CarND-P4-Behavioral_Cloning.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "# Self-Driving Car Engineer Nanodegree\n",
    "## Project 4: (Deep Learning) Driving Behavioral Cloning\n",
    "\n",
    "In this project, I used what I've learned about deep neural networks and convolutional neural networks to clone driving behavior. I trained, validated and tested a model using Keras. The model will output a steering angle to an autonomous vehicle.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.display import display_html\n",
    "display_html(\"<script>Jupyter.notebook.kernel.restart()</script>\",raw=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The Keras library is used here because it makes it super easy to arrange the necessary building blocks without spending a lot of time writing boilerplate code."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Importing useful packages\n",
    "from utils import load_dataset, plot_data_distribution, plot_training_history\n",
    "from utils import create_video, reproduce_dataset\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.utils import shuffle\n",
    "\n",
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Flatten, Dense, Lambda, Cropping2D, Dropout\n",
    "from keras.layers.convolutional import Convolution2D\n",
    "from keras.layers.pooling import MaxPooling2D\n",
    "from keras import optimizers\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import math\n",
    "import cv2\n",
    "import os"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Step 1: Dataset Loading, Summary and Exploration\n",
    "\n",
    "Te data is a dictionary with 4 key/value pairs:\n",
    "\n",
    "- `'features'` is a 4D array containing raw pixel data of car's cameras\n",
    "- `'y_train'` is a 1D array containing the stering value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Hyper Parameters\n",
    "CORRECTION = 0.25 # Correction factor for stearing angles\n",
    "DOWN_LIMIT = 20 # Down Horizontal limits to crop dataset\n",
    "BATCH_SIZE = 128 # Batch size for training\n",
    "UP_LIMIT = 65 # Up Horizontal limits to crop dataset\n",
    "EPOCHS = 20 # NUmber of epochs for traings process"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Dataset Loading and Basic Summay"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "From .... riving-data-capture/track_1(1): 6623 samples\n",
      "From .... riving-data-capture/track_1(2): 6007 samples\n",
      "From .... riving-data-capture/track_2(1): 9450 samples\n",
      "From .... riving-data-capture/track_2(2): 7027 samples\n"
     ]
    }
   ],
   "source": [
    "# Define paths \n",
    "dataset_paths = [\n",
    "    \"self-driving-data-capture/track_1(1)\",\n",
    "    \"self-driving-data-capture/track_1(2)\",\n",
    "    \"self-driving-data-capture/track_2(1)\",\n",
    "    \"self-driving-data-capture/track_2(2)\"\n",
    "    ]\n",
    "\n",
    "# Load dataset\n",
    "data = load_dataset(dataset_paths,  \"driving_log.csv\")\n",
    "data = shuffle(data)\n",
    "\n",
    "# Separate samples for training and testing process\n",
    "train_samples, validation_samples = train_test_split(data, test_size=0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of Dataset samples: 29107\n",
      "Number of training samples: 23285\n",
      "Number of testing samples: 5822\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA4UAAAFNCAYAAAC39MpQAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzt3XuYZVV55/HvL91cBlu5T4dbaIigQR0VW9Axid1qFI0jzkiUjpdGSRgvMTqaGSVqoA1EzCQaTeI4jDK0ZkKLJj4SRoegdMVcRJQoIPgALWoAEeQqjUIE3vnjrNJjUdV1qrvOOVW1v5/nOc/Ze+21137PXrVP1VtrX1JVSJIkSZK66WfGHYAkSZIkaXxMCiVJkiSpw0wKJUmSJKnDTAolSZIkqcNMCiVJkiSpw0wKJUmSJKnDTAolSSOV5INJ3jFPbf1ckq1JlrX5iSS/MR9tt/Y+k2T9fLU3h+2eluTWJN8dsH4leeSw45IkLU0mhZKkeZPkW0l+mOTuJHcm+ackr07y4983VfXqqvr9Adt61rbqVNW/VNWKqnpgHmI/NclfTGn/uVW1cUfbnmMcPwe8GTiiqn52ntte1RLI5fPZ7ri2I0maHyaFkqT59h+q6uHAwcAZwFuAD8/3RpZwwvFzwG1Vdcu4A5EkdYNJoSRpKKrqrqo6D3gJsD7JYwGSnJ3ktDa9T5Lz26ji7Un+PsnPJPkoveTob9rpof+tb/TpxCT/Alw0w4jUzye5JMn3k3wqyV5tW2uS3NAf4+RoZJJjgN8FXtK2d1lb/uPTUVtcb0/y7SS3JPlIkt3bssk41if5l3bq59tm2jdJdm/rf6+19/bW/rOAC