Traffic_Sign_Classifier.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Self-Driving Car Engineer Nanodegree\n",
    "\n",
    "## Deep Learning\n",
    "\n",
    "## Project: Build a Traffic Sign Recognition Classifier\n",
    "\n",
    "In this notebook, a template is provided for you to implement your functionality in stages, which is required to successfully complete this project. If additional code is required that cannot be included in the notebook, be sure that the Python code is successfully imported and included in your submission if necessary. \n",
    "\n",
    "> **Note**: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to  \\n\",\n",
    "    \"**File -> Download as -> HTML (.html)**. Include the finished document along with this notebook as your submission. \n",
    "\n",
    "In addition to implementing code, there is a writeup to complete. The writeup should be completed in a separate file, which can be either a markdown file or a pdf document. There is a [write up template](https://github.com/udacity/CarND-Traffic-Sign-Classifier-Project/blob/master/writeup_template.md) that can be used to guide the writing process. Completing the code template and writeup template will cover all of the [rubric points](https://review.udacity.com/#!/rubrics/481/view) for this project.\n",
    "\n",
    "The [rubric](https://review.udacity.com/#!/rubrics/481/view) contains \"Stand Out Suggestions\" for enhancing the project beyond the minimum requirements. The stand out suggestions are optional. If you decide to pursue the \"stand out suggestions\", you can include the code in this Ipython notebook and also discuss the results in the writeup file.\n",
    "\n",
    "\n",
    ">**Note:** Code and Markdown cells can be executed using the **Shift + Enter** keyboard shortcut. In addition, Markdown cells can be edited by typically double-clicking the cell to enter edit mode."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "## Step 0: Load The Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "numpy.ndarray"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load pickled data\n",
    "import pickle\n",
    "\n",
    "# TODO: Fill this in based on where you saved the training and testing data\n",
    "dataset_dir = './traffic-signs-data/'\n",
    "training_file = dataset_dir + 'train.p'\n",
    "validation_file= dataset_dir + 'valid.p'\n",
    "testing_file = dataset_dir + 'test.p'\n",
    "\n",
    "with open(training_file, mode='rb') as f:\n",
    "    train = pickle.load(f)\n",
    "with open(validation_file, mode='rb') as f:\n",
    "    valid = pickle.load(f)\n",
    "with open(testing_file, mode='rb') as f:\n",
    "    test = pickle.load(f)\n",
    "    \n",
    "X_train, y_train = train['features'], train['labels']\n",
    "X_valid, y_valid = valid['features'], valid['labels']\n",
    "X_test, y_test = test['features'], test['labels']\n",
    "X_test.shape\n",
    "type(X_train[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "## Step 1: Dataset Summary & Exploration\n",
    "\n",
    "The pickled data is a dictionary with 4 key/value pairs:\n",
    "\n",
    "- `'features'` is a 4D array containing raw pixel data of the traffic sign images, (num examples, width, height, channels).\n",
    "- `'labels'` is a 1D array containing the label/class id of the traffic sign. The file `signnames.csv` contains id -> name mappings for each id.\n",
    "- `'sizes'` is a list containing tuples, (width, height) representing the original width and height the image.\n",
    "- `'coords'` is a list containing tuples, (x1, y1, x2, y2) representing coordinates of a bounding box around the sign in the image. **THESE COORDINATES ASSUME THE ORIGINAL IMAGE. THE PICKLED DATA CONTAINS RESIZED VERSIONS (32 by 32) OF THESE IMAGES**\n",
    "\n",
    "Complete the basic data summary below. Use python, numpy and/or pandas methods to calculate the data summary rather than hard coding the results. For example, the [pandas shape method](http://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.shape.html) might be useful for calculating some of the summary results. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Provide a Basic Summary of the Data Set Using Python, Numpy and/or Pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training examples = 104397\n",
      "Number of validation examples = 4410\n",
      "Number of testing examples = 12630\n",
      "Image data shape = (32, 32, 3)\n",
      "Number of classes = 43\n"
     ]
    }
   ],
   "source": [
    "### Replace each question mark with the appropriate value. \n",
    "### Use python, pandas or numpy methods rather than hard coding the results\n",
    "\n",
    "# TODO: Number of training examples\n",
    "n_train = len(X_train)\n",
    "assert len(X_train) == len(y_train)\n",
    "\n",
    "# TODO: Number of validation examples\n",
    "n_validation = len(X_valid)\n",
    "assert len(X_valid) == len(y_valid)\n",
    "\n",
    "# TODO: Number of testing examples.\n",
    "n_test = len(X_test)\n",
    "assert len(X_test) == len(y_test)\n",
    "\n",
    "# TODO: What's the shape of an traffic sign image?\n",
    "image_width = X_test.shape[1]\n",
    "image_height = X_test.shape[2]\n",
    "image_channels = X_test.shape[3]\n",
    "image_shape = (image_width, image_height, image_channels)\n",
    "\n",
    "# TODO: How many unique classes/labels there are in the dataset.\n",
    "n_classes = len(set(y_train))\n",
    "\n",
    "print(\"Number of training examples =\", n_train)\n",
    "print(\"Number of validation examples =\", n_validation)\n",
    "print(\"Number of testing examples =\", n_test)\n",
    "print(\"Image data shape =\", image_shape)\n",
    "print(\"Number of classes =\", n_classes)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Include an exploratory visualization of the dataset"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Visualize the German Traffic Signs Dataset using the pickled file(s). This is open ended, suggestions include: plotting traffic sign images, plotting the count of each sign, etc. \n",
    "\n",
    "The [Matplotlib](http://matplotlib.org/) [examples](http://matplotlib.org/examples/index.html) and [gallery](http://matplotlib.org/gallery.html) pages are a great resource for doing visualizations in Python.\n",
    "\n",
    "**NOTE:** It's recommended you start with something simple first. If you wish to do more, come back to it after you've completed the rest of the sections. It can be interesting to look at the distribution of classes in the training, validation and test set. Is the distribution the same? Are there more examples of some classes than others?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Traffic Sign Id, Traffic Sign Label, Count in Training Set\n",
      "0, Speed limit (20km/h), 180\n",
      "1, Speed limit (30km/h), 1980\n",
      "2, Speed limit (50km/h), 2010\n",
      "3, Speed limit (60km/h), 1260\n",
      "4, Speed limit (70km/h), 1770\n",
      "5, Speed limit (80km/h), 1650\n",
      "6, End of speed limit (80km/h), 360\n",
      "7, Speed limit (100km/h), 1290\n",
      "8, Speed limit (120km/h), 1260\n",
      "9, No passing, 1320\n",
      "10, No passing for vehicles over 3.5 metric tons, 1800\n",
      "11, Right-of-way at the next intersection, 1170\n",
      "12, Priority road, 1890\n",
      "13, Yield, 1920\n",
      "14, Stop, 690\n",
      "15, No vehicles, 540\n",
      "16, Vehicles over 3.5 metric tons prohibited, 360\n",
      "17, No entry, 990\n",
      "18, General caution, 1080\n",
      "19, Dangerous curve to the left, 180\n",
      "20, Dangerous curve to the right, 300\n",
      "21, Double curve, 270\n",
      "22, Bumpy road, 330\n",
      "23, Slippery road, 450\n",
      "24, Road narrows on the right, 240\n",
      "25, Road work, 1350\n",
      "26, Traffic signals, 540\n",
      "27, Pedestrians, 210\n",
      "28, Children crossing, 480\n",
      "29, Bicycles crossing, 240\n",
      "30, Beware of ice/snow, 390\n",
      "31, Wild animals crossing, 690\n",
      "32, End of all speed and passing limits, 210\n",
      "33, Turn right ahead, 599\n",
      "34, Turn left ahead, 360\n",
      "35, Ahead only, 1080\n",
      "36, Go straight or right, 330\n",
      "37, Go straight or left, 180\n",
      "38, Keep right, 1860\n",
      "39, Keep left, 270\n",
      "40, Roundabout mandatory, 300\n",
      "41, End of no passing, 210\n",
      "42, End of no passing by vehicles over 3.5 metric tons, 210\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdAAAAEWCAYAAADW7MapAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12a738710>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import csv\n",
    "import collections\n",
    "import random\n",
    "import numpy as np\n",
    "### Data exploration visualization code goes here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "import matplotlib.pyplot as plt\n",
    "# Visualizations will be shown in the notebook.\n",
    "%matplotlib inline\n",
    "\n",
    "signnames = {}\n",
    "with open('signnames.csv', 'r') as csvfile:\n",
    "    csvreader = csv.reader(csvfile)\n",
    "    next(csvreader) # skip header\n",
    "    for row in csvreader:\n",
    "        # add each sign description to the dictionary\n",
    "        if row[0] in signnames :\n",
    "            raise RuntimeError('key ' + row[0] + ' already exists: ' + signames[row[0]])\n",
    "        signnames[row[0]] = row[1]\n",
    "        \n",
    "counter = collections.Counter(y_train)\n",
    "print(\"Traffic Sign Id, Traffic Sign Label, Count in Training Set\")\n",
    "for elem, count in counter.items() :\n",
    "    label = signnames[str(elem)]\n",
    "    print(\"%d, %s, %d\" %(elem, label, count))\n",
    "\n",
    "print()\n",
    "# Plot counts of five most common signs\n",
    "objects = []\n",
    "counts = []\n",
    "for elem, count in counter.most_common(5) :\n",
    "    label = signnames[str(elem)]\n",
    "    objects.append(label)\n",
    "    counts.append(count)\n",
    "    \n",
    "y_pos = np.arange(len(objects)) \n",
    "plt.barh(y_pos, counts, align='center', alpha=0.5)\n",
    "plt.yticks(y_pos, objects)\n",
    "plt.xlabel('Signs')\n",
    "plt.title('Counts of five most common signs') \n",
    "plt.show()    \n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nbhD/BxxMK47whYUbAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1167d3390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 10 No passing for vehicles over 3.5 metric tons\n",
