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": [],
   "source": [
    "# Load pickled data\n",
    "import pickle\n",
    "\n",
    "# TODO: Fill this in based on where you saved the training and testing data\n",
    "\n",
    "training_file = \"traffic-signs-data/train.p\"\n",
    "validation_file= \"traffic-signs-data/valid.p\"\n",
    "testing_file = \"traffic-signs-data/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']"
   ]
  },
  {
   "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": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of training examples = 34799\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",
    "import numpy as np\n",
    "\n",
    "# TODO: Number of training examples\n",
    "n_train = len(X_train)\n",
    "\n",
    "# TODO: Number of validation examples\n",
    "n_validation = len(X_valid)\n",
    "\n",
    "# TODO: Number of testing examples.\n",
    "n_test = len(X_test)\n",
    "\n",
    "# TODO: What's the shape of an traffic sign image?\n",
    "image_shape = X_train[0].shape\n",
    "\n",
    "# TODO: How many unique classes/labels there are in the dataset.\n",
    "all_y = np.concatenate((y_train, y_valid, y_test), axis=0)\n",
    "n_classes = len(np.unique(all_y))\n",
    "\n",
    "\n",
    "print(\"Number of training examples =\", n_train)\n",
    "print(\"Number of testing examples =\", n_test)\n",
    "print(\"Image data shape =\", image_shape)\n",
    "print(\"Number of classes =\", n_classes)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "4410\n"
     ]
    }
   ],
   "source": [
    "print(n_validation)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "import tensorflow as tf\n",
    "from tqdm import tqdm\n",
    "from sklearn.utils import shuffle\n",
    "\n",
    "X_train, y_train = shuffle(X_train, y_train)\n",
    "\n",
    "# Remove the previous graph, weights, and biases\n",
    "tf.reset_default_graph()"
   ]
  },
  {
   "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": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([  180.,  1980.,  2010.,  1260.,  1770.,  1650.,   360.,  1290.,\n",
       "         1260.,  1320.,  1800.,  1170.,  1890.,  1920.,   690.,   540.,\n",
       "          360.,   990.,  1080.,   180.,   300.,   270.,   330.,   450.,\n",
       "          240.,  1350.,   540.,   210.,   480.,   240.,   390.,   690.,\n",
       "          210.,   599.,   360.,  1080.,   330.,   180.,  1860.,   270.,\n",
       "          300.,   210.,   210.]),\n",
       " array([  0.        ,   0.97674419,   1.95348837,   2.93023256,\n",
       "          3.90697674,   4.88372093,   5.86046512,   6.8372093 ,\n",
       "          7.81395349,   8.79069767,   9.76744186,  10.74418605,\n",
       "         11.72093023,  12.69767442,  13.6744186 ,  14.65116279,\n",
       "         15.62790698,  16.60465116,  17.58139535,  18.55813953,\n",
       "         19.53488372,  20.51162791,  21.48837209,  22.46511628,\n",
       "         23.44186047,  24.41860465,  25.39534884,  26.37209302,\n",
       "         27.34883721,  28.3255814 ,  29.30232558,  30.27906977,\n",
       "         31.25581395,  32.23255814,  33.20930233,  34.18604651,\n",
       "         35.1627907 ,  36.13953488,  37.11627907,  38.09302326,\n",
       "         39.06976744,  40.04651163,  41.02325581,  42.        ]),\n",
       " <a list of 43 Patch objects>)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYEAAAD8CAYAAACRkhiPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n4cJzsnIAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78ed4127f0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "### 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",
    "# Making a histogram for the distribution\n",
    "plt.hist(y_train, bins = n_classes)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAP8AAAD8CAYAAAC4nHJkAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78f71a7390>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "imgplot = plt.imshow(X_train[16758])\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x78f7277be0>"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nHxAr9ooAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78f721e400>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#Test adjust_brightness_and_contrast\n",