4H9Wxxnz7D+f01yU5LvJHnVlGW/muQr7fNfn+TUvsWfb+93tvafmuTnk1yU5LYW9/9Jskdfe29JcmMb/b06yTP79sdbk3yjrXvu5L6ebjsz7QtJ0viZFEqShqqqLgFuAH5pmsVvbsv2BVbSS8yqql4O/Au9UccVVfWHfes8HfgF4DkzbPIVwKuA/YD7gfcPEOP/A/4A+Fjb3uOnqXZCe60FDgVWAH82pc4vAo8Cngn8XpJfmGGTfwrs3tp5eov5lVX1WeC5wHdaHCdMXbElsL8D/ApwGDD1FNt7Wnt7AL8KvCbJC9uyX27ve7T2vwAEeBewP739ehBwatvWo4DfAp7cRn+fA3yrtfF64IUt/v2BO4A/38Z2JEkLlEmhJGkUvgPsNU35j+glbwdX1Y+q6u+rqmZp69SquqeqfjjD8o9W1deq6h7gHcCLJ29Es4NeCrynqq6rqq3AycDxU0YpN1TVD6vqMuAy4CHJZYvleODkqrq7qr4F/DHw8gHjeDHwv/s+46n9C6tqoqquqKoHq+py4Bx6idu0qmpLVV1YVfdV1feA9/TVfwDYBTgiyU5V9a2q+kZb9mrgbVV1Q1Xd1+I4bgmf1itJS5ZJoSRpFA4Abp+m/L8DW4C/TXJdkrcO0Nb1c1j+bWAnYJ+Boty2/Vt7/W0vpzfCOan/bqE/oDeaONU+LaapbR0whzimfsYfS3J0ks3t1NS76CVvM37+JCuTbGqniH4f+IvJ+lW1BXgjvYTvllZv/7bqwcAn26m/dwJfp5dErnzIRiRJC5pJoSRpqJI8mV7C8w9Tl7WRsjdX1aHAC4A3TV6zBsw0YjjbSOJBfdM/R2808lZ6p1Xu1hfXMnqnrQ7a7nfoJUL9bd8P3DzLelPd2mKa2taNA65/Ew/9jP3+EjgPOKiqdgc+SO8UUZj+M/5BK39cVT0CeFlffarqL6vqF1u8Bby7LboeeG5V7dH32rWqbpxhO5KkBcqkUJI0FEkekeT5wCbgL6rqimnqPD/JI5MEuIveSNODbfHN9K65m6uXJTkiyW7AO4FPtEdWXAPs2m7EshPwdnqnRk66GViVvsdnTHEO8F+SHJJkBT+5BvH+uQTXYjkXOD3Jw5McDLyJ3gjdIM4FTuj7jKdMWf5w4PaqujfJUcCv9y37Hr39e+iU+luBu5IcAPzXyQVJHpXkGUl2Ae4FfshP+ueD7TMc3Orum+TYbWxHkrRAmRRKkubb3yS5m95I0tvoXaP2yhnqHgZ8ll5S8gXgA1W1uS17F/D2dnri78xh+x8FzqZ3KueuwG9D726owGuBD9EblbuH3k1uJn28vd+W5J+nafes1vbngW/SS5JeP4e4+r2+bf86eiOof9nan1VVfQb4E+AieqfeXjSlymuBd7Y++D16SeTkuj8ATgf+se3XpwAbgCPpJeX/F/jrvrZ2ofdYkVvp7c9/S+9aSoD30RuR/Nu2rYuBo7exHUnSApXZr+eXJEmSJC1VjhRKkiRJUoeZFEqSJElSh5kUSpIkSVKHmRRKkiRJUoeZFEqSJElShy0fdwDDsM8++9SqVavGHcZD3HPPPTzsYQ8bdxidZh+Mn30wfvbBeLn/x88+GD/7YLzc/+M3ij649NJLb62qfQepuySTwlWrVvHlL3953GE8xMTEBGvWrBl3GJ1mH4yffTB+9sF4uf/Hzz4YP/tgvNz/4zeKPkjy7UHrevqoJEmSJHWYSaEkSZIkdZhJoSRJkiR1mEmhJEmSJHWYSaEkSZIkdZhJoSRJkiR1mEmhJEmSJHWYSaEkSZIkdZhJoSRJkiR1mEmhJEmSJHWYSaEkSZIkddjycQcgSdJitXbj2jmvc8rBpwwhEkmStp8jhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GFDTwqTLEvylSTnt/lDknwxyZYkH0uycyvfpc1vactX9bVxciu/Oslzhh2zJEmSJHXFKEYK3wB8vW/+3cB7q+qRwB3Aia38ROCOVv7eVo8kRwDHA48BjgE+kGTZCOKWJEmSpCVvqElhkgOBXwU+1OYDPAP4RKuyEXhhmz62zdOWP7PVPxbYVFX3VdU3gS3AUcOMW5IkSZK6YtgjhX8C/DfgwTa/N3BnVd3f5m8ADmjTBwDXA7Tld7X6Py6fZh1JkiRJ0g5YPqyGkzwfuKWqLk2yZljb6dveScBJACtXrmRiYmLYm5yzrVu3Lsi4usQ+GD/7YPzsg/mzbsW6Oa/j/h8/+2D87IPxcv+P30Lrg6ElhcDTgBckeR6wK/AI4H3AHkmWt9HAA4EbW/0bgYOAG5IsB3YHbusrn9S/zo9V1ZnAmQCrV6+uNWvWDOMz7ZCJiQkWYlxdYh+Mn30wfvbB/NmwccOc1zll71Pc/2PmMTB+9sF4uf/Hb6H1wdBOH62qk6vqwKpaRe9GMRdV1UuBzcBxrdp64FNt+rw2T1t+UVVVKz++3Z30EOAw4JJhxS1JkiRJXTLMkcKZvAXYlOQ04CvAh1v5h4GPJtkC3E4vkaSqrkxyLnAVcD/wuqp6YPRhS5IkSdLSM5KksKomgIk2fR3T3D20qu4Ffm2G9U8HTh9ehJIkSZLUTaN4TqEkSZIkaYEyKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOWz7uACRJWgjWblw77hAkSRoLRwolSZIkqcNMCiVJkiSpw0wKJUmSJKnDTAolSZIkqcNMCiVJkiSpw0wKJUmSJKnDTAolSZIkqcNMCiVJkiSpw0wKJUmSJKnDTAolSZIkqcNMCiVJkiSpw0wKJUmSJKnDTAolSZIkqcOGlhQm2TXJJUkuS3Jlkg2t/JAkX0yyJcnHkuzcyndp81va8lV9bZ3cyq9O8pxhxSxJkiRJXTPMkcL7gGdU1eOBJwDHJHkK8G7gvVX1SOAO4MRW/0Tgjlb+3laPJEcAxwOPAY4BPpBk2RDjliRJkqTOGFpSWD1b2+xO7VXAM4BPtPKNwAvb9LFtnrb8mUnSyjdV1X1V9U1gC3DUsOKWJEmSpC4Z6jWFSZYl+SpwC3Ah8A3gzqq6v1W5ATigTR8AXA/Qlt8F7N1fPs06kiRJkqQdsHyYjVfVA8ATkuwBfBJ49LC2leQk4CSAlStXMjExMaxNbbetW7cuyLi6xD4YP/tg/OyD6a1bsW4k23H/j599MH72wXi5/8dvofXBUJPCSVV1Z5LNwFOBPZIsb6OBBwI3tmo3AgcBNyRZDuwO3NZXPql/nf5tnAmcCbB69epas2bNkD7N9puYmGAhxtUl9sH42QfjZx9Mb8PGDSPZzil7n+L+HzOPgfGzD8bL/T9+C60Phnn30X3bCCFJ/g3wK8DXgc3Aca3aeuBTbfq8Nk9bflFVVSs/vt2d9BDgMOCSYcUtSZIkSV0yzJHC/YCN7U6hPwOcW1XnJ7kK2JTkNOArwIdb/Q8DH02yBbid3h1Hqaork5wLXAXcD7yunZYqSZIkSdpBQ0sKq+py4InTlF/HNHcPrap7gV+boa3TgdPnO0ZJkiRJ6rqh3n1UkiRJkrSwmRRKkiRJUoeZFEqSJElSh5kUSpIkSVKHmRRKkiRJUoeZFEqSJElSh805KUyyZ5J/N4xgJEmSJEmjNVBSmGQiySOS7AX8M/C/krxnuKFJkiRJkoZt0JHC3avq+8B/Aj5SVUcDzxpeWJIkSZKkURg0KVyeZD/gxcD5Q4xHkiRJkjRCgyaF7wQuALZU1ZeSHApcO7ywJEmSJEmjsHyQSlX1ceDjffPXAS8aVlCSJEmSpNEYKClMsi/wm8Cq/nWq6lXDCUuSJEmSNAoDJYXAp4C/Bz4LPDC8cCRJkiRJozRoUrhbVb1lqJFIkiRJkkZu0BvNnJ/keUONRJIkSZI0coMmhW+glxjem+Tu9vr+MAOTJEmSJA3foHcfffiwA5EkSZIkjd6g1xSS5AXAL7fZiaryIfaSJEmStMgNdPpokjPonUJ6VXu9Icm7hhmYJEmSJGn4Bh0pfB7whKp6ECDJRuArwMnDCkySJEmSNHyD3mgGYI++6d3nOxBJkiRJ0ugNOlL4LuArSTYDoXdt4VuHFpUkSZIkaSQGvfvoOUkmgCe3ordU1XeHFpUkSZIkaSS2efpokke39yOB/YAb2mv/ViZJkiRJWsRmGyl8E3AS8MfTLCvgGfMekSRJkiRpZLaZFFbVSW3yuVV1b/+yJLsOLSpJkiRJ0kgMevfRfxqwTJIkSZK0iGxzpDDJzwIHAP8myRPp3XkU4BHAbkOOTZIkSZI0ZLNdU/gc4ATgQOA9feV3A787pJgkSVqyrrntGjZs3DCndTav3zykaCRJmv2awo3AxiQvqqq/GlFMkiRJkqQRGfTh9Y9N8piphVX1znmOR5IkSZI0QoMmhVv7pncFng98ff7DkSRJkiSN0kBJYVX91HMKk/wRcMFQIpIkSZIkjcygj6SYajd6N5+RJEmSJC1iA40UJrkCqDa7DNgX8HpCSZIkSVrkBr2m8Pl90/cDN1fV/UOIR5IkSZI0QoNeU/jtJEcCv0hvxPAfgK8MMzBJkiRJ0vANdE1hkt8DNgJ7A/sAZyd5+zADkyRJkiQN36Cnj74UeHxV3QuQ5Azgq8BpwwpMkiRJkjR8g9599Dv0nk84aRfgxvkPR5IkSZI0StscKUzyp/SuIbwLuDLJhW3+V4BLhh+eJEmSJGmYZjt99Mvt/VLgk33lE0OJRpIkSZI0UttMCqtq46gCkSRJkiSN3mynj55bVS+e8vD6H6uqfze0yCRJkiRJQzfb6aNvaO/P32YtSZIkSdKiNNvpozclWQacXVVrRxSTJEmSJGlEZn0kRVU9ADyYZPcRxCNJkiRJGqFBH16/FbiiPZLinsnCqvrtoUQlSZIkSRqJQZPCv26vfg+58YwkSZIkaXGZ9fTRZo+q2tj/Avbc1gpJDkqyOclVSa5M8oZWvleSC5Nc2973bOVJ8v4kW5JcnuTIvrbWt/rXJlm/vR9WkiRJkvTTBk0Kp0vETphlnfuBN1fVEcBTgNclOQJ4K/C5qjoM+FybB3gucFh7nQT8D+glkcApwNHAUcApk4mkJEmSJGnHzPacwnXArwOHJDmvb9EjgNu3tW5V3QTc1KbvTvJ14ADgWGBNq7YRmADe0so/UlUFXJxkjyT7tboXVtXtLaYLgWOAcwb+lJIkSZKkac12TeE/0Uvs9gH+uK/8buDyQTeSZBXwROCLwMqWMAJ8F1jZpg8Aru9b7YZWNlO5JEmSJGkHpTcwN0ul5GHAD6vqwSSHA48GPlNVPxpg3RXA3wGnV9VfJ7mzqvboW35HVe2Z5HzgjKr6h1b+OXojiGuAXavqtFb+jhbLH03Zzkn0Tjtl5cqVT9q0adMAH3+0tm7dyooVK8YdRqfZB+NnH4yffTC9a267ZiTb2WvZXtz+wDZPtnmIw/c+fEjRdJPHwPjZB+Pl/h+/UfTB2rVrL62q1YPUHfTuo58Hfqldy/e3wJeAlwAv3dZKSXYC/gr4P1U1effSm5PsV1U3tdNDb2nlNwIH9a1+YCu7kZ+cbjpZPjF1W1V1JnAmwOrVq2vNmjVTq4zdxMQECzGuLrEPxs8+GD/7YHobNm4YyXbWrVjHOVvndgXE5hdtHlI03eQxMH72wXi5/8dvofXBoDeaSVX9APhPwAeq6teAx2xzhSTAh4GvV9V7+hadx09uXLMe+FRf+SvaXUifAtzVTjO9AHh2kj1bUvrsViZJkiRJ2kGDjhQmyVPpjQye2MqWzbLO04CX03vo/Vdb2e8CZwDnJjkR+Dbw4rbs08DzgC3AD4BXAlTV7Ul+n97oJMA7J286I0mSJEnaMYMmhW8ETgY+WVVXJjkU2Oa5LO3awMyw+JnT1C/gdTO0dRZw1oCxSpIkSZIGNFBSWFV/R+9mMZPz1wG/PaygJEmSJEmjMdtzCv+kqt6Y5G+Ah9ymtKpeMLTIJEmSJElDN9tI4Ufb+x9ts5YkSZIkaVHaZlJYVZe2979Lsm+b/t4oApMkSZIkDd+sj6RIcmqSW4GrgWuSfC/J7w0/NEmSJEnSsG0zKUzyJnqPlnhyVe1VVXsCRwNPS/JfRhGgJEmSJGl4ZhspfDmwrqq+OVnQ7jz6MuAVwwxMkiRJkjR8syWFO1XVrVML23WFOw0nJEmSJEnSqMyWFP7rdi6TJEmSJC0Csz2S4vFJvj9NeYBdhxCPJEmSJGmEZnskxbJRBSJJ0nxau3HtuEOQJGlRmPWRFJIkSZKkpcukUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjrMpFCSJEmSOsykUJIkSZI6zKRQkiRJkjpsaElhkrOS3JLka31leyW5MMm17X3PVp4k70+yJcnlSY7sW2d9q39tkvXDileSJEmSumiYI4VnA8dMKXsr8LmqOgz4XJsHeC5wWHudBPwP6CWRwCnA0cBRwCmTiaQkSZIkaccNLSmsqs8Dt08pPhbY2KY3Ai/sK/9I9VwM7JFkP+A5wIVVdXtV3QFcyEMTTUmSJEnSdhr1NYUrq+qmNv1dYGWbPgC4vq/eDa1spnJJkiRJ0jxIVQ2v8WQVcH5VPbbN31lVe/Qtv6Oq9kxyPnBGVf1DK/8c8BZgDbBrVZ3Wyt8B/LCq/miabZ1E79RTVq5c+aRNmzYN7XNtr61bt7JixYpxh9Fp9sH42Qfj15U+uOa2a8YdwrT2WrYXtz8w9USabTt878OHFE03deUYWMjsg/Fy/4/fKPpg7dq1l1bV6kHqLh9qJA91c5L9quqmdnroLa38RuCgvnoHtrIb6SWG/eUT0zVcVWcCZwKsXr261qxZM121sZqYmGAhxtUl9sH42Qfj15U+2LBxw7hDmNa6Fes4Z+s5c1pn84s2DymaburKMbCQ2Qfj5f4fv4XWB6NOCs8D1gNntPdP9ZX/VpJN9G4qc1dLHC8A/qDv5jLPBk4eccySpDFbu3HtuEOQJGnJGlpSmOQceqN8+yS5gd5dRM8Azk1yIvBt4MWt+qeB5wFbgB8ArwSoqtuT/D7wpVbvnVU1t3NuJEmSJEkzGlpSWFXrZlj0zGnqFvC6Gdo5CzhrHkOTJEmSJDWjvvuoJEmSJGkBMSmUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4b9XMKJUkd5zMH52579tnm9T7wXpI0GEcKJUmSJKnDTAolSZIkqcM8fVSStN08FVSaO08HlrTQOFIoSZIkSR3mSKEkLUHbGolYt2IdGzZueEi5IxGSJHWTI4WSJEmS1GEmhZIkSZLUYSaFkiRJktRhJoWSJEmS1GEmhZIkSZLUYSaFkiRJktRhPpJCkiRpO23Pg+glaaExKZSk7bQ9fwz6LEBpdGY6Rn1WpyT9NJNCSZK04DkiJ0nDY1IoSY1/dEqSpC4yKZSkBW5UyapJsbrOY0BSV5kUSpKk7ea1tQuXfSNpUD6SQpIkSZI6zKRQkiRJkjrM00clSVqCPHVQkjQok0JJkjRS3tBFkhYWTx+VJEmSpA5zpFCSJAGO4ElSV5kUSpIkLXAL+XmlXosqLX6ePipJkiRJHeZIoSSNkKfnSZKkhcaRQkmSJEnqMJNCSZIkSeowk0JJkiRJ6jCvKZS04HkdniRJ0vA4UihJkiRJHWZSKEmSJEkdZlIoSZIkSR3mNYWSRuqa265hw8YN4w5DkiRJjSOFkiRJktRhjhRKkiRppEZ11sjm9ZuHvg1pKTAplCRJ0nbbnscGrVuxbgiRSNpeJoXSErQ9v6D9b6okSVI3mRRqwZvPB5evW7FuxtNVup4UjeoB8f53WJIkaWExKZSauSZFXU8iJUlaajzTRl1lUihtp1H94hjVCJ4kSZK6yaRQGiETPEmSlhZHF7UU+JxCSZIkSeowRwolSZK0JC2lM3S88Z6GadEkhUmOAd4HLAM+VFVnjDmkOdueB7V6YEqSJC0tSylZ1dKwKJLCJMuAPwd+BbgB+FKS86rqqvFG1m1+oUmSJC0to/r7zoGPhWVRJIXAUcCWqroOIMkm4FhgySeFJl6SJEmShmmxJIUHANf3zd8AHD2mWCRJkiTtgKU08LEURj1TVeOOYVZJjgOOqarfaPMvB46uqt/qq3MScFKbfRRw9cgDnd0+wK3jDqLj7IPxsw/Gzz4YL/f/+NkH42cfjJf7f/xG0QcHV9W+g1RcLCOFNwIH9c0f2Mp+rKrOBM4cZVBzleTLVbV63HF0mX0wfvbB+NkH4+X+Hz/7YPzsg/Fy/4/fQuuDxfKcwi8BhyU5JMnOwPHAeWOOSZIkSZIWvUUxUlhV9yf5LeACeo+kOKuqrhxzWJIkSZK06C2KpBCgqj4NfHrcceygBX16a0fYB+NnH4yffTBe7v/xsw/Gzz4YL/f/+C2oPlgUN5qRJEmSJA3HYrmmUJIkSZI0BCaF8yzJryW5MsmDSWa8o1CSY5JcnWRLkrf2lR+S5Iut/GPtxjqagyR7JbkwybXtfc9p6qxN8tW+171JXtiWnZ3km33LnjD6T7G4DdIHrd4Dffv5vL5yj4MdMOAx8IQkX2jfV5cneUnfMo+B7TTTd3vf8l3az/SW9jO+qm/Zya386iTPGWXcS8UA+/9NSa5qP/OfS3Jw37Jpv480NwP0wQlJvte3r3+jb9n69r11bZL1o4186RigD97bt/+vSXJn3zKPgx2U5KwktyT52gzLk+T9rX8uT3Jk37LxHQNV5WseX8Av0HtO4gSweoY6y4BvAIcCOwOXAUe0ZecCx7fpDwKvGfdnWmwv4A+Bt7bptwLvnqX+XsDtwG5t/mzguHF/jsX8GrQPgK0zlHscDHn/A4cDh7Xp/YGbgD3avMfA9u33Gb/b++q8Fvhgmz4e+FibPqLV3wU4pLWzbNyfaTG9Btz/a/u+618zuf/b/LTfR77mvQ9OAP5smnX3Aq5r73u26T3H/ZkW22uQPphS//X0buA4Oe9xsON98MvAkcDXZlj+POAzQICnAF9s5WM9BhwpnGdV9fWqunqWakcBW6rquqr6V2ATcGySAM8APtHqbQReOLxol6xj6e07GGwfHgd8pqp+MNSoumWuffBjHgfzYtb9X1XXVNW1bfo7wC3AQA+41Yym/W6fUqe/bz4BPLP9zB8LbKqq+6rqm8CW1p4GN+v+r6rNfd/1F9N77rHmzyDHwEyeA1xYVbdX1R3AhcAxQ4pzKZtrH6wDzhlJZB1RVZ+nN9gwk2OBj1TPxcAeSfZjzMeASeF4HABc3zd/QyvbG7izqu6fUq65WVlVN7Xp7wIrZ6l/PA/9Qjy9Dem/N8ku8x7h0jdoH+ya5MtJLp48fRePg/kwp2MgyVH0/qP8jb5ij4G5m+m7fdo67Wf8Lno/84Osq22b6z48kd5/6ydN932kuRm0D17Uvl8+keSgOa6rbRt4P7bTpw8BLuor9jgYvpn6aKzHwKJ5JMVCkuSzwM9Os+htVfWpUcfTRdvqg/6ZqqokM95it/1n5nFj55taAAAHGUlEQVT0noE56WR6f0jvTO92wW8B3rmjMS8189QHB1fVjUkOBS5KcgW9P5I1i3k+Bj4KrK+qB1uxx4CWtCQvA1YDT+8rfsj3UVV9Y/oWtAP+Bjinqu5L8p/pjZw/Y8wxddXxwCeq6oG+Mo+DjjIp3A5V9awdbOJG4KC++QNb2W30hpCXt/8gT5Zrim31QZKbk+xXVTe1P3hv2UZTLwY+WVU/6mt7coTlviT/G/ideQl6iZmPPqiqG9v7dUkmgCcCf4XHwazmY/8neQTwf+n9Q+vivrY9BrbPTN/t09W5IclyYHd63/2DrKttG2gfJnkWvX+ePL2q7pssn+H7yD+G52bWPqiq2/pmP0TvGujJdddMWXdi3iNc+ubyXXI88Lr+Ao+DkZipj8Z6DHj66Hh8CTgsvTss7kzvoDyveleZbqZ3jRvAesCRx7k7j96+g9n34UPOpW9/RE9e2/ZCYNq7R2mbZu2DJHtOnpaYZB/gacBVHgfzYpD9vzPwSXrXNXxiyjKPge0z7Xf7lDr9fXMccFH7mT8POD69u5MeAhwGXDKiuJeKWfd/kicC/xN4QVXd0lc+7ffRyCJfOgbpg/36Zl8AfL1NXwA8u/XFnsCz+emzeDSYQb6HSPJoejcz+UJfmcfBaJwHvKLdhfQpwF3tn7HjPQZGdUebrryA/0jvHOD7gJuBC1r5/sCn++o9D7iG3n9f3tZXfii9PwS2AB8Hdhn3Z1psL3rX53wOuBb4LLBXK18NfKiv3ip6/5X5mSnrXwRcQe8P4b8AVoz7My221yB9APz7tp8va+8n9q3vcTD8/f8y4EfAV/teT2jLPAa2f98/5Lud3qm3L2jTu7af6S3tZ/zQvnXf1ta7GnjuuD/LYnwNsP8/2343T/7Mn9fKZ/w+8jXvffAu4Mq2rzcDj+5b91Xt2NgCvHLcn2WxvmbrgzZ/KnDGlPU8DuZn/59D747eP6KXE5wIvBp4dVse4M9b/1xB39MKxnkMpAUgSZIkSeogTx+VJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCQtWkneluTKJJcn+WqSo1v5G5PsNo/beXWSV8xje/sk+VGSV+9gO6uS+BxJSdIO8ZEUkqRFKclTgfcAa6rqvvaw5Z2r6jtJvkXv2U+3zsN2llfV/TvazpQ2XwP8OvBgVT19B9pZBZxfVY+dp9AkSR3kSKEkabHaD7i1qu4DqKpbW0L428D+wOYkmwGSPDvJF5L8c5KPJ1nRyp+U5O+SXJrkgiT7tfKJJH+S5MvAG5KcmuR3+pa9O8klSa5J8kutfLck5ya5Ksknk3wxyeoZYl8HvBk4IMmBk4VJtiY5PcllSS5OsrKV/3ybvyLJaUm2Tm0wybIk/z3Jl9rI6X9u5fsl+XwbSf3aZLySJE0yKZQkLVZ/CxzUErMPJHk6QFW9H/gOsLaq1rYRxLcDz6qqI4EvA29KshPwp8BxVfUk4Czg9L72d66q1VX1x9Nse3lVHQW8ETillb0WuKOqjgDeATxpuqCTHATsV1WXAOcCL+lb/DDg4qp6PPB54Ddb+fuA91XV44AbZtgfJwJ3VdWTgScDv5nkEHojkhdU1ROAxwNfnWF9SVJHmRRKkhalqtpKL/E6Cfge8LEkJ0xT9SnAEcA/JvkqsB44GHgU8Fjgwlb+duDAvvU+to3N/3V7vxRY1aZ/EdjUYvsacPkM676EXjJIq7+ub9m/AudP0/ZTgY+36b+cod1nA69on+WLwN7AYcCXgFcmORV4XFXdvY3PJUnqoOXjDkCSpO1VVQ8AE8BEkivoJXxnT6kW4MKqWvdThcnjgCur6qkzNH/PNjZ9X3t/gLn/Ll0H/GySl7b5/ZMcVlXXAj+qn1zsP9e2A7y+qi54yILkl4FfBc5O8p6q+sgcY5YkLWGOFEqSFqUkj0pyWF/RE4Bvt+m7gYe36YuBpyV5ZFvvYUkOB64G9m03rCHJTkkeswMh/SPw4tbWEcDjpon5cGBFVR1QVauqahXwLn56tHA6FwMvatPHz1DnAuA17bRYkhzePuvBwM1V9b+ADwFHzu1jSZKWOpNCSdJitQLY2G7scjm9U0RPbcvOBP5fks1V9T3gBOCcVu8LwKOr6l+B44B3J7mM3rV2/34H4vkAvSTzKuA04Ergril11gGfnFL2V8yeFL6R3nWQlwOPnKZd6CV8VwH/3B5T8T/pjTSuAS5L8hV6p66+b9APJEnqBh9JIUnSPEiyDNipqu5N8vPAZ4FHteRzR9veDfhhVVWS44F1VXXsjrYrSRJ4TaEkSfNlN3qPwdiJ3vV9r52PhLB5EvBnSQLcCbxqntqVJMmRQkmSJEnqMq8plCRJkqQOMymUJEmSpA4zKZQkSZKkDjMplCRJkqQOMymUJEmSpA4zKZQkSZKkDvv/aNjSMa9oTfQAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Print dataset information\n",
    "print(\"Number of Dataset samples: {}\".format(len(data))) # Number of training examples\n",
    "print(\"Number of training samples: {}\".format(len(train_samples))) # Number of training examples\n",
    "print(\"Number of testing samples: {}\".format(len(validation_samples))) # Number of testing examples\n",
    "\n",
    "# Plot dataset distribution\n",
    "y_data_raw = np.asarray([data[idx][\"steering\"] for idx in range(len(data))])\n",