      "per channel means:  [ 124.68847656  106.96972656  108.7109375 ]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12a74b198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 25 Road work\n",
      "per channel means:  [ 47.58789062  39.56640625  35.07617188]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAETBJREFUeJztnFlsHEd6x39fXzPD+9ZBUaIky5It\nyytHPuIjiJFs9nrwJkASZB+CBAjgvCyQAHnIIk95231I8hrAwS4QINkEi+xussBu4niNjU9IPmVZ\nhynRkkmREiVRwyGHc3d35eGrHlISJY5EuS3Y8wFEDburq7q+/td31VclxhjalA45n/ULfJGozewU\nqc3sFKnN7BSpzewUqc3sFKnN7BRpQ8wWka+JyISITIrId+7WS31eSe7UqRERFzgN/A4wA7wNfMsY\nc/Luvd7ni7wNPPs4MGmMOQsgIv8OfBO4KbN7BobM8Oh2xIDv2UklWoSxfvTQGFzia15OXI+GXsLY\nenGsFxwBEbH3tE4URTTqof5uRLabBgAZV+jsyOrvIKNteHJbA2/YsmygXo+5OjvNcn5+3UY2wuxR\n4Pyq/2eAJ66vJCLPA88DDG0d47v/+QaZumHLYIdWCLS4XNYh5MMqPU4dgKFQGRX0DzO3rEyulJWJ\n9WoJgIznkvF9AKpVvVdcKDE3dRWAhTkts1wGYLzf4YkvPQTAnl07AMj1KdNbFaqztjwawvnzJb77\n3G+09NxGmL3Wl7xBJhljXgBeANh94JAxRnAyOUp1Y+9rM2FZGVa7eIX5xiUAKoG+3kjkc2U2D4Bj\ntJ6IPl9zXUJX6xlHy47AZ3gop78dy8i6ormjN8eC7btoZ8fCvFbpG4Cc5Uis35mwDq577Zg8CxDf\ng97ODly3ta+0EWbPAGOr/t8GXLjVA44jdHVliGtQseKgWlNE10tLAOzrc9g6ug+AS2VFb7GQZ+eg\nMi0bKNN8T5luHKhZpoUmGbQw1Nuj7Q5pPxJ3an0TUkJnzrGzdmIa/TCl43mcss6ELb1DAASdvTi2\n2dF+bdPd0gtALXIoLVaIo/hWw14Zf0u11qa3gT0islNEAuCPgJ9toL3PPd0xso0xoYh8G3gRcIEf\nGGNOrPMMtVpEVAXH1ABwfUWeEy4AMDo2xEifore7V9Ecb+rDs0rQtdLLtUIsBkLbflICWCmDa/q0\nffu/wTTlX9LWvNVt2c5NxCWdaQPdKisqFcGzkihf0BmxlNeeGk5A4OaQFjG7ETGCMeYXwC820sYX\niTbE7NsnwXFcxIXOQOVkFFcBcDx9lcB3CZxrUYzrNDWvb8tEZxnA6rJrkW1Lz6JuLewl17KbtIwc\nF7dLe0jeIdcBNVvRzWivkbX9gjLEYbyGWbA2td31FClVZMcCJVeoRVCqKHI8o2jJOVquBsmKbXHj\nNVbdu94G9Rrg11UnkLGmonWiYm4Eou+utO1aZdBs04WGnTLzxQoA+bqyrRr5xJHTKrDTZrahnGlw\n6UqewY5uvWhNvlxg7e41Xt3hRobGq+6Z667VFyPe/eX/AlCN5gB4+ivPAeANb2rWT8oVkbN2P2I/\nRke3Ku5S1X6QsqztbdyE2mIkRUoV2Z4rbOr16M/2kbVdu0tdAPhFqyhXYTtBqnCtQlxdxqvu5Wx5\n7HCNI//x3wBs2qEmZWX/IwB0929CvJV2V5PhRhFjABs1oLikYqRmtAHPyeDfBrjbyE6RUkV2o1pj\nduIcrmfIuuop9FQV0V5dYx/eSK6J7QYryipBxVrITgYxd+YKAMde+jk9BQ08VUK9Vpk9pf3tHQVv\nyLbrr9nm9ZTERvqtk5WYk1KBqFLGldbc9VSZXV6u8P7rxzCZCllfYxXbrCe5d8QyducwiX73VjE7\nmarXM8QFQv1OzBx+C4DSxRcZ6VcPMHa3AHDslZ8C8NjuLH37vmaf9lmPBJqG/FLRRiYbyv3J8+do\nVOao10vrtgNtMZIqpYrserXO9MQsS45hy2aFS5eo6ecPakRNTAPHqkZ/lfeXIPv60tShPFsAoHrq\nAwCy4RT7nvl9vb+sQzx+8kcATB9+l97NT2kbfd3XtHUz5CUR1DjWmTB3WQ3vj8/Os1w4Q6Vaa2n8\nbWSnSOmafo7HQDDAfN5BBrTruHERgMDGm2+2QnX95SayqyEXJiYBuDipK3Lj+4bZ9fWvArBwogxA\nZeJVAOZOzXPfwUUAOvaPalv+2n0klCy3NVSXU1lSjC7ORxSuVIgan348u023Saki24QhYSFPJsxh\nbNzYMyqzO0RDb77jNSPOSTRvTcRZs+Tq9Hk+eFOjvLmsIuzBLz9HsGsXAFlnGYDMe48CkJ8+SeWc\nht1zu4e1EX/Q9rM29hKZ7VkAuzW1RsLFMpXCFeIoXPO56ylVZgsxQbzMgL8IRVVqmS590SS4L7HT\nDK3G1zyrlLCjpqtXTL56kuJ5NfkeeeoQAEN7v0zoq03cvV2H+PCzau6d+bdFPj76cwCyhwYA6OxO\nFmyDG97ZsBKIIrZMtiv2y8sFHAqswOLW1BYjKVK6CtJzGB7M0ROVWCqokvL7u6+pEy0tIV0qYhyL\nztWIaCypwrtw5BwAi0c/YHuPpkVs3rEHALd7mKqj86LT1/q7xu8HoLbrYT4++ToAw+8fA2B8WEWM\n5FaQnThPkQG7Nt3MbSmWdXks6MyRCfxmWHY9aiM7RUpXQcYhYW2BCKhYdJRKWubzGsu4MDVJpl8z\nJCRJ0ECaArx+WRXqzPE39LmrE2zfo8rQ3fwgAPPFmPeP/xKAA9sV9Z3mMQAyIzupnVBlPPO2ptt0\nb9X4ycDBbsRfWWQAqMRQqKlMLtjIZGlZZ8vyYoEgLmDie1BBho06ly5OU/RzLFuva/iqioxaZQSA\nemxWFKPNQ4xECO2639IFFT/TU+8CsJgrUd12HwATsX6c/OmzvHHkFQAq+zU2MpJTYzrK1PAGHtB6\nUxpUKUxokKpj7zCBrzkhSYpZvhEyMasfeHZe36xR03Lx4jyN8hz1WlL71tQWIylSuqafI2QzHksG\nHLtSHbs2DtKhiBrevhsvp1O/bnNFPIHqsk7ViQ+nAZi9/BEA2x57mNgqv58efhOATLZMvqBT/613\ndFlsqOOHABwcf5DN92ls5PKbL2s59SIAm4rboeeAvhcr4qQe2xQ5VGHXQ6sgvUGu5n2isK0g7zlK\nFdlBEDA2Nkam7lMJdfF0pMve7LIRuJ5eYpuVKtaVceoR0yc1/nHiw/cAGNiuSnTw8ScpBtsAmJp/\nGwDT+JhKQZewFmuKwo5RlbvBA7vZcejXAKiePw3AmRP/AsD9E4/TvXk/AJF1G3uzHju39mtbRZ0t\n5boqyGrkUq55xOYuIVtExkTkVyJySkROiMhf2OsDIvKSiJyxZX9LPX6BqRVkh8BfGWPeE5Fu4F0R\neQn4U+BlY8z37BaP7wB/fauGBIfAy5ExHbieIq8ro8jr6rRpv74htC6Fa52JpdkyR1/W+EdlWeMa\nX3rmK3pz7BmmL+hS78CAmoAfnTjNwvxZAHYN6ixpBHqvIlvp36E6YfxRdYKmf6KxkfOvnCS7S6OQ\nuXGdOT7QZVeSszlFdmizuEqVJWrVIia+S8tixpiLwEX7uygip9BE+G8Cz9pq/wz8H+swu9Goc2Fm\nlplygOdpcGMgU9SBPLgVgEBi3MR/K+vg5o7OUJnQ+MeITQHevF0ZtZgdojA3BYC/rIMeG36AoMuG\nUbdrGyeWlOljRZene/V31141GXvHngbg8uQ0ve9pP9tHNUglfrYZmHFtJMr3bCykXkbC+koMdh26\nLQUpIuPAI8ARYJP9EMkHGbnJM8+LyDsi8k61Wrmd7j531LKCFJEu4MfAXxpjlpJ9LOvR6p0H3d29\n5vSZEyxJhsDRRdL9I6ooe0S/eyAOngV2YU7ROXP8CFkb5tu192EAxvapUiyHEV2eTfPdqUgN3YeY\nKtjV+q2a8D4zp6nD2f4ePJs82T2u2zw236cLDUtT3ydvI4jbFtXjdAa3N8fSsJ5iGClonKiGE9/l\nvBER8VFG/6sx5if28iUR2WLvbwG7aaVNN6V1kS0K4e8Dp4wx/7Dq1s+APwG+Z8v/Wrc3MRgnxlTy\nGIsOb0TdadcmZ2ScAC9SDOTPfgzAmVNvMbZFldj9T/46AP6AInWX00XPbynaz08pwk8cmWLp6HEA\nvrFfZ0BmXPdWPbTnCRybSdk3oOXeQ48D8PrrP+bsh+8AcGBW9YDbN0pokz+rFpvFqpp+Ua2K56yV\nibg2tSJGngb+GPhQRI7aa3+DMvlHIvJnwDTwBy31+AWmVqyR17n5p/vt2+nMGGgYyIhDp90VlA00\neOT7KrsdyVIq6L1Tr6kDU5id4KtPfR2AjmAzAMVpjdi5mW56rQNS61BZ3DnewZ7SswAcyOlqTO+g\nyvMghnhRl8pMpO730JDK4p179vPGq0cAmHxDET46uouoUy0luyuQYkWVSr3aQENWrSUNp7vzQARx\nHATBswwKMnbfjKuMKhph4rCNsr2rXmNfeZErb38IQHnGbtHr1I/j+R6xFUGho35V6ULEga2ah+K9\nr6Zl+YR6lyXveHNRQnw1oApnbQSy7pGp6HPHDyvTa3tH6XxSc1AST7HasNsCGxH1RhXzaZh+bdoY\npbt4YAz1egMnCnHtuoDjJtE1RWc5djj7iSrGTT0qAqLqFi5/ogF+OWdLWUm1jG0EPLbxbGNyXDEq\nKhxUEYsdaiwuTrKx0dh8QLsPshZWcFybHCKK9nx+kkyjbN/RRv+scg+jClHcuhhpIztFSldmG0Mc\nGhykmRbsWHmXsVuhu2KHh59UU66i1h6BeYTyssreOFJXOQrtgQFhRMPGl2vNeyEmsgi1W7sSsRrH\nhobN84ibTopdLQobNIy2tdinW7rHR/sxUd0OQGeCJ9qPF9cRz6VVBy9VZjvi05kZIm7EiPWBEnHg\nOTqQHA4PHLQLC4d0t0DIjQk7yckdjgHLd+q2kolWNqUmm5P85BAIZ+XZ66kRg4maC2LaloTMVTQO\nHNsPIVYM+aKKvlXx0BYjKVKqyO7q7OKpx3+Tc5MXWL6qGVENu9m0HCX740Ji0eldi2wyfGyayIzt\nlE3KSIBk+SzRe66QKC1jN48mK1feSvWmWkvW8LMmRgK7GyHWqF/sCPOJRLLzK3SSUyAcYhO1vDWv\njewUKd09NY0aly+cpiPrsGxNMmOdmpqrcnHmyiIU1alJ8jQapRrDSQza7msxnjXXnIBk+5exC7PG\nSNO8E+vwJMmROZHmoQPGlk5i0s0tNJPya4s2HNyZY2nB7vvJaZ+S1TLEa/mQAO2nTalRqsiu1it8\nNP0RXb6Lh6JlsF+jd31bxgH4n5nLnD6l2U7xebVYTN2lN6dOyqPjag9uHdP0hUkTcPaCHoVTnNPS\nhIacdcldWwYZde97sx305zTxvr9TUZzL6WJzEAQwZY/VsdaRZwzLGRsU6dO4TLJFOxKfuBHR6qFm\n6Zp+nkfnyCDR4gKeVSt9nTrwzhFdF5wuuvzwVRsEOqeepMQNBrt1qgenNb+k6ykd8MXeYV5+7TUA\n3LwuNnR7OXo7NcUs163xj55etdP7tw7xoF0O2z2kSjAYs0zsyNz40qUqC/Zwr8tZFXUm0AODIq+D\nMHJWnWCyzvhbqtWmu0J3fK7fHXUmcgUoAfOpdXrnNETr77nDGDO8XqVUmQ0gIu8YYx5NtdM7oE/j\nPdtiJEVqMztF+iyY/cJn0Oed0F1/z9Rl9heZ2mIkRUqN2ffyWdu3yNT9WxGZFZGj9u8bG+onDTFy\nr5+1bTO6tqzO1AV+F/hDYNkY83d3o5+0kN08a9sYUweSs7bvCTLGXDTGvGd/F4EkU/euUlrMXuus\n7bs+mLtB12XqAnxbRI6JyA82mvCfFrNbOmv7s6brM3WBfwR2AwfRHPW/30j7aTH7ts/aTpvWytQ1\nxlwyxkRGU57+CRWHd0xpMfuePmv7Zpm6SUq0pd8Djm+kn1Ti2Xdy1nbKdLNM3W+JyEFU5H0C/PlG\nOml7kClS24NMkdrMTpHazE6R2sxOkdrMTpHazE6R2sxOkdrMTpH+H9lSZBwQkAuhAAAAAElFTkSu\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c7b70f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 30 Beware of ice/snow\n",
      "per channel means:  [ 188.78515625  188.36816406  197.01660156]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\ndos0ZHaLNGR2i/T/r/rshWUqOgIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c8010b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 17 No entry\n",
      "per channel means:  [ 122.86425781   74.44921875   69.5234375 ]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c831cc0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 22 Bumpy road\n",
      "per channel means:  [ 176.02832031  166.70410156  163.08203125]\n",
      "\n",
      "mean of training data\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\npQXwvit9AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c89b2b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "stddev of training data\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nwCitjrMAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c8c7e80>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# show five randomly selected traffic signs\n",
    "for i in range(0,5) :\n",
    "    index = random.randint(0, len(X_train))\n",
    "    image = X_train[index].squeeze()\n",
    "\n",
    "    plt.figure(figsize=(1,1))\n",
    "    plt.imshow(image)\n",
    "    plt.show()\n",
    "    print(\"sign id:\", y_train[index], signnames[str(y_train[index])])\n",
    "    \n",
    "    # per channel mean for each picture\n",
    "    # results show different pictures may have different brightness\n",
    "    channel_means = image.mean(axis=(0,1))\n",
    "    print(\"per channel means: \", channel_means)\n",
    "    \n",
    "print()    \n",
    "# calculate mean and stddev image from training set\n",
    "mean_image = X_train.mean(axis=0)\n",
    "stddev_image = X_train.std(axis=0)\n",
    "# show them\n",
    "print(\"mean of training data\")\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(mean_image)\n",
    "plt.show()\n",
    "print(\"stddev of training data\")\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(stddev_image)\n",
    "plt.show()\n",
    "\n",
    "feature_standardized = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "----\n",
    "\n",
    "## Step 2: Design and Test a Model Architecture\n",
    "\n",
    "Design and implement a deep learning model that learns to recognize traffic signs. Train and test your model on the [German Traffic Sign Dataset](http://benchmark.ini.rub.de/?section=gtsrb&subsection=dataset).\n",
    "\n",
    "The LeNet-5 implementation shown in the [classroom](https://classroom.udacity.com/nanodegrees/nd013/parts/fbf77062-5703-404e-b60c-95b78b2f3f9e/modules/6df7ae49-c61c-4bb2-a23e-6527e69209ec/lessons/601ae704-1035-4287-8b11-e2c2716217ad/concepts/d4aca031-508f-4e0b-b493-e7b706120f81) at the end of the CNN lesson is a solid starting point. You'll have to change the number of classes and possibly the preprocessing, but aside from that it's plug and play! \n",
    "\n",
    "With the LeNet-5 solution from the lecture, you should expect a validation set accuracy of about 0.89. To meet specifications, the validation set accuracy will need to be at least 0.93. It is possible to get an even higher accuracy, but 0.93 is the minimum for a successful project submission. \n",
    "\n",
    "There are various aspects to consider when thinking about this problem:\n",
    "\n",
    "- Neural network architecture (is the network over or underfitting?)\n",
    "- Play around preprocessing techniques (normalization, rgb to grayscale, etc)\n",
    "- Number of examples per label (some have more than others).\n",
    "- Generate fake data.\n",
    "\n",
    "Here is an example of a [published baseline model on this problem](http://yann.lecun.com/exdb/publis/pdf/sermanet-ijcnn-11.pdf). It's not required to be familiar with the approach used in the paper but, it's good practice to try to read papers like these."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Pre-process the Data Set (normalization, grayscale, etc.)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Minimally, the image data should be normalized so that the data has mean zero and equal variance. For image data, `(pixel - 128)/ 128` is a quick way to approximately normalize the data and can be used in this project. \n",