    "image = X_train[16758]\n",
    "\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(image)\n",
    "#print(y_train[0])"
   ]
  },
  {
   "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": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(32, 32, 3)\n"
     ]
    }
   ],
   "source": [
    "import tensorflow as tf\n",
    "import numpy as np\n",
    "\n",
    "### 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",
    "\n",
    "#print((temp.reshape(32,32,1)).shape)\n",
    "    \n",
    "#a = X_train[0].reshape(32,32,1)\n",
    "\n",
    "#print((cv2.cvtColor(X_train[0], cv2.COLOR_BGR2GRAY)).reshape(32,32,1))\n",
    "#print(X_train[1].shape)\n",
    "\n",
    "#init = tf.global_variables_initializer()\n",
    "#with tf.Session() as sess:\n",
    "#    sess.run(init)\n",
    "#    print(sess.run(x))\n",
    "\n",
    "#for x in traindata\n",
    "#    print(x.get_shape())\n",
    "\n",
    "#print(test[0].get_collection())\n",
    "#X_train = tf.image.per_image_standardization(X_train)\n",
    "#print(X_train[0])\n",
    "\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "print(X_train[16758].shape)\n",
    "#gray1 = cv2.cvtColor(X_train[16758], cv2.COLOR_BGR2GRAY)\n",
    "\n",
    "X_train = np.array([cv2.cvtColor(X_train[i], cv2.COLOR_BGR2GRAY) for i in range(len(X_train))],\n",
    "                          dtype = np.uint8)\n",
    "X_valid = np.array([cv2.cvtColor(X_valid[i], cv2.COLOR_BGR2GRAY) for i in range(len(X_valid))],\n",
    "                          dtype = np.uint8)\n",
    "X_test = np.array([cv2.cvtColor(X_test[i], cv2.COLOR_BGR2GRAY) for i in range(len(X_test))],\n",
    "                          dtype = np.uint8)\n",
    "\n",
    "#print(X_train_gs[16758].shape)\n",
    "\n",
    "\n",
    "\n",
    "#print(gray1)\n",
    "#print(gray1.shape)\n",
    "#print((gray1.squeeze()).shape)\n",
    "\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "#plt.figure(figsize=(1,1))\n",
    "#plt.imshow(gray1, cmap=\"gray\")\n",
    "\n",
    "#create CLAHE instantiation called clahe\n",
    "clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))\n",
    "#cl1 = clahe.apply(gray1)\n",
    "\n",
    "#apply CLAHE to all grayscaled images\n",
    "X_train = np.array([clahe.apply(X_train[i]) for i in range(len(X_train))],\n",
    "                          dtype = np.float32)\n",
    "X_valid = np.array([clahe.apply(X_valid[i]) for i in range(len(X_valid))],\n",
    "                          dtype = np.float32)\n",
    "X_test = np.array([clahe.apply(X_test[i]) for i in range(len(X_test))],\n",
    "                          dtype = np.float32)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(32, 32, 1)\n"
     ]
    }
   ],
   "source": [
    "#expand the shape 32x32 to 32x32x1\n",
    "X_train = X_train[:, :, :,np.newaxis]\n",
    "X_valid = X_valid[:, :, :,np.newaxis]\n",
    "X_test = X_test[:, :, :,np.newaxis]\n",
    "\n",
    "print(X_train[16758].shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Augmenting Image Data...\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████████| 43/43 [01:00<00:00,  2.00s/it]\n"
     ]
    }
   ],
   "source": [
    "#from keras.datasets import mnist\n",
    "import keras\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from matplotlib import pyplot\n",
    "from keras import backend as K\n",
    "\n",
    "# load data\n",
    "#(X_train, y_train), (X_test, y_test) = mnist.load_data()\n",
    "# reshape to be [samples][pixels][width][height]\n",
    "#X_train = X_train.reshape(X_train.shape[0], 1, 28, 28)\n",
    "#X_test = X_test.reshape(X_test.shape[0], 1, 28, 28)\n",
    "# convert from int to float\n",
    "#X_train = X_train.astype('float32')\n",
    "#X_test = X_test.astype('float32')\n",
    "\n",
    "# define data preparation\n",
