Td7/sv7l+ql17FCHQLJqf9Z9OPW18AV2RrHUCNmnuLV3AorYKpcf2URbt0TvTQJZHh7sA0G+005O7xyT/6rDGwHQ+D9eupWm6vb29vblFgPP5fGO0hkrkRJ0GBACDvAnXdDJ4ZQX75AUAhQFENNYmeWatTTKnKu5aOkREAjSACEzS+MvSloK7Kek4qnQ0I3CnbOuJvgxeeSsAqGVJi+5g1mWBdY/WfzmsPhFatXwFwPp5MD5srteSO9JbGdsG6FYoGJsi90UAuyxqLSwejyfT6VQbAVRVre1mnLOInWXYrIy69/urhF232ei3Dma6mdSMaC/P7GCQpakzFh0Za02S2ixJk8Sqp3rocmpaFxjnnCNnrSUDVVVZY51b6voNAOuJRp9UEUBiIgSyKXvVltvMU0PqEWNmLXfqWvo0MhWW0va7AyICLdfpJTADLP5FXKLLxMyakHvPp22MKYrCGGONXaT5RY7MWTbsQ7djEp3jYgXwmGYdBTeA1tcETeRARNE0WwUAORYRIADdFYpjipA5yB0NMps6uXMQZzMPCEgmREhSRAxko6CUvq59SYDOubquiawwsWcQA2CDBwjWbqxv5YN0NBolSRKjn4wPDw4OZrNJCDVwYIYYgaNmKYMIuFbhItLmKs2/w2HCzCGGEEA3AnAOrEVqG0O2XogGUbI8oUVAfmG/bmysUc851/PaLMUMujUVDkfPY5tSI70wVEPZCRBZaxLSfOkQq7KKMWpnubLwzB6bFl1ZlrrZrOju3xcc1h7DuhHRLxqIdxKaANga7SThRETtWyJyznmuRUSAiJo2HUmSoCFDFg0hok2TyWSi3XfqovZBACBNk+EwH41y77NQe2aOARhjjAEiaGSHJfhQ4uk1QARrIctpfX19a2tja2NzMMzX1obAgVlt7dBUFwHfunpLA3ZlWRZFVZZQ16BAtQ6cA2MWKjciOLsIPyhakAFs2/OoK1sRRXFlPF6c724CsNDbjYG+hZY66sykfuZCR3ArqTY1Nz03oE1tUM+XUnC3sUSHGYN8Ed3qqx3aX3oFt2DVR92JGNINeEbDNdd2bhMRIusSim2trGl3EdYtnurQdCGfzGe3b9++evXq3t7eeKI7jejGUAMyEOo6hBDZa2ZLXdckYIwJIVQV4/ldiBHYQ4yACNZBYsHapoDMIpBZLKtBOnfmQrdfkGYrZlmihKguUHVzV1WppekHe9c1i30+nxflrCgK7znGpoC400O7g3ywWJ5+eErxphVji0/dMVzLZ5TN6KeKvKP3FwRrQZvHd8JMS0I6RGzRzhBRXS2lR3UA7msD0Jf0eHw61cbGxubm5rlz586fv3Dy5MnRaKT2fVnOFgyGsJNVxpgYtRMzVMEfHh7euHFjb2/vtdduHhwcjMcHADAaDNfWh2tra1mWhlCHWM/nk8n0sKqYGVggBLCHh2AtJAkMM9I+xswhRvCt1WMWgaOIGK9ff1UXF7Xh7KKUo1kgREhTWF9PNzc3h8PhPed286E7fXa4tra2sbm2vr6uHc6MaTaaqKqi6/MWY7x951Zs9xEty7JtehU6TruSdFeUshAZtgltIWJZ3iViiFjVMptLjFp7AtaCMVBVYAw4h8YYAfKBq5qZvbVJX4fqALzwZDXZLg2A79bUdHzl4Matg9t748msCMxnz55V358xFpERWfsua8IN9OpxjDFDax0Z59zW1pbE0VV3dT4tbt6obobZxsbs3Dlz8tRAJMnSNEsHSZKMJ3vzwhcFeA944USiO6HHCCJgG2KFxDXUoJnQHQUYu7Rzg04OAKylGNn7ll1bSFNIEizm0jKAZilNo1c3kZPBIFNOoPHdnZ0t5RBdMrN2G+kyQFbNM2liwBod61Lv1G+g8SjNhtFcvskkaGc5bVukrBJa12NfbOtxkqSdwth3ttzNk2XdEV83MgAkSRKCR8Tt7e17L95z8eLFEydO5Hm+lre5Fdx0a1A3dZIkjVuQGmtCNe3Dcbhy5drzz7/4wvMvvf2WL0tYX4etLXfq1InNrbWNzZxMLKvJeLJ3cDieTgE3U8VcaFNeml50xqAWfYmAtC4tAAhNHs/yJ0JVNZxAG4V0YzBorhFZ8E8R6JRN6GEPAPTaPy/9xDnVFbAfiySijVGTNa35l5q9ZozRNkraHn8wGAwGAz2PNNArVXO+c+fOzZs3Dw8P33jjDe0yd+fOnfl8riBJkkRzKlZMUlj2RfcHS5ea02EAQ7OTEiBBmuL29tbFixc/9KEPnT17NjfYhXWZOXDnn1n2kfUysseH0+vXb7z6+pUXnn/55Zeu3roJLHBiFza30hMntzZ3RoOhAwjT2eF4fIDnd9MYI4fY7BJrwBiySM0GUsvpGYgQwfXfuxFFyO0rRpWFiEqsBMILZcouWHqndnUuT124EKWvW3WQ7kziZbMWTOsJWUGXXvcBUP+tLs9wtEHUtOTe3d09efLkiRMn1tbWLl68qFm6zLy3t/eLX/zitddeu3nz5gsvvLAC4PYpsvLc1dEAuImxiUiWJ0RQFEVZwsaGe+CBBy5evPjRBy5rdoppmkI3uwWHELQfVveURkmMc2Mcgp1O6ldefuOpp5/7yU9euH4NWIAMbG/D6TMb586f2toeMYfZbIIn1nTlWJGm4XvItguMLxqeAQB46cVZe1jWYRzSAs0BwDaes4Vm1GnF0INid3C3qpz256teM7hrUdpdYglm6fW6gyRJ8jzf3Nw8ffr0hQsXLl26dO+9925vbzPz7du3X3311RdeeOG11167detWURQaAuni0Njz/IDYLnbb5rpomxxBFCIC1HR8WFtbW19f/9QjHzt37tzZs2eTJPGh4raRCDc1hqyADyEYS1mWVf62MQ7ERbbR29u3Jj/80c9/+pMXfvzjq5UHAHAW1rbg0qUzZ86eEokdgHWqTK2yoD4qEdEIb4eqHtQj01tW7C9Zr1sTMgBQC+AOuogAeAxF9qntOAAfE9IBAJTj47t0F0xhikexqq8ViwARKntPkuTSpUv33Xffo48+evnyZWPM1atXX3zxxStXrnz961/vbC0VybqvEnDbmLSNj0OL/bGhzkaQJ0mSZcmprfXLly9/+MMf3tnaaiJgpBijEU921oqIym/nXMXXrU0QLEdHJufg7tyeXr22/+wzP37m2Z9cvRaBQPu/nj2bnTt3DnfXGxgt7LnGVCCRxhnTyB4BAPBgjyhZ7w5gWEAXNWiGiLqVRw9m8l4AvHKAcjcKvhuAA/QQrnf/5q9TF1RwqO5pDGxubly8ePGRRx75+Mc/fs8996Rp+pOf/OSJJ5748Y9/fPv2HUTQLDBnRxpQ6TRt3TtrOBzWdan+DTJNuBcRTZDz9+w8+OCDly5dOnXq1DDPAcCHKklsCAGh2bKDW9kM9ra1iYDxNRKmSTICScsapuPqiW9+71vf/N4v3prHCEiQJFBX0AAYAAj6ZNFbx2XvYFxuPQGL6NvxLNpIx6Ib3GkxoM1+6oEZ3olFH0/B8D4BLNRlC0H/oK4bc79/ngjzPA0h1HWoawCANIXhcJBl2Re/+MWHH374kUcesdb+9Kc/feKJJ5555plr125zBOecGpzKwLs0b+XbxhgkTe+NAEAISQK7W1uXL1/+yEcfPHfuXGJt7UvnHAfPomUKTTgAAMSNnUsRjI8sTGRSBOujiYGKMl69tveDHzz99a//4PAQAMBQD8BLMriF9xIRCwF0eyh2o5fK1AaqYNE2vPHldBK0g6tKhN5vBZZJamWsuI0WgHzHvU+PDqGlB3UHadrkXnV+fETdC4HVz4q9Pi+IWJYhTc358+c//elPf+Yzn7lw4UIIYX9//9/9u//r+vXr165dq6swGAySJJnP55NJNRw6ha4Kb+8b2Dvnqyogwu7u5v2X7rt8/6VzZ86sra0hCXDUNFjTL4CmiNaodCdrjDGRoQ5gbVoU0bjhZFx+/Rvf+c9f+8bVa9GYJQDrenVe8t5mni10RRat/Xprf8ySLXbn4L4Axg6uHU9ufojvBF3ooUj7b6vEwfsDcOwRdp9Re79g0dA4QMgYU5S+td2XNP/19fXYdk4fjUYPPvjg448//pGPfORDH3rgF7/4xVNPPfXUU0+/8sork8nMGEyTXGuHNJ+ZmUNgbBzm004iDLPk/D1nH3rooQcuX87yJLEGESV65qamWTik6XaMIUhNRowj6wAIgvBobX1/bxLYJOloNokvvvT61//+m9/+9murAMbOJSOardzq6G2KAi4AvATpFTgtoNVu09EZQgsL7zgt+m6jf9s+BZvjS+bvOuJx5IuIg8GgT8GtNot9haMjQWPMbDbr18sotKy1v/b4I5///Oc///nPG3Lf+973vva1//zcc89NxjNjTF2HGKP26IhNf1EKUloDiBAChABrA7h8/8X7Ll68dN+966PhaDQyKApgQmFmZ87EGFhqJo9Uow1oBCwXZb27ezLN1g8O5/NZAMheeeX1b3/r+7i70a12C4lmNksA7ogY6XiKUd14BRIASwCGlkv3H/keYXwsVOADALh7+PJ9vBdjwFrTT9VARDK4zLqbhBZN1YZeYFSvd2lg5nPnzn3pS1/6g9//b06dOvXiiy/98Ic//LP/+8+n0/l8rtvWGW2JxcwuD7FxpgEzGIDRGm6ur3/u1z9zcnfnxIkTeeoQwRCp7VTOdowB6wRdBCoYS7TRJMggwbOgA0wI8yQZTSf1229fxVNrTSN2AGVKGs9qN7pfBgbACt3eFRJ4BFwropeIYFnn0svoLiz3buB/p/PHIty7vXyP0yAuhyn7B3cLNiRpgw3GmJMnTz722GO/8zu/8+ijjx4eHv793//9X/7lX7722mvYBDkgTW0sg7WmLKMxMBhks1lZlrC5abe2Nz760Q999OEH1tcH1mE+cNZiZA9hsCASaQzl1LoQgrWWyKKQbg4XfKzrGs9sYquONimJyp+sXQYwtDD+xwFwZ1+9bwADHbvH/FHR/s4wFlmgQk9dWH3/7mAFwN1BZK/SOkYggp2djXvvvff06dP/6l/9q42NjRDC17/+9T/7sz97++0rSeLK0juAPM+rqprN2DlIU8PMMcpg6NbXB5fvv/jwxz589twpYwQwOudIMujoQaIGJ7R1kDWOyKLo7n82Bq7rGk+tq7oITWyfF7HYbubva7xfAPdPIuIHATDAURh3AO7/9h1gvALglSkcpeCjU9MDY3XrWNJYWYzq6ne7u7t/9Ed/9IUvfGF3d3c8Hn/ta1/7j//xP966dauaemshy9IY43QaECFJQLslhAA7u/TIxx966KEHT5zczrLEOWeaDTJZQHfUZl1Gi2StRSAUavaEY6jrGs/tLHz6wTeBoCTB/u4Z8H7A/B4B3Do6uqXpWPTxT7o7gHtfLLlO28/3yaJXPuEdoXvMVxi6VLW+WWWtnc2qz33u0//8n//zhx56KEmS559//m//9m+f++GPXnnlCjPs7OR1XRdF1CgGcxONXVvH+x+49MgjD1+4cD7L08RqDxNmZkC2hN1WRolW2Qhqkz9m8N7juR3wvnHWADQbNFirWyQBtD2+Fz2H302peV8A7l3Zsui7oNLdZe3yFwuu0J64S5TiXaegg5bMqmMgvXoSg7SNlrs9b/Srra2tmzdvpWnyJ3/yJ7/7u7+7tbWVZdl3vvHEV7/61R//+MeadGCtQ8T5vLYWiCAyeA/r6/ChBy9/7GMPn7/n7Ob6MM0SY0wINUswBq0hDVFYJEQjIoasNUYYvfd46SxMJlBVkCSQpqCbBC9YtLRlEU2VBAG+S23PewQwLC394uRRM/voor/T+R43XuLM780MOzqO85mvgrk/WeuWUgOgTeqeTiUE2N42RFTX/lOfeuyP//iP77vvvnMnT73yyivf+973nnjimy+88AYADAYmhIgIVQVEkGUQGayFD33o8iOPPHLh4s7m5uZwNJAYInsiIEJhjW0QiQatjbVW96GyX/7yl15++eVXXvnFfN6G+SwYA6yNp6HPnd9r+7X3OGSpHvcD36aJhrU3XcCyE6vS7S3xjjL4Hc7j0l5BDfyW37+59XwWW9pt7qDx7O2tHDWTC0zi6Lvfeebqletf/OIXP/P4p+6///5/du7s7qlTX//611966aXJeIaGmNlHsAhoLYQwnsIrr71VB6j8hfPnz589ezpJHUiTfNH2ipAoDAyIogZd5IDPfv/PvvWtb/3t3/7tiy++ojDWhEgAaLv9gwjqBnQA8Cuk4OPNpPdNwX206+68qlgtfv4+MeluUa8V8u3Nd1Ec1te0RSQEcY6cc5pEoNmln/vsp7/0pS899thjxtiXXnrpq1/96jef+Pbe3oFuU6TIgdjEfpjlk5+89KEH7//Qh+7f2Fgjg86qghwB2WBjrBtAYxwze+/xL//Dvx6O8izL/v2///d/8RdfM61XxRoIAQjBuST4OJ9FEciyhEV3EVjMdsVsOAKAoyvaeiiP02jeL4BXHoRH7Lq7AXjlfH9Gxz7u7ueXAHxM67/eZUeHMF++fPEP/uArv/Ebv7G9vX1wcPDtb3/7r//qb1566RVoqd8YsDZh5roOgzU4c2b7kx//+McffWhjY632hTW4vj6s6oKICKRzw4kgsuA/+8oj//Jf/svI/syZU6+88sq/+Tf/+uCgmk7BuSbjtVZJkA6EsSgK7fQBPd71KwFwd/D+AUw95/YyhwA4xnW6fMO7A3hVqXxnAPde4P0BuKzYWrh06Z7HH3/813/91x999NGiKJ555pm//uu/fvGFl99+e885GA4H3vu68povMBzC+fNnH/7YRz720EdOndpBYl/PiVBDTo1wBQiBvfd4bhfuv//c//K//s/OmRBqY8z/8X/+72+88cYvXp+mKRgC7yEGQAQE02Wa9cH5SwIYlmH8gQCsFyzp5Isf0vE//1UBGJZhfDdA3u382tpwMpmEyFtba4899smvfOUrDz/8MDO/8sor3/rWt7/5xLevX78JLSkTknagT1I4f377sU89+tgnPrazuxmjt0YQQdNFLKkqV8/nczyzDVkGGxuj//F/+h8effThqi7zPP2rv/qr7373+2/84urBAQxysNZMxjFGGOQ2hCYVawXG7xfAAMdISgAwdxGSd1uglQt6BLSkq//KAdwxsJXXe78AjjFqon6InOf20Ucf/YM/+L1f+7VfW19ff+21X3zjG9/4u7/7+9dfu0IEzqUxRtEtrRhcAmfObDzy8Yc++ejDF+49Z4wQSgg+hGAsJtZ572ezmc1znM2kKKb/9t/+23/xL/67z37uM4eHh3/8x3946dK9Tzzxre9/7+nZDBzyaI3qigFjX6l5p5Sz9zaWFN0PrkgfO9r0XsF3td0/8Oi9/3vtErsyigLSVPLcWkfTaXjyyacBQ1HOHn/88XPnT/3BH/xenuff+MYTb/zirbKsmYMxCUt0xBzlrbcOy/IZAHbOnDt/lixa7f3gGYEjgwDZ8aHs7GQC/oUXDv70T/90/2Dv8ccfM3brM5/99NmzZy9cuPDEN7796qu3YhBroWt2+48B2uaWvyyYj6RtA2hTkiPuTFx+ntzl/PFWw3Hv/0HWZOfEelHOJnM/GNBw3U0m/ulnf7Q/3g8SHnrooQfuf/Arf/iVwdroq3/ztz/72c9mJYwy8J7JiLXEnm/frn72sxei97/1W1/Y3tkYDnNEDCGof8U5Z7e2suvXS+vgnvPpW29Vf/qn/6GqZo98/OGLFy9euu/enZ0TW1s7T3zj22/84sp0OisKRlna1ebds0fvPn415NsYb8e+wDKw5fiwxAced3n/u3kLjp9hURQhcFmCCO+eyDY2oKr99evXv/GNvx+PD6y195y/9/HHf01Esjz9+c+enx0AAkuTEg8scP3awXT67HCYfujBBy5fvpSlTf8oZrY22v39cmMDycidO9VwCCLw//w/f/PiS8//8R//YVVV589d+PKXv3zvhfu+971/eOrJZ19//S1+H52M38ca/UrHcUTcPO8fC8YfeFSV39nZygfFdFpOp3MR8AGIqqee+jkRjUajuq4v3nvf7//+l7Wc6Qff/lmSDCNXVV1yAJdAXcOtW/Ld737fh3o0Gpw8saNCnZv+2JzF2qIDg0VdRiRAgJ/+8PW6+PPPf/5zX/7ycGd7+97zO9u//5sffuD8d77zvWeeeX5/v7QGDNFkIkCQZUkxr8kuojrSY3KG76Z8tQfNZ/N/7LVggiVl5y7Ky/H37zo9dL1zuq9WyjuP/nxJVKxI1uY+grR0fhEiIfP+KCBNk+l0BgCJsxJBIFqSuoR84J568rm6giwdra9tbu9sfuyRB9MMmcLPn3v+zu3SJFAHiB7y3EIMN27Kt771k2ImX/zSFy7ed6YsxpUf50NnjRGRGAKzsNZ2AoCP8NqrV6rqGyGEz372s6dPn11b20jTNEmywWD76aefvX5tH0nyHOoIIjHNrA9HJd8vL1D/scb7IbvjVANcmZvO/VfpymXm6RReeOGF0WgQY/zMZz+9u7srIl/6nS+GOv5o9pO6lizTXXwAAIyB6bR+7vmf5wPrw8fPnT+ZuHw+n1qTIHOEHnSFITAcTuDw+dt19c35rH788ccffPDB02fO5YPR9vbp4XD4gx88df3ajVpEY1B1Hcg2pLgkj/+xtNdfarwbUzh2rKhgPam/2PDgl2H+anXxwoREtBb29sKTTz5ZlmWauccff/zkyZO/+Zu/GaMgmOeff+HO7bkh7fcJ1oKv4c03Cl/9gMV/7tc/dfLkDlJqBSoBAQJjDSHGGDgCICQJIMKVa3tf+7tv3Lh957eL+UMPPZRk7tJ992qh9De+8c03fnHbpWiMK4poVrbw+/8Ptt2yrmi/d2Hpd/Ety6oSfbf375vFdPfnvK9xzLPKIuQ5IEJRwHPPPZdlWZIkn/zkJ4ej9c9+9rOJy5jlycMf1iWYFKwF78E4wAg3b8MzP/xJkiSf+MTHzpw9ZX0QrXQwEFlMZAAAQ8Z7zvKkrqsrb8/r+lkF28c+9pABvO/yxcj+4OCgrp+6sz+O0Q9H1odmz4aGgv8L0O6KTH2nhf9ARHx0/Er0tY5m+8VqUJaQE6QpJQnPZvzTn/50c3Mzz4YX77u0s7PzqV/75GQyrUv//PMvlqVPUlMXkRBcBnUFV6+EHzz5IyDH4OxoHeYz8B7QA1FkBkJrKKmqqiwikckHMi/8U08/Y51b21i/dP5Cno/uv//yb/32FwDgiW999/r1cm1k2mR3zWRA7Hb5/dWswl0HL4PqXSGnlTHSXtMddLraKh+4y33wbj7VD0bRPeasYzgEFpjPWftb3LlTPPPMM8z8z/7kv82ybHNz8/Of/3VnLDP/+Mc/L4qYWvC+yfXxNbz9Vunsz2bT2n7kw+euXLl24zrHAAbBEgoYjmhNCiAiwRhLJty6XT/97DPZIP3D3/8962iQjz784QdjkMPp5B++/1RVRSQNLHJvhsuR2v/Kxq+IiH/59+ivz+I4SRKWOJ3GGCHPwXt48807df3U2fPnHnvssQceePDkyZOf+cxn5vOyKIqXXnodDIYgECExYA34AG++tXc4/qn96EfuBamm49vTGRgLIBh9DFGMTdLEFeV0Mi7ykTUG7tyZPfXUMx++fNEYc+qkSRL74Icf2B8fHuyPn3/+haoGabbjRgSUbs+H/7rHf2HoLo0lSjg8rEdrZC2EAMZgmsp8Dnt7h3/9l3+1tbF97sx5yeHEyd3f+I3P7+3dPjw8vHVrT0vvkZAYUXhewLyY23vuOWFILOErL986PARCTp1hcCISvLfWCgiHaC0gwNUre3/zN3+zu3vy9OnTZVmORqNPfeqTt27dGo/Hb7x5VRt/NI5BRN0/HuL7swuP2J0NlrT7+DYRm27UIRrT1CQyiwAY02za0r9bFx3pAvgrErsOkKaou0R4H6XJLsV+HEULq7Q0lJsGuNLpaLJyx/c6judwaQreM2mRYC0AkGXADG+/vfef/tN/SpLkscceS9P0wr33/O6XvyQIf/EX/+9sWoiAAAWO1jmXYFHUNnWwsz2898LJUNdX3jqcTIGjF45EVsCARBBhBIyAAIRw5cq173znO1mW3XfpfmPMaG34yCMP13X4q7/+mxvXJ4hC2idSRDuY/KoopF+lD70YpQJM21o36p0wc33sTUS6aOnqWBvloLmKEQgJmw51GGJ/qzoBwBhF92NYCW0iwgfy2C6NpZscZ4GnKb788svf/e53NzY28jw3Zv3Eqd3Pf/7Xb9y48eyzP3rrzT3noiFbhwhCSMZaCtubuaNzBongratvHxyOwXt2zgtEAIqAACixKWSfTeOTP3j65InTZ06f06STixcvAtCLL710584PFfEhABHUNSTJ3ao7P9jk8ShFZnkmbd+rDupypDVtr8zkeAhPp0VXCuycMV2jd3AISF1eA2gLfSQyIAzIv3zYpZ3dav7X8veLRbhxwz/99NNbW1t5nn74ox/Z2dkajQaT6e/MZtNbt5+MEcgaEohBRNBi9EmSmfUsntmFEC3hW2/v790Ba4RjFIyERgSYSSICGDDx9u3ZT3/6s4sX73vooYeG62tplpw+ffKxT33i5Zdf3tubclTGYqqajcHwK/JddxBaCaoX80q/Mu2ud9577yFNV3ul6mXD0fDY+2tVZ+emr6qqrr023m1/3rirEHTbhXZv4Bbq8I7Z43ePEy+lx/QLGI8dVcl5BtNp+eSTT2ZZsr65dv/99yPKQx/76Ftvv713cPDaq2/OitommXFmNptZFB+9oNDmmjP3nkhT4wxYsz+eNHnwZCORYRBkJDR1DdbCyy+/8v3vf//s2bPbJ3YBYH1j7ZFHHn722Wd/+OyPJxPR3mnWLmb+y4+VBcK2X2hovStawhtCCAx1gMFosZ9eH8Z7+5Nj768Vt1owqOlqRJTlw9jumQEAmlBHBhBQtMmLlu0KAABzlA/gnW2tO81vfHc+H2FtPfWhfuutG8/+8Okz508OhumJEye2t9c/89lPzcriYDy9/eLV0VBcav0hW4vI0Qtg4tzu9lqeOEOYD9JXXr0+PoTZHIABIRI4EKOtUfPMHR6Uzz33s0984hNn7zlvDBqD586dffTRR1599dXDw4mzEKI3BkIIvyoZ3Dahada6ay9rqQGkiIQQ6zrUFQQP165WizXspWHY1Y2+mxECAAiAt9arxjQaceIwy3IR6bwQWrSAJHfu7BkD1pJ1RnfGQCS5u0C6OwX3yFfPLL8zAPS93NaCBPFevIdr1679wz98f21t8Ju//Vtk8L77Ls6K6vXX3rx2/WYMTeqdNSjWakkoI/rBwJ49s5nnqXPm7St3rl8rqwpEEKMIg5AxViUQ7O3t/fjHP