    "\n",
    "Other pre-processing steps are optional. You can try different techniques to see if it improves performance. \n",
    "\n",
    "Use the code cell (or multiple code cells, if necessary) to implement the first step of your project."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Example of rotated image\n",
      "shape of pre-processed and augmentated training data : (104397, 32, 32, 3) (104397,)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAECNJREFUeJztnEuoZVdax3/f2nuffR73nLqPeprE\nJD6m0oLowIkigjhpHSjdA1EQ2kmDggMbRw57oE6FiA0OBBEUbKFBGtFBT6Rj02jH0DGaNrmxktRN\n3cd57dday8H37X1vVarq3tSt7BSp88Fln7sfa6397f/63mtJjJEN9UPu0x7As0QbZvdIG2b3SBtm\n90gbZvdIG2b3SBtm90iXYraI/JKIfF9E3hSRrzypQX1WSR7XqRGRBHgD+EVgH/g28MUY438+ueF9\ntii9xLM/DbwZY/wfABH5a+DzwEOZPcqzOBvnlI2nqr0OIM0AaD96CBERufdcDNyPidP/HwAWEeS+\nGy8EKTn7U58QJ2S5sinLdKypXXM+UPvAYl1RVI1wDl2G2c8B75z5fx/4mftvEpEvAV8CmI4GfOHn\nf4K37pzwgzvHAFy7eguApmwAKIq6Y3bj7VxZUjf6cXzUaz4E7SAEnOjL+5ZBaUrmVEKGugYgmsQM\n97Dd7pdurGfGrc+NJo7rL98E4MatawDcTPS5wfGS90+W/MO33ngEm07pMsx+0Jf8CIBijK8ArwBc\n39mKIUZi9AjKvKTVGnbMEsFb08EbE0Og8S1C9VqHfucIKOOdO73WNPqh7Dt0TI7EBzA3tmM98yLt\nrKpIKHWsUgGwtA+9yhxFlhHkXFADl2P2PvDCmf+fB/7vUQ8E7zk5WUB0OFHupk5fKrFj0VT4lpH2\n3FlBIHI6vfWsENob25eOINY+YjPgHjlkH+U+vER90NpPdDxlzf47H3R9AVy9sQPAuimRpCLKhYTU\npayRbwM/LiIvi8gA+ALw9Uu095mnx0Z2jLERkS8D/wgkwNdijK896hmRhOFwQixWYDLVuRapesxz\nRwyKoKry1hed0DoVFfp/iLHDp7NfTs7IM+snBm1LOtV32kgL+kg4bd+QPR5f4cXnVWbfvHlF+0xV\nnr/08gvcefu/SdNPXowQY/wG8I3LtPEs0aWY/fEpEnxQ5IoixxmCxGSrbzwePRc6cRs7Gd8eWwUm\nRJIW9daG4MDpq8WgijIaikMIp4ox3qsYI+BsJmzNRgDcuj7jxs4QgNyvAVhFbfOdt/c5un2X2szY\n82jjrvdIvSI7IgRJqUJBNES3xyAmU7McCYZeKbvnWjS2VkYreX0MRINMNFQmac5wsKXtNgUAZbkA\noKg80Wz1dkaclbhJqiwZjhXNyUiQTGV0PlKnplipCVgf1+Qx/4hV8zDqldkhBlbVmiqEblqT6BBk\noP82q6pzYEJozT3pXsh1prE5OeIJplAbuyfJR0ymM71etiacigBxadeWdPZ16MZYVsrYO8dLAGrX\nEH0OwGqoH84XKwBmwxnjvRmD9O0Lvf9GjPRIvSI7cY6tUY6b153ZldhxMtBpu04K1mVpT5wiO8TW\nidFzqbnMk8RBra8xnqizMZ3OTmMubgzAcHcPgKaqaGpt31d6XNba1kmM1I21v9Brx5UnLhXt65n2\nc2Nb21yvA8fz442CfBqpV2QLMEwTEhFELJKGCuusVnMqD4K710+3SaByNSYqg9NcZ8LuIGHY2LlW\nDcyPWVh7gy1FYbJWmT2UyDBX1Hsz+Q4THYMfpDROn5ta7GYQHWXrXKF9DkaqfBeHxxRlQwinMv9R\n1LM1AjVQhwjOXtAro6qlKh1pYsfsLsTqIpkxeWdrAsA01+fdegWlKq4qmPgJgdqUYLU0K6dWC0KI\nVL5VttrmdLYNQL49QXIdxwgVHeUy8MHdubZrltDd47mNryYbuy5Ocx5txEiP1LPpFzmpS9Z+TUh0\nejeN2axVi8qmi0+3XiXOMZkooq+P1LNL12o3r1YrskSvJWM19/JxyspQHlKdAdcMvX4x5/DkUJvN\n9Nr2NY1TJ1sppZmU69KOsSGaedqGUus2pJs6SMNpHPcc2iC7R+pXQQoMU8e1acrcnId6qSiLlcrd\nIEkbqGOY6Y/ZIGNinl002R5KPV6ZbTHdeVnbzxWhs72Mtdd2a1QZzkaK7KQpmS1OAGi8or8KOpaj\nw/eoUUXaGHqX8zUtm5LM4jipmYd5wq3r22SDjVPz1FG/yI7CoHHsJkMGQWX1eqWaPctVhuNSojk1\nedThTWNCtlLkR6+ydDhWpO7deoHxznMApBN1XBiU+EaRmYi62iQj+3/EjpmD1VzR/8HBewDUVYVZ\npCQWKkh9Q9Uo8ldrixyaVVKL42S5Ps2HnkP9epAIV8KIUKcU1REA06EO4co1ZdRx0YB5ZEkwz9Cl\nSDTxYdzItzRRnE9/mNG2KkYZKDPKKMRgHmSb6DXrrBbpbHVGyvR8MgVgq66oLYzqzd7eno0ZmPIO\nBpCitFhMBaFqNh7k00i9Ijt3CS+Ot5ByxdxExMhMsunuDQDWx0e4uaJ4gIqAPEmpzXERM9eG26oM\ns8kOpc3i2kQNSUTMM40mAhaFlk4Q+YiplprJOC1KVktFb2XKeXucs22Ini9V5N0pFP1xPAEyHlxo\n8FHaILtH6hXZLnFsbY0IHx50gf7R9LoeZ4bUuiYZmNPRItY3FKbwRuauj6can/B1ibMZcHSoyB5N\nMuqoaCRTebpaK7KLqqHywdrVW3ZyVZ4zyUgtRFBWFv+WlE7EW0xlaqZfmGzhak6zz+dQr8wuG89b\nxyfsnxzjtlQhjnevApCNVVmli5TEUihtTL8WR9MmGczeDkFFzeH+PuOBioHlkQX3JymLqL+3n9OM\n+J2DdwFYrDyFhWS9pXiqsX7AdCg0ZgktTw608yyhGVi+dKyiKR9pQKqRlMV61SU5zqONGOmRekV2\nUdd8b/82BSnP39RajNG2oiqmOqezgWdgoypXipgmCB0uoio8vJqOTfkux/M7AIioCVcXkWRk+UhR\nFGaJPjcZjgELu2YqfsaDNq9Z0JZXBZMxjQ+dAiYJ1pYNoakpinKD7KeR+o36iVBmQ4ZbV9i6qqZe\nOlQ5GDDlNkjJU4WOtwhc6j3OTDhpFZdVuO7s7RHFIoFDi+xR4y1tlpgJ92MvvqT/J7usS1WubRBm\nZPK9OdhnXlgkcHcXgNJFhiazKxuXS/W5ogmI92drrB5J5yJbRF4QkX8WkddF5DUR+V07vysi3xSR\n/7LjzoV6fIbpIshugN+PMX5HVCj+m4h8E/gt4J9ijF+1JR5fAf7gkS2Jww1HzPZukW+plSCJmlNi\nlUuBtKvBdpbnGmUJ3vztulAULo4UTdvXX2L7qraVWNH6uilY2Qxo3fTMUJkPxuxOVU+IVTY1Sy1b\nuBtWXSnzbFvRL+OEOmlj29pWaeXImXesTy6O7HOZHWO8Ddy233MReR0thP888HN2218C/8J5zI4Q\nK8/u9h4DyyFG2vyeTrLDZcldU0513WbZa4atftT3ZG0Z71mVkDltS2x6p6kjs9+l1XU7yzO6NCG1\nlJesNdR69z2t6T8+PKCxoNLQ4jKjxJFnVvRpoylsrJUT/NaQNLmY6vtYClJEXgJ+EvhX4IZ9iPaD\nXH/IM18SkVdF5NXKVgE8q3RhBSkiW8DfAr8XYzyRC3pNZ1ce3Njdiz/y3EtsT8ZdOqyNU7i2brdx\nOFHRUgcVBXeLkqnTc1t2bCr1Mo8O3sA5vW9643lta+S6UKw3b7EVTTF6KoviLW6ro3Pn3X3tz3uw\n2MvJ0orzY8VQ9S/OPMdO5PnIJM9wF+TFhZAtIhnK6L+KMf6dnX5fRG7Z9VvABxfq8Rmmc5EtCuG/\nAF6PMf7pmUtfB34T+Kod//68thJxzLKc1eEBWJFiW6TuDC0zD89ZzPoDQ/sCoTJEt4uUqpXGOqRe\nENb67OpYEXv1h7apvEboFqXOoGCITZzj5O6HABweKD7WpiAHeU5t5uaHrb5Ye7ZHitzpRMd1VOn9\nlSQMJCFccHnjRcTIzwK/AfyHiHzXzv0hyuS/EZHfBt4Gfu1CPT7DdBFr5Fs8PGD7Cx+ns7JY89b3\nX8MPMhhZkY4Vz9AqT98QLBmc2bHxDdEKYbbNiqFRQVpVEV+r/J6f3AXgw/cS/NDKe23orSWRlAWl\nyexoQa3GLI9yVVGJBrgKZ5W0RLCIY7FWqXuw0nvcaMQkz7lgVqxfD9L7hqPjA2KaURxYeddAmd6t\nAhPXlYWt7S0a76mxKF9reydqByckDESndWpeaFV4ilVbRmwhUxtDGj2S6WvXlsBYWYXUuqrIR9rn\n9T213UeTnMVSa1QWVmw5s4hlkiVUjf/IgtiH0SY20iP1XjeSDBzDSU5t9XI75qm1a1my0YiVFcOv\nPrRUll90GexjEy0Lp0jPJWFguL15xTzD0uMrvX/QmpZt4bsvieZpLi3Vdliogl2Hih2rQcwsErg7\nHbK9p2O8O1cTs1pafKZBveInafpt6MlQv1WsAvVAOFoeszBTz5Uqb2/uaBxrMM5oLJOSWKUqK9dW\nDJ+uILMkbIieSrSt2yeK8KYMpFYnkmUtnvTBqq5pap0V88IqVX27NFuoLe6xMoV9VBbduvpgq9gG\npnx9GZDqYmUM0DuzHXWWczAvqG3m7Q0tEm/BHklqKhMV3hIF4hwST5fiYb/aNtvgfpsZj9HhLBA1\nCmZfW0ZiHR1F0XqX7WoGSzTgWCyVsW/+7/sAZLcd+VBtmcmWpt9GA/1/eXzCwAlxkzx4+qhnZAt1\nkuFJwSqOagvjFU5RnIjgknbdzOkSuvsna1t/nkoCvl2IaiZdCMT2upUYt9jzsSCYydfFZdoFrU4Q\nK5Bvd4Qoi0hhouLYCjJbWy9LIy/evIq4TyDqt6HLUb9OTYjMi5qG0+0nGvvea5O7GaebAVi8Hyfx\nzM42tn7GlktnznUpMh9O8d/uWdJYW4kVoaRZSuPbDQisplDa/UrOuMqGXt+ELnGRWl3iyEofprMR\n2Wi4QfbTSP3K7BipqwYR18ncxuIO3jR6WZZgJpazmoGYua5SND3dC0OPSdJNgdhu0ySu24ilvW00\nthh5tWZrpk7K0M6lFgdJYqAxh2pudYNVWVO3GwmYjL8x01KIG7tXWJ+c3LNC+FHU+9K8ROSe/UBs\nRuNrW3o3SLvVV+16djfoLEOit8BVu5OOQGMvG7pFqqHLnwXLnI/HWlMynVylifpx2uecqd8s1N12\nR9Ey6n5dE+wD7O6p6XfzqrYVTuaEcg0b0+/po8fe1++xOhO5AyyBg946fXy6ysXH+WKM8dp5N/XK\nbAAReTXG+FO9dvoY9EmMcyNGeqQNs3ukT4PZr3wKfT4OPfFx9i6zn2XaiJEeqTdmP817bT+iUveP\nRORdEfmu/f3ypfrpQ4w87XttW0XXrbOVusCvAL8OLGKMf/wk+ukL2d1e2zHGCmj32n4qKMZ4O8b4\nHfs9B9pK3SdKfTH7QXttP/GXeRJ0X6UuwJdF5N9F5GuXLfjvi9kX2mv706b7K3WBPwN+FPgcWqP+\nJ5dpvy9mf+y9tvumB1XqxhjfjzH6qDHUP0fF4WNTX8x+qvfaflilblsSbfSrwPcu008v8ezH2Wu7\nZ3pYpe4XReRzqMj7AfA7l+lk40H2SBsPskfaMLtH2jC7R9owu0faMLtH2jC7R9owu0faMLtH+n/q\nBwIfIv04vAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x130526198>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAX3nZO4TSlYIAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x13052c6a0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x131865d68>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### Preprocess the data here. It is required to normalize the data. Other preprocessing steps could include \n",
    "### converting to grayscale, etc.\n",
    "### Feel free to use as many code cells as needed.\n",
    "import cv2\n",
    "\n",
    "def preprocess(images) :\n",
    "    # rescale to [0,1]\n",
    "    images = 0.1 + (images / 255.0)*0.8\n",
    "    # remove per channel brightness from each image\n",
    "    for image in images :\n",
    "        channel_means = image.mean(axis=(0,1))\n",
    "        for channel in range(0,image.shape[2]) : \n",
    "            image[:,:,channel] -= channel_means[channel]\n",
    "    # calculate mean and stddev for each dimension for all images\n",
    "    mean_image = images.mean(axis=0)\n",
    "    stddev_image = images.std(axis=0)\n",
    "    # normalize the images\n",
    "    images = images -  mean_image\n",
    "    images = images / stddev_image\n",
    "    return images\n",
    "\n",
    "# Shuffling the training data\n",
    "from sklearn.utils import shuffle\n",
    "X_train, y_train = shuffle(X_train, y_train)\n",
    "\n",
    "# Data augmentation\n",
    "# for each image in X_train, add one image rotated -15 degrees and another rotated +15 degrees\n",
    "print(\"Example of rotated image\")\n",
    "image = X_train[0]\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(image)\n",
    "rows,cols,ch = image.shape    \n",
    "M = cv2.getRotationMatrix2D((cols/2,rows/2),15,1)\n",
    "rot_image = cv2.warpAffine(image,M,(cols,rows))\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(rot_image)\n",
    "M = cv2.getRotationMatrix2D((cols/2,rows/2),-15,1)\n",
    "rot_image = cv2.warpAffine(image,M,(cols,rows))\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(rot_image)\n",
    "\n",
    "if not feature_standardized :\n",
    "    \n",
    "    # data augmentation only for training set\n",
    "    X_aug = []\n",
    "    y_aug = []\n",
    "    for image, label in zip(X_train, y_train) :\n",
    "        rows,cols,ch = image.shape\n",
    "        # rotate by +15 degrees\n",
    "        M = cv2.getRotationMatrix2D((cols/2,rows/2),15,1)\n",
    "        rot_image = cv2.warpAffine(image,M,(cols,rows))\n",
    "        X_aug.append(rot_image)\n",
    "        y_aug.append(label)\n",
    "        # rotate by -15 degrees\n",
    "        M = cv2.getRotationMatrix2D((cols/2,rows/2),-15,1)\n",
    "        rot_image = cv2.warpAffine(image,M,(cols,rows))\n",
    "        X_aug.append(rot_image)\n",
    "        y_aug.append(label)\n",
    "\n",
    "    X_aug = np.array(X_aug)\n",
    "    y_aug = np.array(y_aug)\n",
    "    X_train = np.concatenate((X_train, X_aug))\n",
    "    y_train = np.concatenate((y_train, y_aug))\n",
    "\n",
    "    # pre-process\n",
    "    X_train = preprocess(X_train)\n",
    "    X_valid = preprocess(X_valid)\n",
    "    X_test = preprocess(X_test)\n",
    "    \n",
    "    feature_standardized = True\n",
    "\n",
    "print(\"shape of pre-processed and augmentated training data :\", X_train.shape, y_train.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Define your architecture here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "# Setting up TensorFlow\n",
    "import tensorflow as tf\n",
    "\n",
    "EPOCHS = 50\n",
    "BATCH_SIZE = 128"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# LeNet Architecture modified for traffic signs\n",
    "\n",
    "from tensorflow.contrib.layers import flatten\n",
    "\n",
    "def LeNet(x):    \n",
    "    # Arguments used for tf.truncated_normal, randomly defines variables for the weights and biases for each layer\n",
    "    mu = 0\n",
    "    sigma = 0.1\n",
    "    \n",
    "    # ColorMap Layer : Input 32x32x3. Output = 32x32x3\n",
    "    # Filter size : 1x1x3x3\n",
    "    WC = tf.Variable(tf.truncated_normal(shape=(1,1,3,3), mean = mu, stddev=sigma))\n",
    "    BC = tf.Variable(tf.zeros(3))\n",
    "    conv_color = tf.nn.conv2d(x, WC, strides=[1, 1, 1, 1], padding='SAME') + BC\n",
    "    \n",
    "    # TODO: Layer 1: Convolutional. Input = 32x32x3. Output = 28x28x6.\n",