    "datagen = keras.preprocessing.image.ImageDataGenerator(\n",
    "    rotation_range=15,\n",
    "    width_shift_range=0.1,\n",
    "    height_shift_range=0.1,\n",
    "    shear_range=0.1,\n",
    "    zoom_range=0.1,\n",
    "    fill_mode='nearest',\n",
    "    horizontal_flip=False,\n",
    "    vertical_flip=False\n",
    ")\n",
    "# fit parameters from data\n",
    "#datagen.fit(X_train)\n",
    "# configure batch size and retrieve one batch of images\n",
    "import os\n",
    "import numpy as np\n",
    "from tqdm import tqdm\n",
    "from sklearn.utils import shuffle\n",
    "\n",
    "total_images_per_class = 3000\n",
    "X_train_aug = np.empty((0,32,32,1),dtype='uint8')\n",
    "y_train_aug = np.empty(0,dtype='uint8')\n",
    "\n",
    "#attaches augmented data to existing data (i.e. X_train, y_train)\n",
    "#for y_train, whicever class is in the for loop (say i=10, then the number 10 is repeated via np.repeat in the array)\n",
    "print('Augmenting Image Data...')\n",
    "for i in tqdm(range(n_classes)):\n",
    "    images_for_i_class = X_train[y_train==i]\n",
    "    y_i_class = np.repeat(i, images_for_i_class.shape[0])\n",
    "    \n",
    "    X_train_aug_i = np.copy(images_for_i_class)\n",
    "    y_train_aug_i = np.copy(y_i_class)\n",
    "        \n",
    "    for X,y in datagen.flow(images_for_i_class, y_i_class, batch_size=len(y_i_class), seed=i):\n",
    "        if len(X_train_aug_i) >= total_images_per_class:\n",
    "            break\n",
    "        X_train_aug_i = np.append(X_train_aug_i, X, axis=0)\n",
    "        y_train_aug_i = np.append(y_train_aug_i, y, axis=0)\n",
    "\n",
    "    X_train_aug = np.append(X_train_aug, X_train_aug_i[:total_images_per_class], axis=0)\n",
    "    y_train_aug = np.append(y_train_aug, y_train_aug_i[:total_images_per_class], axis=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(129000, 32, 32, 1)\n",
      "(34799, 32, 32, 1)\n",
      "(4410, 32, 32, 1)\n",
      "(4410,)\n",
      "(12630, 32, 32, 1)\n",
      "(12630,)\n"
     ]
    }
   ],
   "source": [
    "print(X_train_aug.shape)\n",
    "print(X_train.shape)\n",
    "\n",
    "X_train = X_train_aug\n",
    "y_train = y_train_aug\n",
    "\n",
    "print(X_valid.shape)\n",
    "print(y_valid.shape)\n",
    "print(X_test.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "#not used, later replaced by CLAHE method\n",
    "\n",
    "import cv2\n",
    "#Function for adjust brightness and contrast\n",
    "def adjust_brightness_and_contrast(image):\n",
    "    image[:,:,0] = cv2.equalizeHist(image[:,:,0])\n",
    "    image[:,:,1] = cv2.equalizeHist(image[:,:,1])\n",
    "    image[:,:,2] = cv2.equalizeHist(image[:,:,2])\n",
    "    #image = image/255.-.5\n",
    "        \n",
    "    return image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[7004]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x789585a5c0>"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAFsAAABZCAYAAABR/liSAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78957f0550>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#test random samples from original + augmented data\n",
    "index = np.random.randint(34000, size=1)\n",
    "print(index)\n",
    "#print(X_train[25000])\n",
    "img = X_train[index].squeeze()\n",
    "#print(image.shape)\n",
    "\n",
    "plt.figure(figsize=(1,1))\n",
    "plt.imshow(img, cmap=\"gray\")\n",
    "#print(y_train[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-83-61776cfca50e>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[1;31m#normalize - FINAL: normalize into range of min -0.5 to max 0.5\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 2\u001b[1;33m \u001b[0mX_train\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mX_train\u001b[0m\u001b[1;33m/\u001b[0m\u001b[1;36m255\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m-\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      3\u001b[0m \u001b[0mX_valid\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mX_valid\u001b[0m\u001b[1;33m/\u001b[0m\u001b[1;36m255\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m-\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      4\u001b[0m \u001b[0mX_test\u001b[0m \u001b[1;33m=\u001b[0m \u001b[1;33m(\u001b[0m\u001b[0mX_test\u001b[0m\u001b[1;33m/\u001b[0m\u001b[1;36m255\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;33m-\u001b[0m \u001b[1;36m0.5\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "#normalize - FINAL: normalize into range of min -0.5 to max 0.5\n",