    "plot_data_distribution(y_data_raw, scfc=30, graph_name=\"Distribution of dataset\", save_name=\"writeup_files/dataset_stee_angles.png\")\n",
    "\n",
    "# Plot dataset distribution\n",
    "y_train_raw = np.asarray([train_samples[idx][\"steering\"] for idx in range(len(train_samples))])\n",
    "plot_data_distribution(y_train_raw, scfc=30, graph_name=\"Distribution of training set\",  save_name=\"writeup_files/training_stee_angles.png\")\n",
    "\n",
    "# Plot dataset distribution\n",
    "y_test_raw = np.asarray([validation_samples[idx][\"steering\"] for idx in range(len(validation_samples))])\n",
    "plot_data_distribution(y_test_raw, scfc=30, graph_name=\"Distribution of testing set\",  save_name=\"writeup_files/testing_stee_angles.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generator(samples, batch_size=32, angle_fc = 0.2):\n",
    "    \n",
    "    batch_size_org = batch_size\n",
    "    batch_size = int(np.ceil(batch_size/6))\n",
    "    \n",
    "    while 1: # Loop forever so the generator never terminates\n",
    "        samples = shuffle(samples)\n",
    "        \n",
    "        for offset in range(0, len(samples), batch_size):\n",
    "            batch_samples = samples[offset:offset+batch_size]\n",
    "            images = []; angles = []\n",
    "            \n",
    "            # Read image and steering angle\n",
    "            for batch_sample in batch_samples:\n",
    "                                                    \n",
    "                # Central camera\n",
    "                img_c = cv2.cvtColor(cv2.imread(batch_sample[\"img_c\"]), cv2.COLOR_BGR2RGB) \n",
    "                ang_c =  batch_sample[\"steering\"]\n",
    "                \n",
    "                # Left camera\n",
    "                img_l = cv2.cvtColor(cv2.imread(batch_sample[\"img_l\"]), cv2.COLOR_BGR2RGB)\n",
    "                ang_l =  batch_sample[\"steering\"] + angle_fc\n",
    "\n",
    "                # Right camera\n",
    "                img_r = cv2.cvtColor(cv2.imread(batch_sample[\"img_r\"]), cv2.COLOR_BGR2RGB)\n",
    "                ang_r =  batch_sample[\"steering\"] - angle_fc\n",
    "\n",
    "                # Fliped Left camera\n",
    "                img_lf = cv2.flip(img_l, 1)\n",
    "                ang_lf = -ang_l\n",
    "\n",
    "                # Fliped Right camera\n",
    "                img_rf = cv2.flip(img_r, 1)\n",
    "                ang_rf = -ang_r\n",
    "\n",
    "                # Fliped center camera\n",
    "                img_cf = cv2.flip(img_c, 1)\n",
    "                ang_cf = -ang_c \n",
    "\n",
    "                images.extend((img_c, img_l, img_r, img_cf, img_lf, img_rf))\n",
    "                angles.extend((ang_c, ang_l, ang_r, ang_cf, ang_lf, ang_rf))\n",
    "                      \n",
    "            images, angles = shuffle(images, angles)\n",
    "                    \n",
    "            # trim image to only see section with road\n",
    "            X_train = np.array(images[:32])\n",
    "            y_train = np.array(angles[:32])\n",
    "                        \n",
    "            # Return batch\n",
    "            yield shuffle(X_train, y_train)\n",
    "            "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create data generators\n",
    "train_generator = generator(train_samples, BATCH_SIZE, CORRECTION)\n",
    "validation_generator = generator(validation_samples, BATCH_SIZE, CORRECTION)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Exploratory Visualization of the Dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create videos with current datasets\n",
    "path_idx = 0\n",
    "for cam_label in [\"img_c\", \"img_l\", \"img_r\"]:\n",
    "    create_video(cam_label=cam_label, \n",
    "                 dst_path=\"model/video_results\", \n",
    "                 src_path=dataset_paths[path_idx],\n",
    "                 file_name=\"{}_{}.mp4\".format(dataset_paths[path_idx].split(\"/\")[-1], cam_label),\n",
    "                 video_size=(320, 160),\n",
    "                 fps=15.\n",