    "    # Filter size must be 5x5x3x6 (height x width x input x output)\n",
    "    W1 = tf.Variable(tf.truncated_normal(shape=(5,5,3,6), mean = mu, stddev=sigma))\n",
    "    # Weights must be of size 6\n",
    "    B1 = tf.Variable(tf.zeros(6))\n",
    "    conv1 = tf.nn.conv2d(conv_color, W1, strides=[1, 1, 1, 1], padding='VALID') + B1\n",
    "\n",
    "    # TODO: Activation.\n",
    "    act1 = tf.nn.relu(conv1)\n",
    "\n",
    "    # TODO: Pooling. Input = 28x28x6. Output = 14x14x6.\n",
    "    pool1 = tf.nn.max_pool(act1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')\n",
    "\n",
    "    # TODO: Layer 2: Convolutional. Output = 10x10x16.\n",
    "    # Filter size: 5x5x6x16\n",
    "    W2 = tf.Variable(tf.truncated_normal(shape=(5,5,6,16), mean = mu, stddev=sigma)) \n",
    "    # Bias size: 16\n",
    "    B2 = tf.Variable(tf.zeros(16))\n",
    "    conv2 = tf.nn.conv2d(pool1, W2, strides=[1, 1, 1, 1], padding='VALID') + B2\n",
    "    \n",
    "    # TODO: Activation.\n",
    "    act2 = tf.nn.relu(conv2)\n",
    "\n",
    "    # TODO: Pooling. Input = 10x10x16. Output = 5x5x16.\n",
    "    pool2 = tf.nn.max_pool(act2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')\n",
    "\n",
    "    # TODO: Flatten. Input = 5x5x16. Output = 400\n",
    "    flat_pool2 = flatten(pool2)\n",
    "    \n",
    "    # Input = 14x14x6. Output = 1176\n",
    "    flat_pool1 = flatten(pool1)\n",
    "    \n",
    "    # concat flattened pool1 and pool2\n",
    "    # Input 400 & 1176. Output = 1576\n",
    "    flat = tf.concat(1, [flat_pool2, flat_pool1])\n",
    "    \n",
    "    # TODO: Layer 3: Fully Connected. Input = 1576. Output = 1000.\n",
    "    W3 = tf.Variable(tf.truncated_normal(shape=(1576,1000), mean = mu, stddev=sigma))\n",
    "    B3 = tf.Variable(tf.zeros(1000))\n",
    "    conv3 = tf.add(tf.matmul(flat, W3), B3)\n",
    "    \n",
    "    # TODO: Activation.\n",
    "    act3 = tf.nn.relu(conv3)\n",
    "\n",
    "    # TODO: Layer 4: Fully Connected. Input = 1000. Output = 400.\n",
    "    W4 = tf.Variable(tf.truncated_normal(shape=(1000,400), mean = mu, stddev=sigma))\n",
    "    B4 = tf.Variable(tf.zeros(400))\n",
    "    conv4 = tf.add(tf.matmul(act3, W4), B4)\n",
    "    \n",
    "    # TODO: Activation.\n",
    "    act4 = tf.nn.relu(conv4)\n",
    "\n",
    "    # TODO: Layer 5: Fully Connected. Input = 400. Output = 43.\n",
    "    W5 = tf.Variable(tf.truncated_normal(shape=(400,43), mean = mu, stddev=sigma))\n",
    "    B5 = tf.Variable(tf.zeros(43))\n",
    "    logits = tf.add(tf.matmul(act4, W5), B5)\n",
    "    \n",
    "    return logits"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Train, Validate and Test the Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A validation set can be used to assess how well the model is performing. A low accuracy on the training and validation\n",
    "sets imply underfitting. A high accuracy on the training set but low accuracy on the validation set implies overfitting."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Placeholders for features and labels\n",
    "x = tf.placeholder(tf.float32, (None, 32, 32, 3))\n",
    "y = tf.placeholder(tf.int32, (None))\n",
    "one_hot_y = tf.one_hot(y, 43)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Training Pipeline\n",
    "rate = 0.001\n",
    "\n",
    "logits = LeNet(x)\n",
    "cross_entropy = tf.nn.softmax_cross_entropy_with_logits(labels=one_hot_y, logits=logits)\n",
    "loss_operation = tf.reduce_mean(cross_entropy)\n",
    "optimizer = tf.train.AdamOptimizer(learning_rate = rate)\n",
    "#optimizer = tf.train.AdamOptimizer()\n",
    "training_operation = optimizer.minimize(loss_operation)\n",
    "\n",
    "# Model Evaluation\n",
    "correct_prediction = tf.equal(tf.argmax(logits, 1), tf.argmax(one_hot_y, 1))\n",
    "accuracy_operation = tf.reduce_mean(tf.cast(correct_prediction, tf.float32))\n",
    "saver = tf.train.Saver()\n",
    "\n",
    "def evaluate(X_data, y_data):\n",
    "    num_examples = len(X_data)\n",
    "    total_accuracy = 0\n",
    "    sess = tf.get_default_session()\n",
    "    for offset in range(0, num_examples, BATCH_SIZE):\n",
    "        batch_x, batch_y = X_data[offset:offset+BATCH_SIZE], y_data[offset:offset+BATCH_SIZE]\n",
    "        accuracy = sess.run(accuracy_operation, feed_dict={x: batch_x, y: batch_y})\n",
    "        total_accuracy += (accuracy * len(batch_x))\n",
    "    return total_accuracy / num_examples"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training...\n",
      "\n",
      "EPOCH 1: Validation Accuracy = 0.867\n",
      "Model saved\n",
      "\n",
      "EPOCH 2: Validation Accuracy = 0.895\n",
      "Model saved\n",
      "\n",
      "EPOCH 3: Validation Accuracy = 0.879\n",
      "\n",
      "EPOCH 4: Validation Accuracy = 0.902\n",
      "Model saved\n",
      "\n",
      "EPOCH 5: Validation Accuracy = 0.891\n",
      "\n",
      "EPOCH 6: Validation Accuracy = 0.923\n",
      "Model saved\n",
      "\n",
      "EPOCH 7: Validation Accuracy = 0.924\n",
      "Model saved\n",
      "\n",
      "EPOCH 8: Validation Accuracy = 0.890\n",
      "\n",
      "EPOCH 9: Validation Accuracy = 0.878\n",
      "\n",
      "EPOCH 10: Validation Accuracy = 0.928\n",
      "Model saved\n",
      "\n",
      "EPOCH 11: Validation Accuracy = 0.916\n",
      "\n",
      "EPOCH 12: Validation Accuracy = 0.915\n",
      "\n",
      "EPOCH 13: Validation Accuracy = 0.934\n",
      "Model saved\n",
      "\n",
      "EPOCH 14: Validation Accuracy = 0.922\n",
      "\n",
      "EPOCH 15: Validation Accuracy = 0.905\n",
      "\n",
      "EPOCH 16: Validation Accuracy = 0.929\n",
      "\n",
      "EPOCH 17: Validation Accuracy = 0.911\n",
      "\n",
      "EPOCH 18: Validation Accuracy = 0.918\n",
      "\n",
      "EPOCH 19: Validation Accuracy = 0.942\n",
      "Model saved\n",
      "\n",
      "EPOCH 20: Validation Accuracy = 0.909\n",
      "\n",
      "EPOCH 21: Validation Accuracy = 0.917\n",
      "\n",
      "EPOCH 22: Validation Accuracy = 0.914\n",
      "\n",
      "EPOCH 23: Validation Accuracy = 0.924\n",
      "\n",
      "EPOCH 24: Validation Accuracy = 0.920\n",
      "\n",
      "EPOCH 25: Validation Accuracy = 0.914\n",
      "\n",
      "EPOCH 26: Validation Accuracy = 0.927\n",
      "\n",
      "EPOCH 27: Validation Accuracy = 0.938\n",
      "\n",
      "EPOCH 28: Validation Accuracy = 0.933\n",
      "\n",
      "EPOCH 29: Validation Accuracy = 0.930\n",
      "\n",
      "EPOCH 30: Validation Accuracy = 0.933\n",
      "\n",
      "EPOCH 31: Validation Accuracy = 0.923\n",
      "\n",
      "EPOCH 32: Validation Accuracy = 0.930\n",
      "\n",
      "EPOCH 33: Validation Accuracy = 0.937\n",
      "\n",
      "EPOCH 34: Validation Accuracy = 0.910\n",
      "\n",
      "EPOCH 35: Validation Accuracy = 0.918\n",
      "\n",
      "EPOCH 36: Validation Accuracy = 0.920\n",
      "\n",
      "EPOCH 37: Validation Accuracy = 0.928\n",
      "\n",
      "EPOCH 38: Validation Accuracy = 0.918\n",
      "\n",
      "EPOCH 39: Validation Accuracy = 0.949\n",
      "Model saved\n",
      "\n",
      "EPOCH 40: Validation Accuracy = 0.935\n",
      "\n",
      "EPOCH 41: Validation Accuracy = 0.906\n",
      "\n",
      "EPOCH 42: Validation Accuracy = 0.922\n",
      "\n",
      "EPOCH 43: Validation Accuracy = 0.935\n",
      "\n",
      "EPOCH 44: Validation Accuracy = 0.921\n",
      "\n",
      "EPOCH 45: Validation Accuracy = 0.930\n",
      "\n",
      "EPOCH 46: Validation Accuracy = 0.926\n",
      "\n",
      "EPOCH 47: Validation Accuracy = 0.932\n",
      "\n",
      "EPOCH 48: Validation Accuracy = 0.939\n",
      "\n",
      "EPOCH 49: Validation Accuracy = 0.922\n",
      "\n",
      "EPOCH 50: Validation Accuracy = 0.922\n",
      "\n",
      "Max Validation Accuracy 0.949 achieved at epoch 39\n"
     ]
    }
   ],
   "source": [
    "### Train your model here.\n",
    "### Calculate and report the accuracy on the training and validation set.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "max_valid_accuracy = 0.0\n",
    "max_valid_epoch = 0\n",
    "max_valid_model_ckpt = './max_valid_ckpt'\n",
    "\n",
    "# Training the model and reporting validation accuracy\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    num_examples = len(X_train)\n",
    "    \n",
    "    print(\"Training...\")\n",
    "    print()\n",
    "    for i in range(EPOCHS):\n",
    "        X_train, y_train = shuffle(X_train, y_train)\n",
    "        for offset in range(0, num_examples, BATCH_SIZE):\n",
    "            end = offset + BATCH_SIZE\n",
    "            batch_x, batch_y = X_train[offset:end], y_train[offset:end]\n",
    "            sess.run(training_operation, feed_dict={x: batch_x, y: batch_y})\n",
    "        \n",
    "        epoch = i + 1\n",
    "        validation_accuracy = evaluate(X_valid, y_valid)\n",
    "        print(\"EPOCH {}: Validation Accuracy = {:.3f}\".format(epoch, validation_accuracy))\n",
    "        \n",
    "        if (validation_accuracy > max_valid_accuracy) : \n",
    "            saver.save(sess, max_valid_model_ckpt)\n",
    "            print(\"Model saved\")\n",
    "            max_valid_accuracy = validation_accuracy\n",
    "            max_valid_epoch = epoch\n",
    "            \n",
    "        print()         \n",
    "    \n",
    "print(\"Max Validation Accuracy {:.3f} achieved at epoch {:d}\".format(max_valid_accuracy, max_valid_epoch))    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "restored session\n",
      "Test Accuracy = 0.929\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "### Once a final model architecture is selected, \n",
    "### the accuracy on the test set should be calculated and reported as well.\n",
    "ready_to_submit = True\n",
    "\n",
    "if ready_to_submit == True :\n",
    "    with tf.Session() as sess:\n",
    "        saver.restore(sess, max_valid_model_ckpt)\n",
    "        print(\"restored session\")    \n",
    "        test_accuracy = evaluate(X_test, y_test)\n",
    "        print(\"Test Accuracy = {:.3f}\".format(test_accuracy))\n",
    "print(\"Done\")        "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "restored session\n",
      "Training Accuracy = 0.999\n",
      "Done\n"
     ]
    }
   ],
   "source": [
    "with tf.Session() as sess:\n",
    "    saver.restore(sess, max_valid_model_ckpt)\n",
    "    print(\"restored session\")    \n",
    "    training_accuracy = evaluate(X_train, y_train)\n",
    "    print(\"Training Accuracy = {:.3f}\".format(training_accuracy))\n",