    "X_train = (X_train/255) - 0.5\n",
    "X_valid = (X_valid/255) - 0.5\n",
    "X_test = (X_test/255) - 0.5"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model Architecture"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Define your architecture here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "from tensorflow.contrib.layers import flatten\n",
    "\n",
    "def LeNet(x):\n",
    "\n",
    "    mu = 0\n",
    "    sigma = 0.1\n",
    "\n",
    "    #implement LeNet\n",
    "    #Layer 1: Convolutional. Input = 32x32x1. Output = 28x28x12.\n",
    "    conv1_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 1, 12), mean = mu, stddev = sigma))\n",
    "    conv1_b = tf.Variable(tf.zeros(12))\n",
    "    conv1   = tf.nn.conv2d(x, conv1_W, strides=[1, 1, 1, 1], padding='VALID') + conv1_b\n",
    "\n",
    "    #Activation. Output = 28x28x12\n",
    "    conv1 = tf.nn.relu(conv1)\n",
    "\n",
    "    #Pooling. Input = 28x28x12. Output = 14x14x12.\n",
    "    conv1 = tf.nn.max_pool(conv1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')\n",
    "\n",
    "    #Layer 2: Convolutional. Input = 14x14x12 Output = 10x10x32.\n",
    "    conv2_W = tf.Variable(tf.truncated_normal(shape=(5, 5, 12, 32), mean = mu, stddev = sigma))\n",
    "    conv2_b = tf.Variable(tf.zeros(32))\n",
    "    conv2   = tf.nn.conv2d(conv1, conv2_W, strides=[1, 1, 1, 1], padding='VALID') + conv2_b\n",
    "\n",
    "    #Activation. Output = 10x10x32\n",
    "    conv2 = tf.nn.relu(conv2)\n",
    "\n",
    "    #Pooling. Input = 10x10x32. Output = 5x5x32.\n",
    "    conv2 = tf.nn.max_pool(conv2, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='VALID')\n",
    "\n",
    "    #Flatten. Input = 5x5x32. Output = 800.\n",
    "    fc0   = flatten(conv2)\n",
    "\n",
    "    #Layer 3: Fully Connected. Input = 800. Output = 120.\n",
    "    fc1_W = tf.Variable(tf.truncated_normal(shape=(800, 120), mean = mu, stddev = sigma))\n",
    "    fc1_b = tf.Variable(tf.zeros(120))\n",
    "    fc1   = tf.matmul(fc0, fc1_W) + fc1_b\n",
    "\n",
    "    #Activation.\n",
    "    fc1    = tf.nn.relu(fc1)\n",
    "\n",
    "    #Layer 4: Fully Connected. Input = 120. Output = 84.\n",
    "    fc2_W  = tf.Variable(tf.truncated_normal(shape=(120, 84), mean = mu, stddev = sigma))\n",
    "    fc2_b  = tf.Variable(tf.zeros(84))\n",
    "    fc2    = tf.matmul(fc1, fc2_W) + fc2_b\n",
    "\n",
    "    #Activation.\n",
    "    fc2    = tf.nn.relu(fc2)\n",
    "\n",
    "    #Layer 5: Fully Connected. Input = 84. Output = 43.\n",
    "    fc3_W  = tf.Variable(tf.truncated_normal(shape=(84, 43), mean = mu, stddev = sigma))\n",
    "    fc3_b  = tf.Variable(tf.zeros(43))\n",
    "    logits = tf.matmul(fc2, fc3_W) + fc3_b\n",
    "\n",
    "    return logits\n",
    "\n"
   ]
  },
  {
   "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": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "### Train your model here.\n",
    "### Calculate and report the accuracy on the training and validation set.\n",
    "### Once a final model architecture is selected, \n",
    "### the accuracy on the test set should be calculated and reported as well.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "EPOCHS = 10\n",
    "BATCH_SIZE = 128\n",
    "\n",
    "# MNIST consists of 28x28x1, grayscale images\n",
    "x = tf.placeholder(tf.float32, (None, 32, 32, 1))\n",
    "y = tf.placeholder(tf.int32, (None))\n",
    "one_hot_y = tf.one_hot(y, n_classes)\n",
    "\n",
    "#Set a slow learning rate of 0.001\n",
