R9_Internal_Lab_Questions.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "I-P5m4iA_xe7"
   },
   "source": [
    "# Bounding box detection - Racoon data\n",
    "\n",
    "\n",
    "## Data files\n",
    "- images_racoon.rar: contain images of racoons\n",
    "- train_labels.cv: contains coordinates for bounding box for every image"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "d8M5gzwKWGiU"
   },
   "source": [
    "### Import the necessary libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VOMKpZJcWGiV"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Z_INIZOi_tUW"
   },
   "source": [
    "### Change directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Sxgw3GT1AR9Y"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lwxBtvYRWGii"
   },
   "source": [
    "### Load the training data from train.csv file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "wiTjjKtpWGij"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "-F3CkYxUWGin"
   },
   "source": [
    "### Print the shape of the train dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "uRz_MtnvWGip"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "axf5m10eWGis"
   },
   "source": [
    "### Declare a variable IMAGE_SIZE = 128 as we will be using MobileNet which will be taking Input shape as 128 * 128 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7ayEk2c9WGit"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "wbrrYb5IWGix"
   },
   "source": [
    "### With the help of csv.reader write a for loop which can load the train.csv file and store the path, width, height, x0,y0,x1,y1 in induvidual variables. <br>\n",
    "1. Create a list variable known as 'path' which has all the path for all the training images\n",
    "2. Create an array 'coords' which has the resized coordinates of the bounding box for the training images\n",
    "\n",
    "<u>Note:</u> All the training images should be downsampled to 128 * 128 as it is the input shape of MobileNet (which we will be using for Object detection). Hence the corresponding coordinates of the bounding boxes should be changed to match the image dimension of 128 * 128 "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "twSTiPpRWGiy"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Z9iZWEpKWGi2"
   },
   "source": [
    "### Write a for loop which can load all the training images into a variable 'batch_images' using the paths from the 'paths' variable\n",
    "<u>Note:</u> Convert the image to RGB scale as the MobileNet accepts 3 channels as inputs   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "h6e99qquWGi3"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "C9KU6pUTWGi9"
   },
   "source": [
    "### Import MobileNet and load MobileNet into a variable named 'model' which takes input shape of 128 * 128 * 3. Freeze all the layers. Add convolution and reshape layers at the end to ensure the output is 4 coordinates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "nosEtf1lWGi-"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "nsQSHWgpWGjC"
   },
   "source": [
    "### Define a custom loss function IoU which calculates Intersection Over Union"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7Ig_Gsn-WGjD"
   },
   "outputs": [],
   "source": [
    "def loss(gt,pred):\n",
    "    intersections = 0\n",
    "    unions = 0\n",
    "    diff_width = np.minimum(gt[:,0] + gt[:,2], pred[:,0] + pred[:,2]) - np.maximum(gt[:,0], pred[:,0])\n",
    "    diff_height = np.minimum(gt[:,1] + gt[:,3], pred[:,1] + pred[:,3]) - np.maximum(gt[:,1], pred[:,1])\n",
    "    intersection = diff_width * diff_height\n",
    "    \n",
    "    # Compute union\n",
    "    area_gt = gt[:,2] * gt[:,3]\n",
    "    area_pred = pred[:,2] * pred[:,3]\n",
    "    union = area_gt + area_pred - intersection\n",
    "\n",
    "#     Compute intersection and union over multiple boxes\n",
    "    for j, _ in enumerate(union):\n",
    "        if union[j] > 0 and intersection[j] > 0 and union[j] >= intersection[j]:\n",
    "            intersections += intersection[j]\n",
    "            unions += union[j]\n",
    "\n",
    "    # Compute IOU. Use epsilon to prevent division by zero\n",
    "    iou = np.round(intersections / (unions + epsilon()), 4)\n",
    "    iou = iou.astype(np.float32)\n",
    "    return iou\n",
    "\n",
    "def IoU(y_true, y_pred):\n",
    "    iou = tf.py_func(loss, [y_true, y_pred], tf.float32)\n",
    "    return iou"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "2Zs7VcSVWGjF"
   },
   "source": [
    "### Write model.compile function & model.fit function with: <br>\n",
    "1. Optimizer = Adam, Loss = 'mse' and metrics = IoU\n",
    "2. Epochs = 30, batch_size = 32, verbose = 1"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "f35dLEJlWGjG"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "II1itmqiWGjK"
   },
   "source": [
    "### Pick a test image from the given data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0ULErYeSWGjK"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "h99SC9MjWGjO"
   },
   "source": [
    "### Resize the image to 128 * 128 and preprocess the image for the MobileNet model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "J3jbKm7_WGjP"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "aGK4IjdaWGjR"
   },
   "source": [
    "### Predict the coordinates of the bounding box for the given test image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "abnXH5F7WGjS"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CXbjVdLWWGjW"
   },
   "source": [
    "### Plot the test image using .imshow and draw a boundary box around the image with the coordinates obtained from the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "t4Ah05MlWGjY"
   },
   "outputs": [],
   "source": [
    "x0 = int(region[0] * image_width / IMAGE_SIZE) # Scale the BBox\n",
    "y0 = int(region[1] * image_height / IMAGE_SIZE)\n",
    "\n",
    "x1 = int((region[2]) * image_width / IMAGE_SIZE)\n",
    "y1 = int((region[3]) * image_height / IMAGE_SIZE)\n",
    "\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.patches as patches\n",
    "from PIL import Image\n",
    "import numpy as np\n",
    "\n",
    "\n",
    "# Create figure and axes\n",
    "fig,ax = plt.subplots(1)\n",
    "\n",
    "# Display the image\n",
    "ax.imshow(unscaled)\n",