    "print(\"Done\")   "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "## Step 3: Test a Model on New Images\n",
    "\n",
    "To give yourself more insight into how your model is working, download at least five pictures of German traffic signs from the web and use your model to predict the traffic sign type.\n",
    "\n",
    "You may find `signnames.csv` useful as it contains mappings from the class id (integer) to the actual sign name."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load and Output the Images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1419ba4e0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of the image:  (32, 32, 3)\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c8b6ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of the image:  (32, 32, 3)\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAABuRJREFUeJztnF1sFFUUx3+nu2st2G75WlpbEWLk\nQ0NEqGJsqonGREiI+qDRB+ODCT6AYEQT4pOPPiA+qDHBaOKDiTHRRB80xhAjoAYoiFbEViKNtrTF\nUnaLhsJue3yY2WXBlp3dmb0zbe8v2czszJ05hz+nZ++de+aKqmIxQ03YDswmrNgGsWIbxIptECu2\nQazYBrFiG8SX2CLysIh0i8hJEdkZlFMzFal0UCMiMaAHeAjoAw4DT6nqr8G5N7OI+7j2buCkqv4B\nICIfAY8AU4q9cOFCXbp0qQ+Tl8kHSS6XI5FIOMfGx53tpWyhnVznnJNYDIBsLkusxtmvqQkmi/b2\n9jI8PCyl2vkRuwX4q+h7H7D+6kYishnYDLBkyRI6Ozt9mLzM2NgYAJlMmsWpJgBymVEALvb3FdrV\ntrQAEG9sAGDozBD1Nzj7c+bMCcSXtrY2T+38iD3Z/+T/cpKq7gH2uE5V/CAmk84AUDuSBuBQW3vh\nXHc5NxJF1HH9rqPfATDWmAQgmUwiUjJAK8aP2H3ATUXfW4HT/ty5kq6uLgAW/XScnm0vBXNTlUJE\nHFrbfsWp5W/uIr3uDgBWrlwVjL0i/CStw8CtIrJMRK4DngQ+D8atmUnFka2qORHZCnwFxID3VfV4\nEE7lI+/c/RucbRA39UDP85f/elaO9BV8CSqx+EkjqOoXwBcB+TLj8SV2NZjIZjmweFnYbrBvfisA\nHYOnwO0++sUO1w0SGbE1m0UjEtXF7G9aVvANZZLOrXcik0b2R0zkYvK+dZx1xnBS4U9mZCJ7NhBa\nZPf09AAweM8DYblQNvsXOGO4Rd99DcCqVeUNfGxkGyS0yE4eOgrAYFgO+KDxyE/OTpmRHYrYmUyG\n7u0vh2E6EH7f5vg+Z9MGksmk5+tsGjFIKGLXjqRB1flMS5wO9/Xn0mVdZSPbIEbFnpiY4MKFCxxa\n11668TTg4Np2xsbG8DqPayPbIMYj+/z58yZNVhURIZ0+Ry6X89TeaNcvHo+TSqX4DYqeyFdvzq96\nOGlDVWla3FyY3S+FTSMGMRrZOj5OLuPMkt93tq9E6+izb34rudHRQr1KKWxkG8So2Hopy8W+QKsd\nQudif78zseABs5EtTM/fw4CwacQgRsWWRILalhtNmqw6tS0tiO36RQ+jXT+JxYi7z3/3LWjNHzXp\nQkBcfhYSTzYUypFLUTKyReQmEflGRE6IyHER2e4eny8iX4vI7+52XqWuzxa8pJEcsENVVwH3AFtE\n5DZgJ7BXVW8F9rrfr0k2m2VoaNApy83XYOSfa0+rj+O7iDAwcJpsUF0/VR1Q1aPu/nngBE4h/CPA\nB26zD4BHS90rFovR0JBkYtpOGlyJqtLYOI9YUGmkGBFZCtwJHAQWq+qAa3QASE1xzWYR6RSRzuHh\n4XLMzTg8iy0iNwCfAC+o6qjX61R1j6q2qWpbKpWirq6O9T9+X4mvkePuYz9QV1fn+d0cT61EJIEj\n9Ieq+ql7eEhEmt3zzcCZCvydVXjpjQjwHnBCVXcXnfoceMbdfwb4zKvRi41JZsLY/VKj9zIG8NbP\nbgeeBrpE5Jh77BXgNeBjEXkW+BN4vCzLs5CSYqvqAaYOwQcrMZpMJlnx1i4AurfuqOQWobL8rTcA\naGhoKOu60MrP0uvWhGXaNyNrbgegqczr7LMRg4QW2StWrHC27ltZ+XdYosx9I/6m8mxkGyQyr3l0\nnDkFwP5U9F736BjqDeQ+kYlsiSeQeIKOoVNhu3IFHWf+QBJxJBFHVT2Xmk1GZMSeDUQmjeSRRKLw\nQxTmj2aH60PxAMPvig02sg0SucguZt63XwKwqOtXure8CPiPrmux/O3djLqDrWpYsZFtkEhH9urV\nq/M7zN20EYDEiLMgxsG19wJQg1TcQ8g/Vy9eSafcIXg5RFrsYurr6/M7AKzvPwlAOnOOplQzALlR\np2izuMStttWpU8nP6g8ODlBf7zxAqp0719lW2fc8No0YpOJ1/SoyJvI38C8wHSYjF+Ldz5tVdVGp\nRkbFBhCRTlX1tjZbiFTDT5tGDGLFNkgYYu8JwWYlBO6n8Zw9m7FpxCDGxI7yWtvXqNR9VUT6ReSY\n+9noy46JNBL1tbbdiq5mVT0qIvXAEZxC0SeAf1R1VxB2TEV2Ya1tVb0E5NfajgTXqNQNFFNiT7bW\nduD/mCC4qlIXYKuI/Cwi7/st+Dcltqe1tsNmkkrdd4BbgDXAAPC6n/ubErvqa237ZbJKXVUdUtVx\nVZ0A3sVJhxVjSuxIr7U9VaVuviTa5THgFz92jDzPruZa2wExVaXuUyKyBifl9QLP+TFiR5AGsSNI\ng1ixDWLFNogV2yBWbINYsQ1ixTaIFdsg/wFS9oIwUYxO3AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c8c30b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of the image:  (32, 32, 3)\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYnoAm6APfQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x12c8f7908>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of the image:  (32, 32, 3)\n",
      "\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAButJREFUeJztnF1sHNUVx3//eA00gFSwDQQcZw0y\nIuElSFahlHUkKiRUVQJUUcED4gGRPhQEEi+oT33kgfYVKVUj8YCEKoEED0gVIKQ4BYWYKOIrwiBq\nsCExjlMlzeeuvYeHmVl/7Mzu7M7s3Z3N/Kzx7t65O/f4v8dn7rn37pWZkeOGLd024HIiF9shudgO\nycV2SC62Q3KxHZKL7ZBEYkt6UNJXkr6R9GJaRvUrajepkTQAzAIPAAvAYeBxM/syPfP6i0KC9/4K\n+MbMvgWQ9DrwEBAp9vDwsBWLxQRNxmP17Lna84Frru54e3Nzc5w8eVLN6iUR+xZgft3rBeDuzZUk\n7QX2AoyNjTEzM5OgyXic/vAQRhWAX9776463Nzk5GateErHDPsm6mGRm+4B9vlEdHYg5cWIRgNnf\n/6FmyJ1fHwVgeGi4k03HIonYC8D2da9HgR+TmZOM2V1rHhZ4wpcTuwGYOrXQBYs2kqQ3chiYkDQu\n6QrgMeDtdMzqT9r2bDNbkfQM8G9gANhvZl+kZlkLnD5zOrCq7lxQcv78eQC2bt3qxqgQkoQRzOwd\n4J2UbOl7EondK3x6+12R54LYfWT8TgDuW/yvA4vCydN1h2Re7DOnz2DlMlYuN6xXrVSoVipcuHDB\nkWX1ZF7so+M7W6p/+JYJzIxuzL1mXuwskdkb5LlzwfhHfSI7tTyPmZeuHxjZ4dXyHVmiFkpcdwNz\nz3ZIZj37SHFX9EkJbwR4zaMDzOCTHV6cLy191ynzQsmc2EEmaKsrfslaGAkb/5ha9soODI3Wymx1\n1Xv0b5JS09HRVMjDiEMy59kzo7f7z2J6o1/NNr4E4MD1nrfv+d8PaZjWlNyzHZIZz754MTrzm1qe\njzy3uc700NoQfBCrL166BMBVV16ZxMSm5J7tkMx49sc3T4SWC/MylWY0qPLxttsAKC3Pd7RnkgnP\nvlQuex3mzZ1moHQq3s1N/k+j6bFOj5dkQux+IRNiH7rpVjB5h4/VjvS8cf3NsxNkQux+oadvkJVK\nJfLcngRLE4K4HSQ1AZJYWfGGAQqF9KXpabE/uunWurKwzkLYhzI4ONjgytGh56MbxwGvZ5I2eRhx\nSE96dq0LFtIVKy1vDB+FQoFVfxSvEdWqN5ng9aO9f4+wcBK0HUw+SOn5Y+7ZDulJz26lCxbHqwG2\nbPH8qlqtxsoSp4fGgHSzyqaeLWm7pA8kHZP0haTn/PLrJb0r6Wv/8bpULOpj4oSRFeAFM9sJ3AP8\nWdIu4EXgfTObAN73XycmylOnTi2EptpmRqFQqDuiCDw8zrXTpqnYZnbczI74z/8PHMNbCP8Q8Kpf\n7VXg4TQM+o8/G74eDQywljPWU6lU6o5gbYikRGEgzayypRukpCJwF3AIuNHMjoP3gQA3RLxnr6QZ\nSTNLS0vJrM04scWWdA3wBvC8mZ2J+z4z22dmk2Y2OTIyEl2xWvWOEO77aQ5DWNypsA2Xrda6fQ2s\nBIxSVCipmnckJJbYkgbxhH7NzN70ixclbfPPbwN+SmxNn9O06ycv4P0TOGZmf1936m3gSeAl//Gt\nJIYcCInVQW7e+aUGWve7nukRL25vTqhaJU4/+zfAE8Bnko76ZX/BE/lfkp4CvgceTWTJZUBTsc3s\nINEf+m+TGlCLpyGpeZyJ3LQJT+GDx402tvof1/UM8uDwWF2Zrmg0YueGsHUmga3tjgjmYyMO6Zpn\n28pK5LnSiW9Tb29iwpudLxaLsdb47WkwIlj1bd/S4gRD7tkO6YpnmxnTNxRDz2mwQOx1fC0wOzvb\n3htF3SjBQd/2qeWFlkztOc8uLc5124QNlJbnI7+D0+rMfs+J3c84DyNmxvRw/Uha4CObZ7x7gagb\n6fTQdkrL38e+Tu7ZDnHr2WZUy+XQYWk3X7ToACuV0Ow3jNyzHeLUs89+9jkfjjf4llfGEDA9egdn\nT8Xb08ap2Fft2snO997j+NPPgjUb0M8AEqP7X+EX998fq3oeRhzS9r5+bTUmLQHngJPOGm2fYeLb\nucPMGsz5eTgVG0DSjJnF25uti3TCzjyMOCQX2yHdEHtfF9psh9TtdB6zL2fyMOIQZ2L38l7bDVbq\n/lXSD5KO+sfvErXjIoz0+l7b/oqubWZ2RNK1wCd4C0X/CJw1s5fTaMeVZ9f22jazMhDstd0TNFip\nmyquxA7bazv1PyYNNq3UBXhG0qeS9idd8O9K7Fh7bXebkJW6rwC3AbuB48Dfklzfldg9t9f2ZsJW\n6prZopmtmvfVsX/ghcO2cSV2T++1HbVSN1gS7fMI8HmSdpyMZ/fSXtsRRK3UfVzSbryQNwf8KUkj\neQbpkDyDdEgutkNysR2Si+2QXGyH5GI7JBfbIbnYDvkZRNl0Hr94pVkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14300a278>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "shape of the image:  (32, 32, 3)\n",
      "\n"
     ]
    }
   ],
   "source": [
    "### Load the images and plot them here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "import glob\n",
    "\n",
    "image_paths = glob.glob('web_examples/*.jpeg')\n",
    "web_images = []\n",
    "for image_path in image_paths :\n",
    "    #print(image_path)\n",
    "    # cv2 reads as BGR\n",
    "    bgr_image = cv2.imread(image_path)\n",
    "    # convert to RGB\n",
    "    b,g,r = cv2.split(bgr_image)\n",
    "    image = cv2.merge([r,g,b])\n",
    "    # resize to 32x32\n",
    "    image = cv2.resize(image, (32, 32)) \n",
    "    # plot image\n",
    "    plt.figure(figsize=(1,1))\n",
    "    plt.imshow(image)\n",
    "    plt.show()\n",
    "    print(\"shape of the image: \", image.shape)\n",
    "    web_images.append(image)\n",
    "    print()\n",
    "\n",
    "X_web = np.asarray(web_images)\n",
    "y_web = np.array([3,9,17,11,19])  # manually populated based on the signnames table"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict the Sign Type for Each Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Predicted signs for each of the following images are :\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x108913978>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 3 Speed limit (60km/h)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1419cbfd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 9 No passing\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAABuRJREFUeJztnF1sFFUUx3+nu2st2G75WlpbEWLk\nQ0NEqGJsqonGREiI+qDRB+ODCT6AYEQT4pOPPiA+qDHBaOKDiTHRRB80xhAjoAYoiFbEViKNtrTF\nUnaLhsJue3yY2WXBlp3dmb0zbe8v2czszJ05hz+nZ++de+aKqmIxQ03YDswmrNgGsWIbxIptECu2\nQazYBrFiG8SX2CLysIh0i8hJEdkZlFMzFal0UCMiMaAHeAjoAw4DT6nqr8G5N7OI+7j2buCkqv4B\nICIfAY8AU4q9cOFCXbp0qQ+Tl8kHSS6XI5FIOMfGx53tpWyhnVznnJNYDIBsLkusxtmvqQkmi/b2\n9jI8PCyl2vkRuwX4q+h7H7D+6kYishnYDLBkyRI6Ozt9mLzM2NgYAJlMmsWpJgBymVEALvb3FdrV\ntrQAEG9sAGDozBD1Nzj7c+bMCcSXtrY2T+38iD3Z/+T/cpKq7gH2uE5V/CAmk84AUDuSBuBQW3vh\nXHc5NxJF1HH9rqPfATDWmAQgmUwiUjJAK8aP2H3ATUXfW4HT/ty5kq6uLgAW/XScnm0vBXNTlUJE\nHFrbfsWp5W/uIr3uDgBWrlwVjL0i/CStw8CtIrJMRK4DngQ+D8atmUnFka2qORHZCnwFxID3VfV4\nEE7lI+/c/RucbRA39UDP85f/elaO9BV8CSqx+EkjqOoXwBcB+TLj8SV2NZjIZjmweFnYbrBvfisA\nHYOnwO0++sUO1w0SGbE1m0UjEtXF7G9aVvANZZLOrXcik0b2R0zkYvK+dZx1xnBS4U9mZCJ7NhBa\nZPf09AAweM8DYblQNvsXOGO4Rd99DcCqVeUNfGxkGyS0yE4eOgrAYFgO+KDxyE/OTpmRHYrYmUyG\n7u0vh2E6EH7f5vg+Z9MGksmk5+tsGjFIKGLXjqRB1flMS5wO9/Xn0mVdZSPbIEbFnpiY4MKFCxxa\n11668TTg4Np2xsbG8DqPayPbIMYj+/z58yZNVhURIZ0+Ry6X89TeaNcvHo+TSqX4DYqeyFdvzq96\nOGlDVWla3FyY3S+FTSMGMRrZOj5OLuPMkt93tq9E6+izb34rudHRQr1KKWxkG8So2Hopy8W+QKsd\nQudif78zseABs5EtTM/fw4CwacQgRsWWRILalhtNmqw6tS0tiO36RQ+jXT+JxYi7z3/3LWjNHzXp\nQkBcfhYSTzYUypFLUTKyReQmEflGRE6IyHER2e4eny8iX4vI7+52XqWuzxa8pJEcsENVVwH3AFtE\n5DZgJ7BXVW8F9rrfr0k2m2VoaNApy83XYOSfa0+rj+O7iDAwcJpsUF0/VR1Q1aPu/nngBE4h/CPA\nB26zD4BHS90rFovR0JBkYtpOGlyJqtLYOI9YUGmkGBFZCtwJHAQWq+qAa3QASE1xzWYR6RSRzuHh\n4XLMzTg8iy0iNwCfAC+o6qjX61R1j6q2qWpbKpWirq6O9T9+X4mvkePuYz9QV1fn+d0cT61EJIEj\n9Ieq+ql7eEhEmt3zzcCZCvydVXjpjQjwHnBCVXcXnfoceMbdfwb4zKvRi41JZsLY/VKj9zIG8NbP\nbgeeBrpE5Jh77BXgNeBjEXkW+BN4vCzLs5CSYqvqAaYOwQcrMZpMJlnx1i4AurfuqOQWobL8rTcA\naGhoKOu60MrP0uvWhGXaNyNrbgegqczr7LMRg4QW2StWrHC27ltZ+XdYosx9I/6m8mxkGyQyr3l0\nnDkFwP5U9F736BjqDeQ+kYlsiSeQeIKOoVNhu3IFHWf+QBJxJBFHVT2Xmk1GZMSeDUQmjeSRRKLw\nQxTmj2aH60PxAMPvig02sg0SucguZt63XwKwqOtXure8CPiPrmux/O3djLqDrWpYsZFtkEhH9urV\nq/M7zN20EYDEiLMgxsG19wJQg1TcQ8g/Vy9eSafcIXg5RFrsYurr6/M7AKzvPwlAOnOOplQzALlR\np2izuMStttWpU8nP6g8ODlBf7zxAqp0719lW2fc8No0YpOJ1/SoyJvI38C8wHSYjF+Ldz5tVdVGp\nRkbFBhCRTlX1tjZbiFTDT5tGDGLFNkgYYu8JwWYlBO6n8Zw9m7FpxCDGxI7yWtvXqNR9VUT6ReSY\n+9noy46JNBL1tbbdiq5mVT0qIvXAEZxC0SeAf1R1VxB2TEV2Ya1tVb0E5NfajgTXqNQNFFNiT7bW\nduD/mCC4qlIXYKuI/Cwi7/st+Dcltqe1tsNmkkrdd4BbgDXAAPC6n/ubErvqa237ZbJKXVUdUtVx\nVZ0A3sVJhxVjSuxIr7U9VaVuviTa5THgFz92jDzPruZa2wExVaXuUyKyBifl9QLP+TFiR5AGsSNI\ng1ixDWLFNogV2yBWbINYsQ1ixTaIFdsg/wFS9oIwUYxO3AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14199fdd8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 17 No