773vksXL14IIeR5/uijH//2t79788bEWJhOVQ/8VYYbmNl70ZZ6RGwMGwNFVQEAc1Nf26H/9vZiq6W+JL5biwitANMuC0Xhp1MYj2E+n4SwoHhs97AHgCwH5yDLOANwidWO+kT0ARStu0rfZtDyxagdDp2D2ax+7rnn1tdHZ8+fue+++wej9Qc+dPm3v/jbh+P5T3/68+mkIAe2aSRAVHnv68ACeZ6kSZ5lQ8K0mL95cBCCp6A910MMAN7XaQrB86uvvnbt2rUHHrjsvScD9957787OVp4TkImH3loM4ZdWK9vBzCFIXUNdN4V7WtyBXUfBtpFikoBzsNL2ebFad4FwWdbeB2PMcDjI80h0MJ1CCDAagXPas9QSEUvQDehv3pIQNIecLQNZJLLMDPJBlI73roHHGNM0Q4shhrqG2c3w8+efO/uDM7u7u1HwxInTn/nsp69evfHWW1f3Dt621lqxSQQSsNZYIyIcmAPYCMPi8sU0z3bevHZ45VpZ7IGP0aROCihLWBuul2V57e3x09//yWMf/1Q+tMwFSvFH//S3n/qHp4sJJwCpDLwvwbw/NZrtYmONluFpW3wz93VRAiLkQ8jzXHEZrKnrUBXRRwEgFALMjKMI0RkRrH30IGhtAkDBs7WoXdURUQuBtK1Vng8FsQ7sxzUikhkmaVmWcX9KxvAgi4OBSTMUIYYUMB/k87L0432ItdvZWc8SF2JR1RNSGYC9bkPAsLB2+12IBABwEYxacpb2cHJpAYOjubBF0r158gzG+/Ov/c3fX7xw+dFHH63ms9Ew/b3f/6298a1bf/bnh4fBtmZFsycwtIWd6+vrUSiKE7sucFj52wdjVu8BEcR2t+Q7d+5cuXLl/g/doy+9sbGxvm7KIiZJ06Xg/Q7vtV8QIHadRwgRDw5qLZLTuqlOnfbRx8idi5SwCfXHGA2Bbs0HIMwMQjEyYmyNKOx2KRPB+XweY9TN17HZeEv3QOYYYT6vQ6xtAaAFHwzCECIQQFn62XwK5JDCL5/c966k3Lf6AMB7mE5r7/e/+tWvnj595p4L9x5OZ9baf/JP/snNW3f+/M//s1XYNu1nBdQJRYbSJBsOhewwHUGIt2/ervYODjl6I6Dak4iEEK5evfr8888/8OAFEDDGnDhx4vTp0zdvXElT8lXAu6WBv+MMsW0D324SpgADY8Ak4CwkSWIM6YY8k1kpDMCNbBRh5uA91H5eWSBsOuER1cK1tkzufE/adbFZ1raJ5kqGiAFABDKgyTRKkMIQYyMbnYM0daPRcDhIrIMb7S6p73fgIt76LjBWkGFb1V1VUNfy858//3d/93dr6+sXLt23ubmJ1j322CdefPFF2xIvA4vW62qftrquASDP80jk3GEIoa6BELAHYGbWbo5fiV9U14+2RP7xj65Qswe3i+/TUup6yTIzR4lRtCxd2UEI4D0URQ1Q61osJ1IzQ/AeQoyRwYfG9gC1WBFAIL4LfSz5FdR3JQIctH2LdiFqqtn0uUUJV69Mrl2bWAvGwtnT7vhb3/WJC4rsw/gdBnNAFMTWImj6tcp3v/vd3RMn/unO7mC4JiIXLlz4whe+0ABY3xrbTBQ1KEGImefzejyezGZFjGATkhARm11yRWQ+L998882iKPKBJaIkSe69917EJ7suMvF9erLYNzARgRgb3TgGsG3JnrT9Z5tOBA6MgRgackRAlsCBrdW+6c1tRYC0f2IP35QI9IbNv83muU26ESLkWTIYplluyDAiW6P2GFtwIibGGH0dYski79oJ6h1GD7R3LSHQYQwyI7Ng26sdAYhgOvUxHrz11pWXXnrpngsXd0+dvHjx4ltvvWWlebOFja2kmSRJrMJkPLt2be/69evzOSMAoRWMncefmRHhzp07t27dOnf+ZJLYGOOlS5cGA5iOxVr3gd14iB2vBmtBBIaDZteS1hGJ6roq/NRaGzxy9HUFxhAASODG7axAFWAtyiazkuesb0ioCyddzABbMpoXtbFsrKMYiLjZNDbEij2hM8a43CINECNSQJL3q0QfBe2KlF0ZbXk/MENQjUQACYoSDPHVq1ffeOON02fOpWk6Whvcc889qvJJ4/MBEBaNyZCxRVFcu3bj1VeuXr82CQGMARFst2RAAAghuMTOZrNr166dPLWVps57f++9925srB3sTbKBK4r3bQib3r5G3WwBIIRai9Bbs5ZiZGY0FqwlYSAEQjDYZDwqSyeCNAUQqkoOAsxCAM41e4tXlRIeJInJskFd12VZ6bo2ezkjpi4OB1meWxYQqJwjS+QRnMtBLLCICEuM7CVEEUjc+069e+/kCwDqS2/QsVUFUCBJwHt45ZVXTp098+GPPHR4eFgHf/n+S1YjCoiN8iGtQ76qqvF4ev36zbffnuzvgzAYg8FHlzZuQu2vn9vM+3Jvb08V7BDC7u7ucDhknlhrmSs4vm/2O0xAAEBZUN9Noe0eRZqng1rGMRrUSEi7ZwMJAnd+gyzFbDgAsYJVnHsBJMv5cJhlWV3XlZ80zekHwyzPIwhXVUPTRGQtIlrDRCgSva9Y2GAlBJEhFAHYqH/DGnTGoANEeZ/R0Wa8F9DqiDGS0f2sUCILAAEgQhQYDpPbt4unn3r24Y997MTpE2mejUZr/x/JGei5YUOxbAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=160x120 at 0x2119C0EA780>"
      ]
     },
     "execution_count": 194,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "img_name = df['cam/image_array'][150]\n",
    "img = Image.open(f'{PATH}/{img_name}')\n",
    "img"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [],
   "source": [
    "val = np.array(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(120, 160, 3)"
      ]
     },
     "execution_count": 196,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "val.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-0.4285714285714286, 0.31011686)"
      ]
     },
     "execution_count": 198,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#prediction\n",
    "run(val)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-0.42, 0.27)"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "#compare to actual data\n",
    "df['user/angle'][150], df['user/throttle'][150]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Tensorflow-GPU",
   "language": "python",
   "name": "tf-gpu"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}