    "                 )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "From .... riving-data-capture/track_2(1): 9450 samples\n"
     ]
    }
   ],
   "source": [
    "# Concatenate cameras videos and reproduce  -> A: Stop video - Q: Quit\n",
    "reproduce_dataset(fps=30, loop=True, sc_fc=1.5, up_limit=UP_LIMIT, down_limit=DOWN_LIMIT, \n",
    "                  dataset_path = dataset_paths[2], CORRECTION=CORRECTION)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Step 2: Designing, Training and Testing a Model Architecture"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Training"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:10: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(24, (5, 5), strides=(2, 2), activation=\"relu\")`\n",
      "  # Remove the CWD from sys.path while we load stuff.\n",
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:13: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(36, (5, 5), strides=(2, 2), activation=\"relu\")`\n",
      "  del sys.path[0]\n",
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:16: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(48, (5, 5), strides=(2, 2), activation=\"relu\")`\n",
      "  app.launch_new_instance()\n",
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:19: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3), activation=\"relu\")`\n",
      "/usr/local/lib/python3.5/dist-packages/ipykernel_launcher.py:22: UserWarning: Update your `Conv2D` call to the Keras 2 API: `Conv2D(64, (3, 3), activation=\"relu\")`\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/20\n",
      "1092/1092 [==============================] - 194s 178ms/step - loss: 0.0652 - val_loss: 0.0435\n",
      "Epoch 2/20\n",
      "1092/1092 [==============================] - 76s 69ms/step - loss: 0.0405 - val_loss: 0.0352\n",
      "Epoch 3/20\n",
      "1092/1092 [==============================] - 76s 69ms/step - loss: 0.0346 - val_loss: 0.0310\n",
      "Epoch 4/20\n",
      "1092/1092 [==============================] - 77s 71ms/step - loss: 0.0314 - val_loss: 0.0264\n",
      "Epoch 5/20\n",
      "1092/1092 [==============================] - 98s 89ms/step - loss: 0.0292 - val_loss: 0.0246\n",
      "Epoch 6/20\n",
      "1092/1092 [==============================] - 85s 78ms/step - loss: 0.0276 - val_loss: 0.0332\n",
      "Epoch 7/20\n",
      "1092/1092 [==============================] - 75s 69ms/step - loss: 0.0271 - val_loss: 0.0255\n",
      "Epoch 8/20\n",
      "1092/1092 [==============================] - 74s 68ms/step - loss: 0.0253 - val_loss: 0.0234\n",
      "Epoch 9/20\n",
      "1092/1092 [==============================] - 73s 67ms/step - loss: 0.0242 - val_loss: 0.0233\n",
      "Epoch 10/20\n",
      "1092/1092 [==============================] - 86s 79ms/step - loss: 0.0242 - val_loss: 0.0273\n",
      "Epoch 11/20\n",
      "1092/1092 [==============================] - 83s 76ms/step - loss: 0.0232 - val_loss: 0.0202\n",
      "Epoch 12/20\n",
      "1092/1092 [==============================] - 123s 113ms/step - loss: 0.0225 - val_loss: 0.0216\n",
      "Epoch 13/20\n",
      "1092/1092 [==============================] - 122s 112ms/step - loss: 0.0224 - val_loss: 0.0210\n",
      "Epoch 14/20\n",
      "1092/1092 [==============================] - 123s 113ms/step - loss: 0.0214 - val_loss: 0.0218\n",
      "Epoch 15/20\n",
      "1092/1092 [==============================] - 87s 80ms/step - loss: 0.0213 - val_loss: 0.0218\n",
      "Epoch 16/20\n",
      "1092/1092 [==============================] - 91s 83ms/step - loss: 0.0208 - val_loss: 0.0207\n",
      "Epoch 17/20\n",
      "1092/1092 [==============================] - 143s 131ms/step - loss: 0.0205 - val_loss: 0.0196\n",
      "Epoch 18/20\n",
      "1092/1092 [==============================] - 190s 174ms/step - loss: 0.0201 - val_loss: 0.0201\n",
      "Epoch 19/20\n",
      "1092/1092 [==============================] - 158s 144ms/step - loss: 0.0198 - val_loss: 0.0228\n",
      "Epoch 20/20\n",
      "1092/1092 [==============================] - 142s 130ms/step - loss: 0.0192 - val_loss: 0.0176\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "\n",