entry\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAABztJREFUeJztnF1sFFUUx3+HgiKtidqWD1tKC4EI\nJA1NGsWPFqIxIcQExWj0gWhixBeMJj5IDA++AQn6SoKRxBATI9FEH0yMGkOLSqGQhqpEIFilCKUU\nA/K5bff4MDPb7e5Md3Zn9u5uuT+ynd1778w9/Pfsmfs1V1QVixlmlNqAOwkrtkGs2AaxYhvEim0Q\nK7ZBrNgGiSS2iKwTkT9E5LSIbI3LqOmKFNqpEZEq4CTwNDAIHAFeVtXf4zNvejEzwrkPA6dV9QyA\niHwGbAACxa6rq9Pm5uYIVYZj/Nr11Puqmuqi1zcwMMClS5ckV7koYjcAZ9M+DwKPZBYSkc3AZoCm\npiZ6e3sjVBmOKz/3oCQBuO+xR4teX3t7e6hyUcT2+yazYpKq7gH2uEYVdSDmwoUhAE4+83zKkJWn\n+gCoq60rZtWhiCL2ILAw7XMj8E80c6JxcsWEh3me8PvSVQB0Xh4sgUWTidIaOQIsFZEWEbkLeAn4\nOh6zpicFe7aqjonIFuBboArYq6q/xWZZHly5esWzKivPS7lx4wYAc+bMMWOUD1HCCKr6DfBNTLZM\neyKJXS4cX9YWmOfF7mMtKwF4YuhPAxb5Y7vrBql4sa9euYomEmgiMWW55OgoydFRbt68aciybCpe\n7L6W5XmVP9KwFFWlFHOvFS92JVGxN8jr173xj+yObOfIWVSd7npX/SKnlOvIIqRCielmoPVsg1Ss\nZx9rXhGcKYIzAjzh0R6qcHSRE+c7hv8qlnm+VJzYXk9Qx8fclIkw4jf+0TnipHXVNqbSdHzcObo3\nSZGco6OxYMOIQSrOs3sbl7nvQnqjW0wnfwSg6wHH29f8ey4O03JiPdsgFePZt24F9/w6R84G5mWW\n6a6dGIL3YvWt27cBmH333VFMzIn1bINUjGcffnCpb7qgTk8lF1MUObxgCQAdI2eL2jKpCM++nUg4\nDebMRjPQcTn75lZTU0N1dTXV1RMz6+L+m2p6rNjjJRUh9nShIsJIz/zFZMYBTR0VcfPWrl0LpI+b\nwOrVqwE4dOhQznq6axcWtRloPdsgZe3Zo6OjgXlr0mLv/v37AThw4EBWuZ6eHgD27dsHwKZNm1Jx\n2+vUeIgIY2POMMDMmfFLU9Zi/zJ/cVaaX2Nh+/btOa+1a9cuwBHbbxY+Vee8FsBpmcSNDSMGKUvP\nTjXBfJpiHSNxrGxyfh5+4cSr25t8EInPH61nG6QsPTt9/CIu8u2wdNc2AfH2KnN6togsFJEfReSE\niPwmIm+56Q+IyHcicso93h+LRdOYMJ49BryjqsdE5F7gqIh8B7wK/KCqO9xHPLYC70Y1aNydRckk\ns5udr7f19/dnned5e1BTMG5yeraqnlfVY+77/4ATOAvhNwCfuMU+AZ6Nw6Cf3NnwdKSqCqe5Zn6t\nR5whLa8bpIg0A21ADzBPVc+D84UAcwPO2SwivSLSOzw8HM3aCie02CJSA3wBvK2qV8Oep6p7VLVd\nVdvr6+uDCyaTzsuHJy4OoAjK5BDgvdra2mhrC15cCdDa2kpra+uk89KsBJSOoBHBpDqviIQSW0Rm\n4Qj9qap+6SYPicgCN38BcDGyNdOcnDdIce4oHwMnVPXDtKyvgVeAHe7xqyiGdPnEaq9vXvylBpL2\nN5vueiduR+1QhWmNPA5sAvpFpM9New9H5M9F5DXgb+CFSJbcAeQUW1UPEvylPxXVgKQXp306HWEm\ncuPGvwvvHSfbmO8vruQ9yIN1TVlpctesSNdMF6XQEOS3zsSztdARQTs2YpCSebaOjQXmdVw4U9A1\nz5zJPm9w0AkL69evz+taa6YYEUy6ts/Ic4LBerZBSuLZqkr33GbfPJk1k9Dr+Fy2bdsGQEtLS1Ze\nQ0MDADt37szrmhMGkTVKcNC1vXNkMC9TS36DzKRjaCDvczZu3JizzLp16wqwxrkZdt3vhJLMm236\nzH4YbBgxiHHPVlW667JH0rxfarGHOQshqPnYXbuQjpG/Q1/HerZBzHq2KslEwndY2syDFkVgbNS3\n9+uH9WyDGPXsa/2/8nPLFE95VRgCdDc+xLXL4fa0MSr27BXLWf7995x//U1Q/4mCikKExr27uefJ\nJ0MVt2HEIAXv61dQZSLDwHXgkrFKC6eO8HYuUtUp5vwcjIoNICK9qhpub7YSUgw7bRgxiBXbIKUQ\ne08J6iyE2O00HrPvZGwYMYgxsct5r+0pVuq+LyLnRKTPfeU3t5ZZj4kwUu57bbsruhakr9TFWSj6\nInBNVXfFUY8pz07tta2qCcDba7ssmGKlbqyYEttvr+3Y/zNxkLFSF2CLiBwXkb1RF/ybEjvUXtul\nxmel7m5gCbAKOA98EOX6psQuu722M/FbqauqQ6o6rs6jYx/hhMOCMSV2We+1HbRS11sS7fIc8GuU\neoyMZ5fTXtsBBK3UfVlEVuGEvAHgjSiV2B6kQWwP0iBWbINYsQ1ixTaIFdsgVmyDWLENYsU2yP/l\nYnoAm6APfQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1592229b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 11 Right-of-way at the next intersection\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1419ba9b0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "sign id: 19 Dangerous curve to the left\n"
     ]
    }
   ],
   "source": [
    "### Run the predictions here and use the model to output the prediction for each image.\n",
    "### Make sure to pre-process the images with the same pre-processing pipeline used earlier.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "# preprocess\n",
    "\n",
    "X_web = preprocess(X_web)\n",
    "\n",
    "prediction = tf.argmax(logits, 1)\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    saver.restore(sess, max_valid_model_ckpt)\n",
    "    predicted_signs = sess.run(prediction, feed_dict={x: X_web})\n",
    "    \n",
    "print(\"Predicted signs for each of the following images are :\")\n",
    "for sign_id, web_img in zip(predicted_signs, web_images) :\n",
    "    # plot image\n",
    "    plt.figure(figsize=(1,1))\n",
    "    plt.imshow(web_img)\n",
    "    plt.show()\n",
    "    # print label\n",
    "    label = signnames[str(sign_id)]\n",
    "    print(\"sign id:\", sign_id, label)\n",
    "    "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Analyze Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy % : 100.0\n"
     ]
    }
   ],
   "source": [
    "### Calculate the accuracy for these 5 new images. \n",
    "### For example, if the model predicted 1 out of 5 signs correctly, it's 20% accurate on these new images.\n",
    "num_correct = 0\n",
    "for i, j in zip(predicted_signs, y_web) :\n",
    "    if i == j :\n",
    "        num_correct += 1\n",
    "\n",
    "accuracy_pc = 100.0 * num_correct / len(y_web)\n",
    "print(\"Accuracy % :\", accuracy_pc)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Output Top 5 Softmax Probabilities For Each Image Found on the Web"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "For each of the new images, print out the model's softmax probabilities to show the **certainty** of the model's predictions (limit the output to the top 5 probabilities for each image). [`tf.nn.top_k`](https://www.tensorflow.org/versions/r0.12/api_docs/python/nn.html#top_k) could prove helpful here. \n",
    "\n",
    "The example below demonstrates how tf.nn.top_k can be used to find the top k predictions for each image.\n",
    "\n",
    "`tf.nn.top_k` will return the values and indices (class ids) of the top k predictions. So if k=3, for each sign, it'll return the 3 largest probabilities (out of a possible 43) and the correspoding class ids.\n",
    "\n",
    "Take this numpy array as an example. The values in the array represent predictions. The array contains softmax probabilities for five candidate images with six possible classes. `tf.nn.top_k` is used to choose the three classes with the highest probability:\n",
    "\n",
    "```\n",
    "# (5, 6) array\n",
    "a = np.array([[ 0.24879643,  0.07032244,  0.12641572,  0.34763842,  0.07893497,\n",
    "         0.12789202],\n",
    "       [ 0.28086119,  0.27569815,  0.08594638,  0.0178669 ,  0.18063401,\n",
    "         0.15899337],\n",
    "       [ 0.26076848,  0.23664738,  0.08020603,  0.07001922,  0.1134371 ,\n",
    "         0.23892179],\n",
    "       [ 0.11943333,  0.29198961,  0.02605103,  0.26234032,  0.1351348 ,\n",
    "         0.16505091],\n",
    "       [ 0.09561176,  0.34396535,  0.0643941 ,  0.16240774,  0.24206137,\n",
    "         0.09155967]])\n",
    "```\n",
    "\n",
    "Running it through `sess.run(tf.nn.top_k(tf.constant(a), k=3))` produces:\n",
    "\n",
    "```\n",
    "TopKV2(values=array([[ 0.34763842,  0.24879643,  0.12789202],\n",
    "       [ 0.28086119,  0.27569815,  0.18063401],\n",
    "       [ 0.26076848,  0.23892179,  0.23664738],\n",
    "       [ 0.29198961,  0.26234032,  0.16505091],\n",
    "       [ 0.34396535,  0.24206137,  0.16240774]]), indices=array([[3, 0, 5],\n",
    "       [0, 1, 4],\n",
    "       [0, 5, 1],\n",
    "       [1, 3, 5],\n",
    "       [1, 4, 3]], dtype=int32))\n",
    "```\n",
    "\n",
    "Looking just at the first row we get `[ 0.34763842,  0.24879643,  0.12789202]`, you can confirm these are the 3 largest probabilities in `a`. You'll also notice `[3, 0, 5]` are the corresponding indices."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TopKV2(values=array([[  9.97144878e-01,   1.48523925e-03,   1.08707347e-03,\n",
      "          2.29188387e-04,   2.60076358e-05],\n",
      "       [  9.05143559e-01,   8.71173516e-02,   7.70291500e-03,\n",
      "          3.60482336e-05,   1.94363821e-07],\n",
      "       [  1.00000000e+00,   3.38034022e-21,   3.05689849e-23,\n",
      "          5.96438702e-27,   2.84115184e-28],\n",
      "       [  9.99998093e-01,   1.71179101e-06,   1.40379228e-07,\n",
      "          7.17800432e-08,   5.22941779e-09],\n",
      "       [  9.99998093e-01,   1.66717109e-06,   1.98016849e-07,\n",
      "          4.16706669e-10,   1.83252788e-10]], dtype=float32), indices=array([[ 3, 36,  5, 38, 35],\n",
      "       [ 9, 41, 20, 10, 35],\n",
      "       [17, 22, 14, 34, 13],\n",
      "       [11,  1, 30, 21, 34],\n",
      "       [19,  2, 23, 30, 34]], dtype=int32))\n"
     ]
    }
   ],
   "source": [
    "### Print out the top five softmax probabilities for the predictions on the German traffic sign images found on the web. \n",
    "### Feel free to use as many code cells as needed.\n",
    "softmax = tf.nn.softmax(logits)\n",
    "top_5 = tf.nn.top_k(softmax, k=5)\n",
    "with tf.Session() as sess:\n",
    "    saver.restore(sess, max_valid_model_ckpt)\n",
    "    top_5_out = sess.run(top_5, feed_dict={x: X_web})\n",
    "    print(top_5_out)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x143d9f9e8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdAAAAEWCAYAAADW7MapAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nQCMiImpIgEZERNSQAI2IiKjh/wMLyjTI1g898AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14312d518>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAACLpJREFUeJztnGtsFNcVgL/jB8YQezE2BmMbDBGu\no4pigoNpeKRqVTVJqWh/tEp+VFSqRH6EJBVENIpUqVKlKFSQPxGqStVI/VGpqtRKjSBVg/qykwaD\nyzOE2jwLNmDXMV7TigWvffpjZtbrx87M7s7e3cB8krWzM2fvOT579t6595y5oqqEmKEo3wY8TITO\nNkjobIOEzjZI6GyDhM42SOhsg2TlbBF5WkR6ROSiiLwWlFEPKpLppEZEioFe4KtAH3AceF5VPwnO\nvAeLkiw+ux64qKqXAUTkN8A2IKWza2pqtKmpKQuVkzhBEo/HKS0ttc6Nj1uv98cScjLHuibFxQCM\nxccoLrKOi4qC6UWvXr3K0NCQeMll4+x64HrS+z6gfbqQiOwAdgAsW7aM7u7uLFROEovFAIhGR1hc\nuwSAeHQUgHv9fQm5svp6AEoWVAIwMDhAxSPW8bx58wKxpa2tzZdcNs6e7Zuc0Sep6kHgoG1Uxgsx\n0ZEoAGXDIwAca9uYuNaTTkOiiFqmP3HiQwBiCyIARCIRRDwDNGOycXYf0Jj0vgG4kZ05Uzl79iwA\ni06fo/flV4NpVCUREcce3zjlUvPb+xhZtwaAlpbHgtGXRDad1nFglYisEJE5wHPAu8GY9WCScWSr\nalxEdgJ/AoqBd1T1XBBGOZF3+6lnrNcgGvVB70uTv56W4T4XyczIphtBVd8D3gvIlgeerJydCybG\nxvhg8YrA23UGPr/zio6FDQBsHrgCJSVT2siUcLpukIKJbB2zJiLpRnX7mS7KGup9y9/r66frC9Z0\nwAlyt4DtXLzCim4Ae/KUKQXj7E6fTt6S5cBV1lA/ow2ny0iFY1u2usNuxCB5i+ze3l4Abm34srew\nCFs+ve4tlyFOxHZWN7oOoM4voPbDIwC0PJbexCeMbIPkLbIru6wFqVtuQvbItfnT64mJTu5WLiw9\nHdX2CoRLhFedPGMdpBnZeXF2NBql95UfespNdh1Kbt08U6fboNmzczcAc7/+NJFIxHfbYTdikLxE\n9pzhEWZZjbURtgxfm3HONFuGnQhvTCkzdyQKYWQXJkadPTExwd27dzm+bqOLlNM/m4/mqXjb0LX2\nSWKxmO/1ljCyDWK0z56YGOfOnVFEJGU0tJ/pSnpnydy9dBnXKLNvEctXpp7yX7hwwWrRIwqbm5un\n2nO6i641M1KrCW7fHiYej7u26WDU2SUlpdTWLuZfLv9wWUN9Yuw8+bVtAKx93z0BpNMPkr4Xx7mr\nVq3yZeOGDRsAOHr0qGVPo/siV13d0kR234uwGzGI0cjW8XHidpbcFbGicWxk1Fe7TiB3VltRuHm4\nP3HNqQ3xO4hFoz7sSyI+OpqoV/EijGyDmI3s+2PE+r2rHT5+bjsAsYsXAe/15gSSOnbSTWlt3boV\ngEOHDrnK3eu7kUh8eFEwyYNkht//S2YfDPBhrMOHDwfWlkPYjRjEaGTLnFLm+sgXNr2+B4B/v7EX\ngAl17wJEgl+A3bt3ry+5svqlSHjrV3iYjeziYkoilZ5yy159GYCrb/wUgKduuyda+w78HIDLP/pJ\nSpn9+/cDsGvXLncb7YF0z549nnYClEQqE+XIXnhGtog0ishfReS8iJwTkVfs8wtF5IiIXLBfq3xp\nfIjxE9lxYLeqnhCRCuCfInIE+B7wZ1V9037E4zXANf0Sj8cZHBx0VXbven9iivzE8Q4AFEVc+uOG\nF18AoGbrMzOuXbp0CYCVK1e66nXo6ZlagBy7Nn1tfSo3b91gLKhbP1W9Cdy0j++IyHmsQvhtwJds\nsV8Bf8PD2UVFRVRUVLjq61rTnsh2lz/qz0EOc5cvm3HOr5Mdpi9EHWt90lV+QaSK4qC6kWREpAlY\nC3QBi+0vwvlCalN8ZoeIdItI99DQUDrqHjh8O1tEHgF+B/xAVf0tWmA9eaCqbaraVltbS3l5eaLi\n3+VTpE6bmUFVPddT1p/6iPLyct/P5viSEpFSLEf/WlV/b58eEJE6+3od4N4Zh/i6GxHgl8B5VX0r\n6dK7wHb7eDvwB79K71ctwC3t1LGw0TXRaoLO6kY6q91tuL/Af7IX/N2NbAS+C5wVkVP2udeBN4Hf\nisj3gWvAt9PS/BDi527kA