    "rate = 0.001\n",
    "\n",
    "#feed in grayscale images\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",
    "training_operation = optimizer.minimize(loss_operation)\n",
    "\n",
    "\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",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "#define evaluate function\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": 18,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Training...\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:31<00:00,  6.75it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 1 ...\n",
      "Validation Accuracy = 0.893\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:35<00:00,  6.79it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 2 ...\n",
      "Validation Accuracy = 0.943\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:37<00:00,  6.78it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 3 ...\n",
      "Validation Accuracy = 0.945\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:35<00:00,  6.81it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 4 ...\n",
      "Validation Accuracy = 0.956\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:35<00:00,  7.01it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 5 ...\n",
      "Validation Accuracy = 0.956\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:35<00:00,  7.12it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 6 ...\n",
      "Validation Accuracy = 0.956\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:33<00:00,  7.14it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 7 ...\n",
      "Validation Accuracy = 0.966\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:34<00:00,  7.09it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 8 ...\n",
      "Validation Accuracy = 0.962\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:35<00:00,  6.79it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 9 ...\n",
      "Validation Accuracy = 0.972\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|██████████████████████████████████████| 1008/1008 [02:34<00:00,  6.96it/s]\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "EPOCH 10 ...\n",
      "Validation Accuracy = 0.974\n",
      "\n",
      "Model saved\n"
     ]
    }
   ],
   "source": [
    "#Train \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 tqdm(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",
    "        validation_accuracy = evaluate(X_valid, y_valid)\n",
    "        print(\"EPOCH {} ...\".format(i+1))\n",
    "        print(\"Validation Accuracy = {:.3f}\".format(validation_accuracy))\n",
    "        print()\n",
    "        \n",
    "    saver.save(sess, './lenet')\n",
    "    print(\"Model saved\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n",
      "Test Accuracy = 0.950\n"
     ]
    }
   ],
   "source": [
    "#use the test set to get the final result\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver2 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver2.restore(sess, \"./lenet\")\n",
    "\n",
    "    test_accuracy = evaluate(X_test, y_test)\n",
    "    print(\"Test Accuracy = {:.3f}\".format(test_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n",
      "Train Accuracy = 0.991\n"
     ]
    }
   ],
   "source": [
    "#for the write-up, calculate the train accuracy, which should be very very close to 100%\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver10 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver10.restore(sess, \"./lenet\")\n",
    "\n",
    "    train_accuracy = evaluate(X_train, y_train)\n",
    "    print(\"Train Accuracy = {:.3f}\".format(train_accuracy))"
   ]
  },
  {
   "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": 115,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdkAAADDCAYAAADOfWtLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x789543b0b8>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "import os\n",
    "from os import listdir\n",
    "from PIL import Image\n",