    "# # Preprocess incoming data, centered around zero with small standard deviation \n",
    "model.add(Lambda(lambda x: (x/255.) - 0.5, input_shape=(160, 320, 3)))\n",
    "\n",
    "# # trim image to only see section with road\n",
    "model.add(Cropping2D(cropping=((DOWN_LIMIT, UP_LIMIT),(0, 0))))\n",
    "\n",
    "#layer 1- Convolution, no of filters- 24, filter size= 5x5, stride= 2x2\n",
    "model.add(Convolution2D(24, 5, 5, subsample=(2,2), activation=\"relu\"))\n",
    "\n",
    "#layer 2- Convolution, no of filters- 36, filter size= 5x5, stride= 2x2\n",
    "model.add(Convolution2D(36, 5, 5, subsample=(2,2), activation=\"relu\"))\n",
    "\n",
    "#layer 3- Convolution, no of filters- 48, filter size= 5x5, stride= 2x2\n",
    "model.add(Convolution2D(48, 5, 5, subsample=(2,2), activation=\"relu\"))\n",
    "\n",
    "#layer 4- Convolution, no of filters- 64, filter size= 3x3, stride= 1x1\n",
    "model.add(Convolution2D(64, 3, 3, activation=\"relu\"))\n",
    "\n",
    "#layer 5- Convolution, no of filters- 64, filter size= 3x3, stride= 1x1\n",
    "model.add(Convolution2D(64, 3, 3, activation=\"relu\"))\n",
    "\n",
    "# model.add(MaxPooling2D())\n",
    "\n",
    "#flatten image from 2D to side by side\n",
    "model.add(Flatten())\n",
    "\n",
    "#layer 6- fully connected layer 1\n",
    "model.add(Dense(100, activation=\"relu\"))\n",
    "\n",
    "#Adding a dropout layer to avoid overfitting. Here we are have given the dropout \n",
    "# rate as 25% after first fully connected layer\n",
    "model.add(Dropout(0.25))\n",
    "\n",
    "#layer 7- fully connected layer 1\n",
    "model.add(Dense(50, activation=\"relu\"))\n",
    "\n",
    "#layer 8- fully connected layer 1\n",
    "model.add(Dense(10, activation=\"relu\"))\n",
    "\n",
    "#layer 9- fully connected layer 1\n",
    "#here the final layer will contain one value as this is a regression problem and not classification\n",
    "model.add(Dense(1))\n",
    "\n",
    "# the output is the steering angle using mean squared error loss function is the right choice \n",
    "# for this regression problem adam optimizer is used here\n",
    "optimizer = optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)\n",
    "model.compile(loss=\"mse\", optimizer=optimizer)\n",
    "history_object  = model.fit_generator(\n",
    "        train_generator, \n",
    "        steps_per_epoch=math.ceil(len(train_samples*6)/BATCH_SIZE), \n",
    "        validation_data=validation_generator, \n",
    "        validation_steps=math.ceil(len(validation_samples)/BATCH_SIZE), \n",
    "        epochs=EPOCHS, \n",
    "        verbose=1)\n",
    "\n",
    "# Save the model whe n finish\n",
    "model.save('model/behavioral_model.h5')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Plot Testing Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lambda_1 (Lambda)            (None, 160, 320, 3)       0         \n",
      "_________________________________________________________________\n",
      "cropping2d_1 (Cropping2D)    (None, 75, 320, 3)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 36, 158, 24)       1824      \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 16, 77, 36)        21636     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 6, 37, 48)         43248     \n",
      "_________________________________________________________________\n",
      "conv2d_4 (Conv2D)            (None, 4, 35, 64)         27712     \n",
      "_________________________________________________________________\n",
      "conv2d_5 (Conv2D)            (None, 2, 33, 64)         36928     \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 4224)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 100)               422500    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 50)                5050      \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 10)                510       \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 1)                 11        \n",
      "=================================================================\n",
      "Total params: 559,419\n",