1LPg7+SidJIJELz2/sA6H1pd0o5pzopl3V+MDXB0+ERzQDNB6wfeGWl\n9wQtmbyt+jlPZbliD1Ad1Y25dbijp2a5r5XD4dbPA7AkTTXh2ohB8hbZLS0tAHzOKdd1SxCoJhII\n2Rakz4bXQOgQFsN/hsh7psYZeTfZz614PVPjRKEz4Vh/+iPmNvq/TYxd7+PYmg1p2bhp8Epa8qnI\nu7MdiuwF+E0DV1wdPn1Wd2zNFyffON9cgJPPzYOXEfvRPEd3po/ohd2IQQomsh2KSksTA5HvrLpD\nkBFt25Acw9lu2BBGtkEKLrKTqfr7HwFYdPYTel600lm53A+k+cBbjK5rtfTkoP0wsg1S0JG9evVq\n54D533gWgNJha0OMrsetDEoRqcuPvWg/+Q9g6k466U7B06GgnZ1MIp1mv7b3W6VpI9HbLKmtAyA+\nahVF3uubLHEra1gKQIn97MutWzepqLAWkMrmz7dec2y7Q9iNGCTjff0yUibyH+B/wGchGVmDfzuX\nq+oiLyGjzgYQkW5V9bc3Wx7JhZ1hN2KQ0NkGyYezD+ZBZyYEbqfxPvthJuxGDGLM2YW817ZLpe6P\nRaRfRE7Zf89mpcdEN1Loe23bFV11yZW6wDeB7wD/VdV9QegxFdmJvbZV9T7g7LVdEKjqTVU9YR/f\nAZxK3UAx5ezZ9toO/J8JgmmVugA7ReSMiLyTbcG/KWf72ms738xSqfsz4FGgFatGfX827Ztyds73\n2s6W2Sp1VXVAVcdVdQL4BVZ3mDGmnF3Qe22nqtR1SqJtvgV8nI0eI+vZudxrOyBSVeo+LyKtWF3e\nVeCFbJSEM0iDhDNIg4TONkjobIOEzjZI6GyDhM42SOhsg4TONsj/AXkoyUXa+su9AAAAAElFTkSu\nQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x1430b45c0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAkMAAAEWCAYAAABlkpNCAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x14137a400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x141373400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAbQAAAEWCAYAAAAO4GKjAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAG8xJREFUeJzt3XmYZVV97vHvK4MgzeC1UUGBNlEk\nCglgoUKQMRo0xCESEVGBoFxNDNEkXrkxijhEDRGDEq9BRIYYJhOUGCNGmUSGUAw2LSIigggCjTI1\nAjL87h97t54uq7uqm6pT1Krv53nq6T2svfdvnSrOW2vtzalUFZIkzXaPm+kCJEmaCgaaJKkJBpok\nqQkGmiSpCQaaJKkJBpokqQkGmvQYk2Rekv9KcneSE2e6numSZK0kleTpq3j8LUl2XM6+30vy7fHa\nJjksyVErOO+BSf5jVWrSzDLQNKslWTLw9UiS+wbW953ia52c5IGB89+5grZrJflEkpv6ttcl+egk\nL7UPMA94YlW9Iclbknx9SjrxKPW1PNT36e4klyXZY6brGquqvl5Vv7OcfYdW1dsAkmyR5KEx+z9b\nVX84jDo1tQw0zWpVNW/pF/Aj4A8Htn1+Gi75gYHzb7CCdocCvwVsC6wL/B6wcJLX2Az4XlU9/OhK\nnTbn9K/3E4GTgNOSzBvbKMnqQ69Mc5qBpqYlWTvJPyX5SZIfJzk8yRr9vj2SXNtPQf2sH0X98RRd\nejvg36rq1upcNxiwSbZK8s0kdyZZmOSl/faPAv8H2K8fBb0b+Edgl379lr7dyUmOTPLfSe5Nck6S\nJyf5VH/O7yTZauB6703ywyT3JFmU5A8G9n0uyWBtRyb5z4k62AfusXSjyQUDr+d7ktwK/L/+fH+W\n5AdJfprk35M8ZcypXpnk+iSLk3woSfrjtuj79bN+3/FJ1h1z7A5Jru7bHJ3k8f2xeyS5dry6k3wk\nyTH96nnAagOj7m3GjoiTbJnkrCR3JPluklcO7HtFf/17ktyY5OCJXjdNHwNNrTsM+G1gK+B5wC50\ngbHUAmBN4KnAQcDxSZ6xgvO9o39jHk3y8hW0uwh4V//m+NzBHUnWAr4MfBHYEHgn3SjnGVX1LuAI\n4Ph+FPgh4O30o6KqeurAqfYG/hqYD6zeX/Nc4EnAV4C/H2j7PWAHYH3go8DJSeb3+w6mC4bXJtmd\nbsrzwBX0bWk/Vu/b3QX8sN+8AFgD2AQ4OMnLgPcArwKeBtwO/MuYU/0hsDXw/P7ag1PF76f73mwF\nPBt495hj9wF26/dtQ/daroydgIcHRt2Xj+njesB/A5+le53fCByb5Jl9k2OBN1bVun0fvrmS19cU\nMtDUun2BQ6vq9qq6Ffgg8IaB/Q8Bh1XVL6rq68DXgb2Wc67DgWfSvcF+APjXJCPLaXsY3chqP+Cy\nfnS4T7/vRf2/R1TVg1V1Jt2b5t4r2bfTqurbVXUf8CXgrqo6pR85nUr3Bg9Av/0nVfVIVZ0I3EQX\n8FTVPXRv1EcBxwNvqapbVnDdnfv7h7cArwBeWVX39vseoJuW/UVf177A0VW1sKrup/tlYvckg8H8\n4aq6s6p+2NewT1/X1VV1Vn+uW+hez53H1HJkVd1cVYuBDy89dgq9ClhUVZ+vqoer6hLgP4BX9/sf\nAp6bZN2q+unYQNRwGWhqVj919VTghoHNN9CNFJZa3L/RDu7feLzzVdWlVXVHH0JfAr5A94Y3XtsH\nq+rIqtqe7l7TEcAJSX6zP/+PatlPBh9b12TcOrB83zjrv7yvle7JvYX9dOSddME8f6D9+cBPgPuB\n0ye47rlVtUFVza+q362qcwb23VJVDw6sb8zA619VdwJ3s2xfbxxY/uXrn2TjJKele7DmbuCYMTUv\n99gptBmw09LXrX/tXg1s1O9/Zb/+o35acnm/4GgIDDQ1qw+MW+jelJbalG50stT8fgpwcP/Nk70E\nkEnU8fOqOoJu9LJFf/5NxzQbW9fY66yyJJsDn6SbUv1f/cMs17Js7X8JPEgXNm9/FJcbW+vNDLz+\nSdYH1mPZvm4ysDz4+h8O3AtsWVXrAW/i11/v5R27qvWOdSPwtT7Al37Nq6q3A1TVhVW1J/AU4Gt0\nD8lohhhoat1JwKFJnpTkyXT3YAbv4awBvCfJmkl2A14M/NvYkyRZPcmrkqyTZLX+oYo/ppt++jVJ\n/irJi9I9vr9GkoOA1YBv091neVySt/fnfTHwEuC05fThVmCT9A+zrIJ5wCPA4v66b6EboS2tdUvg\nb4HX91/vTfKcVbzWWCcBb+4frFiL7v7dWWOmNN+VZP0kC4C3Aaf029cFlgB3J9mULnTHOjjJRv39\nwEMGjp2s2+geChn7C8ZSXwS2SbJ3/31cM8kLk2ze/yy8tr/P9iBwD/BYfTJ1TjDQ1Lr3AlcB3wGu\nAL7Fsg9LXE93H+QWuhv8B1TVdeOcJ3QPYNwM3EF3L26/qrpwOdd9APgEXRjdBhxAd6/px/0U5550\n9+p+SjcduXdV/WA55/pqX+dtSX48cZeXVVWXAZ8GRummFZ/RL5NkTbqAP6yqrqqqq+gexDjxUQTo\n4LW/THdv6wy61+6pLHsPE+A/6YJ+lC7Ul/7C8V5gR7qHTk5nnF80gJOBs4HvA1ey7Pd2MvXd0R9z\naT+luPU4+3+f7vv3k74PH6T7RQjgT+imOu+iuw+538pcX1Mr/oFPzVXp/ofgo6rqmRM2lvSY5whN\nktQEA02S1ASnHCVJTXCEJklqgh8eOkTz58+vBQsWzHQZkjSrXHrppbdX1YYTtTPQhmjBggWMjo7O\ndBmSNKskuWHiVk45SpIaYaBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCg\nSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmS\nmmCgSZKaMOcCLZ3zk7x0YNtrknw1yQWTOH7JcrYfl2SvqaxVkjR5q890AcNWVZXkLcBpSc4GVgM+\nBOxRVT+Y2eokSatqzgUaQFUtSvIfwLuAdYATquoHSZZU1TyAJO8EXgM8Hji9qg4dPEeSAJ8EdgN+\nCGSYfZAkLWtOBlrvMOAy4BfAyOCOJC8BngU8ny6ozkiyU1WdN9DsVcCzga2ApwBXAceOvUiSg4CD\nADbddNOp74UkCZjDgVZV9yY5BVhSVQ+M2f2S/uvyfn0eXcANBtpOwElV9TBwc5KzlnOdo4GjAUZG\nRmoKuyBJGjBnA633SP81VoAPV9U/T3C8ASVJjxFz7inHSToT+JMkS++nPS3Jk8e0OQ94bZLVkmwE\n7DrsIiVJvzLXR2jjqqqvJfkt4MLu2Q+WAK8HbhtodjrdAyFXAtcA5w67TknSr6TKWbNhGRkZqdHR\n0ZkuQ5JmlSSXVtXIRO2ccpQkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCT\nJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1\nwUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFAkyQ1wUCTJDXBQJMkNcFA\nkyQ1YYWBluRJSa7ov25JctPA+ppTVUSS1ZPcOYl2f5nku0lOSLJbkheu5HXelOQfV73S5Z73mUmu\nmOrzSpImb/UV7ayqnwJbAyR5H7Ckqv5hsidPsnpVPfSoKlzWnwK7VtWNST4I3A5cNIXnlyTNUqs0\n5Th2RJLkkCR/2y+fn+RDSc4D3pbkX5IcmeSCJNcledUkzn9Ikv9JsjDJe/ttxwCbAl9J8hfAm4B3\n9qPFHcYc/8IkFya5PMm3kjxrYPfTk5yZ5PtJPjxwzEv7Yy5LckqSdfrthyW5JMmiJJ9Okn77dn19\nFwJvWZXXUZI0dabrHtp6VbVTVS2d3nsy8LvAK4EPL/8wSPIyuuB6Ad3ocIckO1TVm4DbgBdV1ZHA\nMcDhVbV1VV0w5jTfBXasqm2ADwAfHNj3O8BewG8Dr0+ycZInA4cAu1fVtsBC4C/69kdW1XbAVsD6\nwB799uOAt1bV9sBqK+jPQUlGk4wuXrx4RV2XJD0KK5xyfBROHrP+xaoqYGGSp01w7EuAlwKX9+vz\ngM2BsaG1IhsAJyT5zXH2fb2q7gFIcjVdeD4VeA5wQT8AWxM4v2+/e5J3AmsB84FLk1wCrF1V3+rb\nnAjsOl4hVXU0cDTAyMhIrUQfJEkrYVUD7SGWHd2t1W9b6t4x7R8YWM4E5w7wwar67CrWBvAh4Myq\n+lSSZwJfXU4tD9O9BgG+WlVvWKaQ5AnAUcC2VXVTf99urX634SRJjyGrOuV4C7BxkicmWQv4gyms\n6UzgwIF7WE9PMn+cdvcA6y7nHOsDN/XL+0/imhcAOyf5jf6a6/T33dYGHgFuT7Iu8GqAqroduD/J\n9v3x+07iGpKkabRKgVZV9wN/B1wCnAFcNVUFVdVXgC8AFyW5EjiVbtpxrC8Br+kf/NhhzL6PAocn\n+dY4x413zVuBA4FTknybLuA275/yPB5YBJwOXDxw2AHAP/cPhSyZdAclSdMi3a0tDcPIyEiNjo7O\ndBmSNKskubSqRiZq5yeFSJKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmS\nmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppg\noEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSZKaYKBJ\nkpowpwMtybuTfCfJwiRXJHlBkrcnecJM1yZJWjmrz3QBMyXJ9sCewLZV9UCS+cCawCnAvwA/n8n6\nJEkrZy6P0DYCbq+qBwCq6nZgL2Bj4OwkZwMk2SfJlUkWJfno0oOTLEnysSSXJflGkg1nohOSpM5c\nDrSvAZskuSbJp5LsXFWfAG4Gdq2qXZNsDHwU2A3YGtguySv749cBLquqbYFzgUPHu0iSg5KMJhld\nvHjxtHdKkuaqORtoVbUEeB5wELAYOCXJ/mOabQecU1WLq+oh4PPATv2+R+imJ6GbotxxOdc5uqpG\nqmpkww0dxEnSdJmz99AAquph4BzgnCRXAvuNaZKVOd1U1SVJWnlzdoSW5NlJnjWwaWvgBuAeYN1+\n28XAzknmJ1kN2IduehG6126vfvl1wPnTX7UkaXnm8ghtHvDJJBsADwHX0k0/7gP8V5Kf9PfR/i9w\nNt1o7StV9aX++HuB5ya5FLgL2HvoPZAk/VKqnClbFUmWVNW8lTlmZGSkRkdHp6skSWpSkkuramSi\ndnN2ylGS1BYDbRWt7OhMkjS9DDRJUhMMNElSEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMMNElS\nEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMM\nNElSEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMMNElSEww0SVITDDRJUhMMNElSEww0SVITpj3Q\nkjyc5Iok305yWZIdpvuaw5ZkQZJFM12HJM1lqw/hGvdV1dYASX4f+DCw8xCuO2lJVquqh2e6DknS\nqhv2lON6wB0ASXZJ8uWlO5IclWT/fvn6JH+X5MIko0m2TXJmkh8kecvA8eclOT3JVUk+neRxSQ5M\n8vGB8745yRFjC0myJMn7k1wMbJ9k9ySXJ7kyybFJHt+3e2+SS5IsSnJ0kvTbn9ePOi8E/mz6XjJJ\n0mQMI9DW7qccrwaOAT4wyeNurKrtgW8CxwF7AS8E3j/Q5vnAXwFbAb8J/BFwMvDyJGv0bQ4APjfO\n+dcBFlXVC4DR/hp7V9VWdCPXt/btjqqq7apqS2BtYM9+++eAg/salyvJQX0ojy5evHiSXZckraxh\nBNp9VbV1VW0B7AGcsHSUM4Ez+n+vBC6uqnuqajFwf5IN+n3/U1XX9dOFJwE7VtW9wFnAnkm2ANao\nqivHOf/DwL/1y88GflhV1/TrxwM79cu7Jrk4yZXAbsBzk6wPbFBV5/ZtTlxeJ6rq6KoaqaqRDTfc\ncBLdliStimHcQ/ulqrowyXxgQ+Ahlg3UtcY0f6D/95GB5aXrS+uusZfo/z0G+BvgasYfnQHcP3Df\nbNyATbIW8ClgpKpuTPK+vs6Mc21J0gwa6j20fsS0GvBT4AbgOUke3494dl+FUz4/yTOSPA7YGzgf\noKouBjYBXkc3cpvI1cCCJM/s198AnMuvQvb2JPPopj2pqjuBu5Ls2O/fdxVqlyRNoWGM0NZOckW/\nHGC/fmR0Y5JTgYXA94HLV+HcFwIfobuHdh5w+sC+U4Gtq+qOiU5SVfcnOQA4LcnqwCXAp6vqgSSf\noZv2vL7fvtQBwLFJfg6cuQq1S5KmUKpm58xZkl2Av66qPZez/8vAx6vqG0MtbAVGRkZqdHR0psuQ\npFklyaVVNTJRu+Y+KSTJBkmuoXsY5TETZpKk6TXUh0KmUlWdA5wzzvY7gc2HXY8kaWY1N0KTJM1N\nBpokqQkGmiSpCQaaJKkJBpokqQkGmiSpCQaaJKkJBpokqQkGmiSpCQaaJKkJBpokqQkGmiSpCQaa\nJKkJBpokqQkGmiSpCQaaJKkJBpokqQmz9i9Wz0a33n0/H//va2a6DEkaqne8ePOhXMcRmiSpCQaa\nJKkJBpokqQkGmiSpCQaaJKkJBpokqQkGmiSpCQaaJKkJBpokqQkGmiSpCQaaJKkJBpokqQmzJtCS\nVJKPDaz/dZL3TfM1/2Y6zy9JmjqzJtCAB4A/SjJ/iNccN9DSmU2vnSQ1bza9KT8EHA28Y+yOJJsl\n+UaShf2/m47TZp0kxya5JMnlSV7Rb98/yb8n+WqS7yf5+377R4C1k1yR5PNJFiT5bpJPAZcB70ny\n8YHzvznJEdPUd0nSBGZToAH8E7BvkvXHbD8KOKGqfhv4PPCJcY59N3BWVW0H7AocnmSdft/WwN7A\nVsDeSTapqkOA+6pq66rat2/37P462wD/ALw8yRr9vgOAz429aJKDkowmGb33rjtWtd+SpAnMqkCr\nqruBE4CDx+zaHvjXfvlEYMdxDn8JcEiSK4BzgLWApSO5b1TVXVV1P3AVsNlySrihqi7qa7kXOAvY\nM8kWwBpVdeU4NR9dVSNVNbLO+k+cZE8lSStrNv7F6n+km/L7tdHQgBpnW4BXV9X3ltmYvIDu/txS\nD7P81+XeMevH0N1nu3qCeiRJ02xWjdAAqupnwKnAgQObLwBe2y/vC5w/zqFnAn+eJABJtpnE5R4c\nmFIcr5aLgU2A1wEnTeJ8kqRpMusCrfcxYPBpx4OBA5IsBN4A/MU4x3wAWANYmGRRvz6Ro/v2n19B\nm1OBb1WVN8gkaQbNminHqpo3sHwr8ISB9euB3SY4/j7gf4+z/TjguIH1PQeW3wW8a6D5luOcekfg\n4+NslyQN0Wwdoc24JBskuYbuSchvzHQ9kjTXzZoR2mNNVd0JbD7TdUiSOo7QJElNMNAkSU0w0CRJ\nTTDQJElNMNAkSU0w0CRJTTDQJElNMNAkSU0w0CRJTfCTQoboKeutxTte7IeLSNJ0cIQmSWqCgSZJ\naoKBJklqgoEmSWqCgSZJaoKBJklqgoEmSWqCgSZJaoKBJklqQqpqpmuYM5LcA3xvpuuYQfOB22e6\niBkyl/sO9t/+P7r+b1ZVG07UyI++Gq7vVdXITBcxU5KMztX+z+W+g/23/8Ppv1OOkqQmGGiSpCYY\naMN19EwXMMPmcv/nct/B/tv/IfChEElSExyhSZKaYKBJkppgoE2DJHsk+V6Sa5McMs7+xyc5pd9/\ncZIFw69yekyi73+Z5KokC5N8I8lmM1HndJmo/wPt9kpSSZp6lHsy/U/ymv5n4DtJ/nXYNU6nSfz8\nb5rk7CSX9/8NvGwm6pwOSY5NcluSRcvZnySf6F+bhUm2nfIiqsqvKfwCVgN+APwGsCbwbeA5Y9r8\nKfDpfvm1wCkzXfcQ+74r8IR++a2t9H2y/e/brQucB1wEjMx03UP+/j8LuBx4Yr/+5Jmue8j9Pxp4\na7/8HOD6ma57Cvu/E7AtsGg5+18G/BcQ4IXAxVNdgyO0qfd84Nqquq6qfgGcDLxiTJtXAMf3y18A\ndk+SIdY4XSbse1WdXVU/71cvAp4+5Bqn02S+9wAfAP4euH+YxQ3BZPr/ZuCfquoOgKq6bcg1TqfJ\n9L+A9frl9YGbh1jftKqq84CfraDJK4ATqnMRsEGSjaayBgNt6j0NuHFg/cf9tnHbVNVDwF3Ak4ZS\n3fSaTN8HHUj3G1srJux/km2ATarqy8MsbEgm8/3fHNg8ybeSXJRkj6FVN/0m0//3Aa9P8mPgK8Cf\nD6e0x4SVfX9YaX701dQbb6Q19v+NmEyb2WjS/UryemAE2HlaKxquFfY/yeOAjwP7D6ugIZvM9391\numnHXehG599MsmVV3TnNtQ3DZPq/D3BcVX0syfbAiX3/H5n+8mbctL/vOUKbej8GNhlYfzq/Pq3w\nyzZJVqebeljRUH22mEzfSfJ7wLuBl1fVA0OqbRgm6v+6wJbAOUmup7uPcEZDD4ZM9mf/S1X1YFX9\nkO7Dup81pPqm22T6fyBwKkBVXQisRffBvXPBpN4fHg0DbepdAjwryTOSrEn30McZY9qcAezXL+8F\nnFX9XdNZbsK+91Nu/0wXZi3dP4EJ+l9Vd1XV/KpaUFUL6O4hvryqRmem3Ck3mZ/9L9I9GESS+XRT\nkNcNtcrpM5n+/wjYHSDJb9EF2uKhVjlzzgDe2D/t+ELgrqr6yVRewCnHKVZVDyV5G3Am3VNPx1bV\nd5K8HxitqjOAz9JNNVxLNzJ77cxVPHUm2ffDgXnAaf1zMD+qqpfPWNFTaJL9b9Yk+38m8JIkVwEP\nA++sqp/OXNVTZ5L9/yvgM0neQTfdtn8jv8yS5CS6qeT5/T3CQ4E1AKrq03T3DF8GXAv8HDhgymto\n5LWUJM1xTjlKkppgoEmSmmCgSZKaYKBJkppgoEmSmmCgSQ1K8u7+0+wXJrkiyQuSHJPkOTNdmzRd\nfGxfakz/kUpHALtU1QP9/8C8ZlU180G40ngcoUnt2Qi4fenHilXV7VV1c5Jzln7MVpIDk1zTb/tM\nkqP67cf1f7PqgiTXJdmr375RkvP60d6iJC+asd5Jy2GgSe35GrBJH1ifSrLMB0An2Rh4D91nSb4Y\n2GLM8RsBOwJ7Ah/pt70OOLOqtgZ+B7hiGuuXVomBJjWmqpYAzwMOovucwFOS7D/Q5PnAuVX1s6p6\nEDhtzCm+WFWPVNVVwFP6bZcAByR5H7BVVd0znX2QVoWBJjWoqh6uqnOq6lDgbcCrB3ZP9MdkB/8C\nQvrznUf3F4lvovsc0jdOZb3SVDDQpMYkeXaSwT/JsjVww8D6/wA7J3li/+eLXs0EkmwG3FZVn6H7\ncO1tp7JmaSr4aftSe+YBn0yyAfAQ3aebHwR8AaCqbkryd8DFdH+P6iq6v5q+IrsA70zyILAEcISm\nxxwf25fmoCTzqmpJP0I7ne5PnZw+03VJj4ZTjtLc9L4kVwCLgB/S/eFNaVZzhCZJaoIjNElSEww0\nSVITDDRJUhMMNElSEww0SVIT/j88pN3MXdRE7AAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x116a636d8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYnoAm6APfQAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x143de5780>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAh8AAAEWCAYAAADGuvWEAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nMDMzs6wcPszMzCwrhw8zMzPL6v8D/7//yCgDXlEAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x159726470>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMS4wLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvpW3flQAAButJREFUeJztnF1sHNUVx3//eA00gFSwDQQcZw0y\nIuElSFahlHUkKiRUVQJUUcED4gGRPhQEEi+oT33kgfYVKVUj8YCEKoEED0gVIKQ4BYWYKOIrwiBq\nsCExjlMlzeeuvYeHmVl/7Mzu7M7s3Z3N/Kzx7t65O/f4v8dn7rn37pWZkeOGLd024HIiF9shudgO\nycV2SC62Q3KxHZKL7ZBEYkt6UNJXkr6R9GJaRvUrajepkTQAzAIPAAvAYeBxM/syPfP6i0KC9/4K\n+MbMvgWQ9DrwEBAp9vDwsBWLxQRNxmP17Lna84Frru54e3Nzc5w8eVLN6iUR+xZgft3rBeDuzZUk\n7QX2AoyNjTEzM5OgyXic/vAQRhWAX9776463Nzk5GateErHDPsm6mGRm+4B9vlEdHYg5cWIRgNnf\n/6FmyJ1fHwVgeGi4k03HIonYC8D2da9HgR+TmZOM2V1rHhZ4wpcTuwGYOrXQBYs2kqQ3chiYkDQu\n6QrgMeDtdMzqT9r2bDNbkfQM8G9gANhvZl+kZlkLnD5zOrCq7lxQcv78eQC2bt3qxqgQkoQRzOwd\n4J2UbOl7EondK3x6+12R54LYfWT8TgDuW/yvA4vCydN1h2Re7DOnz2DlMlYuN6xXrVSoVipcuHDB\nkWX1ZF7so+M7W6p/+JYJzIxuzL1mXuwskdkb5LlzwfhHfSI7tTyPmZeuHxjZ4dXyHVmiFkpcdwNz\nz3ZIZj37SHFX9EkJbwR4zaMDzOCTHV6cLy191ynzQsmc2EEmaKsrfslaGAkb/5ha9soODI3Wymx1\n1Xv0b5JS09HRVMjDiEMy59kzo7f7z2J6o1/NNr4E4MD1nrfv+d8PaZjWlNyzHZIZz754MTrzm1qe\njzy3uc700NoQfBCrL166BMBVV16ZxMSm5J7tkMx49sc3T4SWC/MylWY0qPLxttsAKC3Pd7RnkgnP\nvlQuex3mzZ1moHQq3s1N/k+j6bFOj5dkQux+IRNiH7rpVjB5h4/VjvS8cf3NsxNkQux+oadvkJVK\nJfLcngRLE4K4HSQ1AZJYWfGGAQqF9KXpabE/uunWurKwzkLYhzI4ONjgytGh56MbxwGvZ5I2eRhx\nSE96dq0LFtIVKy1vDB+FQoFVfxSvEdWqN5ng9aO9f4+wcBK0HUw+SOn5Y+7ZDulJz26lCxbHqwG2\nbPH8qlqtxsoSp4fGgHSzyqaeLWm7pA8kHZP0haTn/PLrJb0r6Wv/8bpULOpj4oSRFeAFM9sJ3AP8\nWdIu4EXgfTObAN73XycmylOnTi2EptpmRqFQqDuiCDw8zrXTpqnYZnbczI74z/8PHMNbCP8Q8Kpf\n7VXg4TQM+o8/G74eDQywljPWU6lU6o5gbYikRGEgzayypRukpCJwF3AIuNHMjoP3gQA3RLxnr6QZ\nSTNLS0vJrM04scWWdA3wBvC8mZ2J+z4z22dmk2Y2OTIyEl2xWvWOEO77aQ5DWNypsA2Xrda6fQ2s\nBIxSVCipmnckJJbYkgbxhH7NzN70ixclbfPPbwN+SmxNn9O06ycv4P0TOGZmf1936m3gSeAl//Gt\nJIYcCInVQW7e+aUGWve7nukRL25vTqhaJU4/+zfAE8Bnko76ZX/BE/lfkp4CvgceTWTJZUBTsc3s\nINEf+m+TGlCLpyGpeZyJ3LQJT+GDx402tvof1/UM8uDwWF2Zrmg0YueGsHUmga3tjgjmYyMO6Zpn\n28pK5LnSiW9Tb29iwpudLxaLsdb47WkwIlj1bd/S4gRD7tkO6YpnmxnTNxRDz2mwQOx1fC0wOzvb\n3htF3SjBQd/2qeWFlkztOc8uLc5124QNlJbnI7+D0+rMfs+J3c84DyNmxvRw/Uha4CObZ7x7gagb\n6fTQdkrL38e+Tu7ZDnHr2WZUy+XQYWk3X7ToACuV0Ow3jNyzHeLUs89+9jkfjjf4llfGEDA9egdn\nT8Xb08ap2Fft2snO997j+NPPgjUb0M8AEqP7X+EX998fq3oeRhzS9r5+bTUmLQHngJPOGm2fYeLb\nucPMGsz5eTgVG0DSjJnF25uti3TCzjyMOCQX2yHdEHtfF9psh9TtdB6zL2fyMOIQZ2L38l7bDVbq\n/lXSD5KO+sfvErXjIoz0+l7b/oqubWZ2RNK1wCd4C0X/CJw1s5fTaMeVZ9f22jazMhDstd0TNFip\nmyquxA7bazv1PyYNNq3UBXhG0qeS9idd8O9K7Fh7bXebkJW6rwC3AbuB48Dfklzfldg9t9f2ZsJW\n6prZopmtmvfVsX/ghcO2cSV2T++1HbVSN1gS7fMI8HmSdpyMZ/fSXtsRRK3UfVzSbryQNwf8KUkj\neQbpkDyDdEgutkNysR2Si+2QXGyH5GI7JBfbIbnYDvkZRNl0Hr94pVkAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x159748550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfEAAAEWCAYAAAB2c65HAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niIioqYR4RERETSXEIyIiaur/A6Ls45K6t3pfAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x141c874a8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "num_web_images = len(web_images)   \n",
    "for i in range(0, num_web_images):\n",
    "    probs = top_5_out.values[i,:]\n",
    "    indices = top_5_out.indices[i,:]\n",
    "    labels = [signnames[str(elem)] for elem in indices]\n",
    "    # plot image\n",
    "    plt.figure(figsize=(1,1))\n",
    "    plt.imshow(web_images[i])\n",
    "    plt.show()\n",
    "    # plot barchart\n",
    "    y_pos = np.arange(len(labels)) \n",
    "    plt.barh(y_pos, probs, align='center', alpha=0.5)\n",
    "    plt.yticks(y_pos, labels)\n",
    "    plt.xlabel('Signs')\n",
    "    plt.title('Top 5 Softmax Probabilities')\n",
    "    plt.show()    \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Project Writeup\n",
    "\n",
    "Once you have completed the code implementation, document your results in a project writeup using this [template](https://github.com/udacity/CarND-Traffic-Sign-Classifier-Project/blob/master/writeup_template.md) as a guide. The writeup can be in a markdown or pdf file. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> **Note**: Once you have completed all of the code implementations and successfully answered each question above, you may finalize your work by exporting the iPython Notebook as an HTML document. You can do this by using the menu above and navigating to  \\n\",\n",
    "    \"**File -> Download as -> HTML (.html)**. Include the finished document along with this notebook as your submission."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "---\n",
    "\n",
    "## Step 4 (Optional): Visualize the Neural Network's State with Test Images\n",
    "\n",
    " This Section is not required to complete but acts as an additional excersise for understaning the output of a neural network's weights. While neural networks can be a great learning device they are often referred to as a black box. We can understand what the weights of a neural network look like better by plotting their feature maps. After successfully training your neural network you can see what it's feature maps look like by plotting the output of the network's weight layers in response to a test stimuli image. From these plotted feature maps, it's possible to see what characteristics of an image the network finds interesting. For a sign, maybe the inner network feature maps react with high activation to the sign's boundary outline or to the contrast in the sign's painted symbol.\n",
    "\n",
    " Provided for you below is the function code that allows you to get the visualization output of any tensorflow weight layer you want. The inputs to the function should be a stimuli image, one used during training or a new one you provided, and then the tensorflow variable name that represents the layer's state during the training process, for instance if you wanted to see what the [LeNet lab's](https://classroom.udacity.com/nanodegrees/nd013/parts/fbf77062-5703-404e-b60c-95b78b2f3f9e/modules/6df7ae49-c61c-4bb2-a23e-6527e69209ec/lessons/601ae704-1035-4287-8b11-e2c2716217ad/concepts/d4aca031-508f-4e0b-b493-e7b706120f81) feature maps looked like for it's second convolutional layer you could enter conv2 as the tf_activation variable.\n",
    "\n",
    "For an example of what feature map outputs look like, check out NVIDIA's results in their paper [End-to-End Deep Learning for Self-Driving Cars](https://devblogs.nvidia.com/parallelforall/deep-learning-self-driving-cars/) in the section Visualization of internal CNN State. NVIDIA was able to show that their network's inner weights had high activations to road boundary lines by comparing feature maps from an image with a clear path to one without. Try experimenting with a similar test to show that your trained network's weights are looking for interesting features, whether it's looking at differences in feature maps from images with or without a sign, or even what feature maps look like in a trained network vs a completely untrained one on the same sign image.\n",
    "\n",
    "<figure>\n",
    " <img src=\"visualize_cnn.png\" width=\"380\" alt=\"Combined Image\" />\n",
    " <figcaption>\n",
    " <p></p> \n",
    " <p style=\"text-align: center;\"> Your output should look something like this (above)</p> \n",
    " </figcaption>\n",
    "</figure>\n",
    " <p></p> \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 301,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Visualize your network's feature maps here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "# image_input: the test image being fed into the network to produce the feature maps\n",
    "# tf_activation: should be a tf variable name used during your training procedure that represents the calculated state of a specific weight layer\n",
    "# activation_min/max: can be used to view the activation contrast in more detail, by default matplot sets min and max to the actual min and max values of the output\n",
    "# plt_num: used to plot out multiple different weight feature map sets on the same block, just extend the plt number for each new feature map entry\n",
    "\n",
    "def outputFeatureMap(image_input, tf_activation, activation_min=-1, activation_max=-1 ,plt_num=1):\n",
    "    # Here make sure to preprocess your image_input in a way your network expects\n",
    "    # with size, normalization, ect if needed\n",
    "    # image_input =\n",
    "    # Note: x should be the same name as your network's tensorflow data placeholder variable\n",
    "    # If you get an error tf_activation is not defined it may be having trouble accessing the variable from inside a function\n",
    "    activation = tf_activation.eval(session=sess,feed_dict={x : image_input})\n",
    "    featuremaps = activation.shape[3]\n",
    "    plt.figure(plt_num, figsize=(15,15))\n",
    "    for featuremap in range(featuremaps):\n",
    "        plt.subplot(6,8, featuremap+1) # sets the number of feature maps to show on each row and column\n",
    "        plt.title('FeatureMap ' + str(featuremap)) # displays the feature map number\n",
    "        if activation_min != -1 & activation_max != -1:\n",
    "            plt.imshow(activation[0,:,:, featuremap], interpolation=\"nearest\", vmin =activation_min, vmax=activation_max, cmap=\"gray\")\n",
    "        elif activation_max != -1:\n",
    "            plt.imshow(activation[0,:,:, featuremap], interpolation=\"nearest\", vmax=activation_max, cmap=\"gray\")\n",
    "        elif activation_min !=-1:\n",
    "            plt.imshow(activation[0,:,:, featuremap], interpolation=\"nearest\", vmin=activation_min, cmap=\"gray\")\n",
    "        else:\n",
    "            plt.imshow(activation[0,:,:, featuremap], interpolation=\"nearest\", cmap=\"gray\")"
   ]
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "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"
  },
  "widgets": {
   "state": {},
   "version": "1.1.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}