    "\n",
    "# Create a Rectangle patch\n",
    "rect = patches.Rectangle((x0, y0), (x1 - x0) , (y1 - y0) , linewidth=2, edgecolor='r', facecolor='none')\n",
    "\n",
    "# Add the patch to the Axes\n",
    "ax.add_patch(rect)\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "70uye82rMg3S"
   },
   "source": [
    "# Time Series Prediction using LSTM"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CBdggSYjMg3U"
   },
   "source": [
    "### Download Data\n",
    "Link: https://datamarket.com/data/set/2324/daily-minimum-temperatures-in-melbourne-australia-1981-1990#!ds=2324&display=line\n",
    "\n",
    "#### Description\n",
    "Daily minimum temperatures in Melbourne, Australia, 1981-1990\n",
    "\n",
    "Units: Degrees Celcius\n",
    "\n",
    "#### Steps before loading\n",
    "- Rename the column name with temprature values to \"Temprature\"\n",
    "- In the last, there is one extra row in the data, remove it by opening the file and save it again.\n",
    "- There are some values in Temprature column which have a \"?\" before them, they will give error, remove \"?\" before them and save the file\n",
    "- If you don't want to do these steps, just load the data file given by Great Learning."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "xXaEyK4qMg3W"
   },
   "source": [
    "#### Mount google drive"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VjoLRSg57qlt"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "BghKtCKlMg3b"
   },
   "source": [
    "#### Change your present working directory"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JOYCPgwJMg3c"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "UMP39JKEMg3f"
   },
   "source": [
    "### Load your data file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "AanHqGBu7Jw2"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "EAfx3bsQMg3j"
   },
   "source": [
    "#### Plot data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "REU4UapqMg3k"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LWGBEOdNMg3n"
   },
   "source": [
    "#### Descibe your dataframe"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "oBWNA_kG7JxC"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "HlPYgFPLMg3r"
   },
   "source": [
    "#### Check for null values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "97MVIy_N7JxH"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qo6HzStBMg3v"
   },
   "source": [
    "#### Drop null values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "fgh7mdOI7JxS"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LcJgcz37Mg30"
   },
   "source": [
    "#### Get the representation of the distribution of data in the form of histogram"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "659sLAx27Jxc"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "GpH88qtBMg34"
   },
   "source": [
    "#### Check the maximum and minimum values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WmkjCx8c7Jxn"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "AQVmR3ENMg37"
   },
   "source": [
    "### Normalize the data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "fb4SNzD27Jxw"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Amc8j4P-Mg3-"
   },
   "source": [
    "#### Check the maximum and minimum values of scaled data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "clv2rSfL7Jx4"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "mPwfKLsbMg4B"
   },
   "source": [
    "#### Look into some of the scaled values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Xq9CBinO7Jx9"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "4i6aEeORMg4F"
   },
   "source": [
    "### Split data into Training and Testing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-0Uh7lCO7JyD"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "MYxgFlAHMg4I"
   },
   "source": [
    "#### Print train and test size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "T9pRem3FMg4J"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "-UKovqKBMg4L"
   },
   "source": [
    "## Create the sequential data\n",
    "Map the temprature at a particular time t to the temprature at time t+n, where n is any number you define.\n",
    "\n",
    "For example: to map tempratures of consecutive days, use t+1, i.e. loop_back = 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "m7uoSqtKMg4M"
   },
   "source": [
    "#### Define your function to create dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jWIvInPB7JyK"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "WePfiYR6Mg4R"
   },
   "source": [
    "#### Use function to get training and test set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "uarOJcPk7JyQ",
    "scrolled": true
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "w_vztTvsMg4X"
   },
   "source": [
    "#### Transform the prepared train and test input data into the expected structure using numpy.reshape()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UJZ4VB4N7JyV"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "2ApGy3FBMg4a"
   },
   "source": [
    "## Define Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "kx0ctZLiMg4b"
   },
   "source": [
    "### Define sequntial model, add LSTM layer and compile the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yRW-Txvt7Jya"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "zyAxLVdxMg4f"
   },
   "source": [
    "### Summarize your model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0I_UO7GOMg4g"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "orlJW-1_Mg4j"
   },
   "source": [
    "### Train the model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "yyzESxsmMg4j"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eSQgnvPKMg4m"
   },
   "source": [
    "## Make Predictions and Evaluate your model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "85Ip5wrq7Jyi"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Ewi6Zyl9Mg4o"
   },
   "source": [
    "## Plot the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FHbIWp-D7Jyn"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "colab": {
   "collapsed_sections": [],
   "name": "R9_Internal_Lab_Questions.ipynb",
   "provenance": [],
   "version": "0.3.2"
  },
  "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": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}