      "Trainable params: 559,419\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 936x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# keras method to print the model summary\n",
    "model.summary()\n",
    "\n",
    "# Plot training history\n",
    "plot_training_history(history_object, fig_size=(13,8), save_name=\"writeup_files/training_loss_graph.png\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Run Simulation Enviroment, Model and Other Utils\n",
    "NOTE: If you recently trained a model you must restart the kernel to free memory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.display import display_html\n",
    "display_html(\"<script>Jupyter.notebook.kernel.restart()</script>\",raw=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from utils import setClipboardData, create_model_result_video\n",
    "from IPython.display import clear_output\n",
    "from multiprocessing import Pool \n",
    "import os\n",
    "\n",
    "def run_process(process):                                                             \n",
    "    os.system('{}'.format(process)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ffmpeg -video_size 1000x650 -framerate 25 -f x11grab -i :0.0+100,200 /home/kiwivision/Documents/John_Udacity_Nanodegree_Selfdrivingcars/CarND-P4-Behavioral_Cloning/model/video_results/behavioral_driving_stg1_test1.mp4\n"
     ]
    }
   ],
   "source": [
    "# Create Video from desktop\n",
    "video_name = \"behavioral_driving_stg{}_test{}.mp4\".format(1, 1)\n",
    "command = \"ffmpeg -video_size {}x{} -framerate 25 -f x11grab -i :0.0+100,200 {}\".format(\n",
    "    1000, 650, os.path.join(os.getcwd(), \"model/video_results\", video_name))\n",
    "print(command); setClipboardData(command.encode())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create gif from videovideo_name, \n",
    "video_name = \"stage_1_03312019_img_r\"\n",
    "command = \"ffmpeg -i {}.mp4 -ss 15 -t 20 {}.gif -hide_banner\".format(video_name, video_name)\n",
    "print(command); setClipboardData(command.encode())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "clear && /home/kiwivision/Documents/John_Udacity_Nanodegree_Selfdrivingcars/CarND-P4-Behavioral_Cloning/beta_simulator_linux/beta_simulator.x86_64\n",
      "clear && termdown 10 && clear && python3 /home/kiwivision/Documents/John_Udacity_Nanodegree_Selfdrivingcars/CarND-P4-Behavioral_Cloning/drive.py /home/kiwivision/Documents/John_Udacity_Nanodegree_Selfdrivingcars/CarND-P4-Behavioral_Cloning/model/behavioral_model.h5 /home/kiwivision/Documents/John_Udacity_Nanodegree_Selfdrivingcars/CarND-P4-Behavioral_Cloning/model/image_results \n"
     ]
    }
   ],
   "source": [
    "# Run simulator and socket to drive autonomously\n",
    "try:\n",
    "    time_for_server = 10\n",
    "    processes = (\n",
    "        \"clear && {}\".format(os.path.join(os.getcwd(), \"beta_simulator_linux\", \"beta_simulator.x86_64\")),\n",
    "        \"clear && termdown {} && clear && python3 {} {} {}\".format(\n",
    "            time_for_server,\n",
    "            os.path.join(os.getcwd(), \"drive.py\"), \n",
    "            os.path.join(os.getcwd(), \"model\", \"behavioral_model.h5\"), \n",
    "            os.path.join(os.getcwd(), \"model\", \"image_results \")))\n",
    "    print(processes[0])\n",
    "    print(processes[1])\n",
    "    pool = Pool(processes=len(processes))                                                        \n",
    "    pool.map(run_process, processes)\n",
    "except:\n",
    "    print(\"Process has finished\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Run only simulation environment\n",
    "command = os.path.join(os.getcwd(), \"beta_simulator_linux\", \"beta_simulator.x86_64\")\n",
    "os.system('{}'.format(command)) \n",
    "print(command)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Create model's results video   \n",
    "create_model_result_video(src_path = os.path.join(os.getcwd(), \"model\", \"image_results\"), \n",
    "                          dst_path = os.path.join(os.getcwd(), \"model\", \"video_results\"),\n",
    "                          video_size = (360, 160))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}