    "import glob\n",
    "\n",
    "# Read sample image files, resize them, convert to numpy arrays w/ dtype=uint8\n",
    "image_files  = ['sign_examples/' + image_file for image_file in os.listdir('sign_examples')]\n",
    "images = []\n",
    "for image_file in image_files:\n",
    "    image = Image.open(image_file)\n",
    "    image = image.convert('RGB')\n",
    "    image = image.resize((32, 32), Image.ANTIALIAS)\n",
    "    image = np.array(list(image.getdata()), dtype='uint8')\n",
    "    image = np.reshape(image, (32, 32, 3))\n",
    "\n",
    "    images.append(image)\n",
    "X_sample = np.array(images, dtype='uint8')\n",
    "\n",
    "# Visually inspect sample images\n",
    "for i, image in enumerate(X_sample):\n",
    "    plt.subplot(3, 3, i+1)\n",
    "    plt.imshow(image)\n",
    "    plt.title(image_files[i])\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "y_sample = np.array([17,12,7,14,13]) #manually looked up in the csv file comparing each image\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5, 32, 32, 3)\n",
      "(5,)\n",
      "(12630,)\n"
     ]
    }
   ],
   "source": [
    "print(X_sample.shape)\n",
    "print(y_sample.shape)\n",
    "print(y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Predict the Sign Type for Each Image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 117,
   "metadata": {},
   "outputs": [],
   "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",
    "X_sample = np.array([cv2.cvtColor(X_sample[i], cv2.COLOR_BGR2GRAY) for i in range(len(X_sample))],\n",
    "                          dtype = np.uint8)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 118,
   "metadata": {},
   "outputs": [],
   "source": [
    "#create CLAHE instantiation called clahe\n",
    "clahe = cv2.createCLAHE(clipLimit=2.0, tileGridSize=(8,8))\n",
    "#cl1 = clahe.apply(gray1)\n",
    "\n",
    "#apply CLAHE to all grayscaled images\n",
    "X_sample = np.array([clahe.apply(X_sample[i]) for i in range(len(X_sample))],\n",
    "                          dtype = np.float32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 119,
   "metadata": {},
   "outputs": [],
   "source": [
    "#expand the shape 32x32 to 32x32x1\n",
    "X_sample = X_sample[:, :, :,np.newaxis]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 120,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAaMAAAC8CAYAAADVYsEOAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nANjJCICdjADYBfW6OrbsrkHOAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78960f6ef0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#visualize preprocessed sample images\n",
    "# Visually inspect sample images\n",
    "for i, image in enumerate(X_sample):\n",
    "    image = image.squeeze()\n",
    "    plt.subplot(3, 3, i+1)\n",
    "    plt.imshow(image, cmap='gray')\n",
    "    plt.title(image_files[i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 121,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n"
     ]
    }
   ],
   "source": [
    "predict_proba_operation = tf.nn.softmax(logits=logits)\n",
    "\n",
    "#sess = tf.get_default_session()\n",
    "predicted_proba = list()\n",
    "    \n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver4 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver4.restore(sess, \"./lenet\")\n",
    "    predicted_proba.extend(sess.run(predict_proba_operation, feed_dict={x: X_sample}))\n",
    "\n",
    "#print(predicted_proba[0].where(1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 122,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[17 12  7 14 13]\n",
      "[23, 12, 23, 3, 23]\n"
     ]
    }
   ],
   "source": [
    "print(y_sample)\n",
    "pred_result_arr = []\n",
    "for i in range(len(predicted_proba)):\n",
    "    pred_result_arr.append(predicted_proba[i].tolist().index(1))\n",
    "    #print(predicted_proba[i].tolist().index(1))\n",
    "#[y for x in predicted_proba for y in x].index(0)\n",
    "#print(predicted_proba[0].tolist().index(1.0))\n",
    "print(pred_result_arr)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n",
      "Prediction [False  True False False False]\n",
      "Accuracy 0.20000000298023224\n"
     ]
    }
   ],
   "source": [
    "soft = tf.nn.softmax(logits)\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver30 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver30.restore(sess, \"./lenet\")\n",
    "        \n",
    "    res = sess.run(correct_prediction, feed_dict={x : X_sample, y : y_sample})\n",
    "    test_accuracy = evaluate(X_sample, y_sample)\n",
    "    \n",
    "    print(\"Prediction {}\".format(res))\n",
    "    print(\"Accuracy {}\".format(test_accuracy))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 133,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA4sAAACrCAYAAADPVwnIAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x78a0153fd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "#reportImg = X_sample[5:15]\n",
    "reportImg = X_sample.reshape((-1,32,32))\n",
    "plt.figure(figsize=(16,16))\n",
    "for i in range(0,len(reportImg)):\n",
    "    subp = plt.subplot(6,5, i+1)\n",
    "    plt.imshow(reportImg[i], cmap='gray')\n",
    "    plt.axis('off')\n",
    "    subp.set_title( \"Class %s, Predited %s\" % (y_sample[i], res[i]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Analyze Performance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 123,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n",
      "Test Set Accuracy = 0.200\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",
    "### Run the predictions here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver3 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver3.restore(sess, \"./lenet\")\n",
    "    my_accuracy = evaluate(X_sample, y_sample)\n",
    "    print(\"Test Set Accuracy = {:.3f}\".format(my_accuracy))"
   ]
  },
  {
   "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": 74,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAdkAAADDCAYAAADOfWtLAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7892c95f28>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Visually inspect sample images\n",
    "for i, image in enumerate(images):\n",
    "    plt.subplot(3, 3, i+1)\n",
    "    plt.imshow(image)\n",
    "    plt.title(image_files[i])\n",
    "\n",
    "plt.tight_layout()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Restoring parameters from ./lenet\n",
      "Prediction [False  True False False False]\n",
      "Accuracy 0.20000000298023224\n",
      "TOP 5 TopKV2(values=array([[ 1.,  0.,  0.,  0.,  0.],\n",
      "       [ 1.,  0.,  0.,  0.,  0.],\n",
      "       [ 1.,  0.,  0.,  0.,  0.],\n",
      "       [ 1.,  0.,  0.,  0.,  0.],\n",
      "       [ 1.,  0.,  0.,  0.,  0.]], dtype=float32), indices=array([[23,  0,  1,  2,  3],\n",
      "       [12,  0,  1,  2,  3],\n",
      "       [23,  0,  1,  2,  3],\n",
      "       [ 3,  0,  1,  2,  4],\n",
      "       [23,  0,  1,  2,  3]]))\n"
     ]
    }
   ],
   "source": [
    "### Visualize the softmax probabilities here.\n",
    "### Feel free to use as many code cells as needed.\n",
    "\n",
    "soft = tf.nn.softmax(logits)\n",
    "top_5 = tf.nn.top_k(soft,5)\n",
    "\n",
    "with tf.Session() as sess:\n",
    "    sess.run(tf.global_variables_initializer())\n",
    "    saver4 = tf.train.import_meta_graph('./lenet.meta')\n",
    "    saver4.restore(sess, \"./lenet\")\n",
    "    \n",
    "    res = sess.run(correct_prediction, feed_dict={x : X_sample, y : y_sample})\n",
    "    top = sess.run(top_5, feed_dict={x : X_sample, y : y_sample})\n",
    "    test_accuracy = evaluate(X_sample,  y_sample)\n",
    "    \n",
    "    print(\"Prediction {}\".format(res))\n",
    "    print(\"Accuracy {}\".format(test_accuracy))\n",
    "    print(\"TOP 5 {}\".format(top))\n",
    "\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": null,
   "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\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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
}