Face+Recognition+for+the+Happy+House+-+v2.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Face Recognition\n",
    "\n",
    "Many of the ideas presented here are from [FaceNet](https://arxiv.org/pdf/1503.03832.pdf). \n",
    "\n",
    "Face recognition problems commonly fall into two categories: \n",
    "\n",
    "- **Face Verification** - \"is this the claimed person?\". For example, at some airports, you can pass through customs by letting a system scan your passport and then verifying that you (the person carrying the passport) are the correct person. A mobile phone that unlocks using your face is also using face verification. This is a 1:1 matching problem. \n",
    "- **Face Recognition** - \"who is this person?\". For example, the video lecture showed a face recognition video (https://www.youtube.com/watch?v=wr4rx0Spihs) of Baidu employees entering the office without needing to otherwise identify themselves. This is a 1:K matching problem. \n",
    "\n",
    "FaceNet learns a neural network that encodes a face image into a vector of 128 numbers. By comparing two such vectors, you can then determine if two pictures are of the same person.\n",
    "\n",
    "In this notebook, we will be using a pre-trained model which represents ConvNet activations using a \"channels first\" convention, as opposed to the \"channels last\" convention used in lecture and previous programming assignments. In other words, a batch of images will be of shape $(m, n_C, n_H, n_W)$ instead of $(m, n_H, n_W, n_C)$. \n",
    "\n",
    "Let's load the required packages. \n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Sequential\n",
    "from keras.layers import Conv2D, ZeroPadding2D, Activation, Input, concatenate\n",
    "from keras.models import Model\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.layers.pooling import MaxPooling2D, AveragePooling2D\n",
    "from keras.layers.merge import Concatenate\n",
    "from keras.layers.core import Lambda, Flatten, Dense\n",
    "from keras.initializers import glorot_uniform\n",
    "from keras.engine.topology import Layer\n",
    "from keras import backend as K\n",
    "K.set_image_data_format('channels_first')\n",
    "import cv2\n",
    "import os\n",
    "import numpy as np\n",
    "from numpy import genfromtxt\n",
    "import pandas as pd\n",
    "import tensorflow as tf\n",
    "from fr_utils import *\n",
    "from inception_blocks_v2 import *\n",
    "\n",
    "%matplotlib inline\n",
    "%load_ext autoreload\n",
    "%autoreload 2\n",
    "\n",
    "np.set_printoptions(threshold=np.nan)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 0 - Naive Face Verification\n",
    "\n",
    "In Face Verification, we're given two images and you have to tell if they are of the same person. The simplest way to do this is to compare the two images pixel-by-pixel. If the distance between the raw images are less than a chosen threshold, it may be the same person! "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "Of course, this algorithm performs really poorly, since the pixel values change dramatically due to variations in lighting, orientation of the person's face, even minor changes in head position, and so on. \n",
    "\n",
    "You'll see that rather than using the raw image, you can learn an encoding $f(img)$ so that element-wise comparisons of this encoding gives more accurate judgements as to whether two pictures are of the same person."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1 - Encoding face images into a 128-dimensional vector \n",
    "\n",
    "### 1.1 - Using an ConvNet  to compute encodings\n",
    "\n",
    "The FaceNet model takes a lot of data and a long time to train. So following common practice in applied deep learning settings, let's just load weights that someone else has already trained. The network architecture follows the Inception model from [Szegedy *et al.*](https://arxiv.org/abs/1409.4842). \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The key things you need to know are:\n",
    "\n",
    "- This network uses 96x96 dimensional RGB images as its input. Specifically, inputs a face image (or batch of $m$ face images) as a tensor of shape $(m, n_C, n_H, n_W) = (m, 3, 96, 96)$ \n",
    "- It outputs a matrix of shape $(m, 128)$ that encodes each input face image into a 128-dimensional vector\n",
    "\n",
    "Run the cell below to create the model for face images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "FRmodel = faceRecoModel(input_shape=(3, 96, 96))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Params: 3743280\n"
     ]
    }
   ],
   "source": [
    "print(\"Total Params:\", FRmodel.count_params())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "By using a 128-neuron fully connected layer as its last layer, the model ensures that the output is an encoding vector of size 128. \n",
    "\n",
    "By computing a distance between two encodings and thresholding, we can determine if the two pictures represent the same person\n",
    "\n",
    "So, an encoding is a good one if: \n",
    "- The encodings of two images of the same person are quite similar to each other \n",
    "- The encodings of two images of different persons are very different\n",
    "\n",
    "The triplet loss function formalizes this, and tries to \"push\" the encodings of two images of the same person (Anchor and Positive) closer together, while \"pulling\" the encodings of two images of different persons (Anchor, Negative) further apart. \n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "\n",
    "### 1.2 - The Triplet Loss\n",
    "\n",
    "For an image $x$, we denote its encoding $f(x)$, where $f$ is the function computed by the neural network.\n",
    "\n",
    "\n",
    "<!--\n",
    "We will also add a normalization step at the end of our model so that $\\mid \\mid f(x) \\mid \\mid_2 = 1$ (means the vector of encoding should be of norm 1).\n",
    "!-->\n",
    "\n",
    "Training will use triplets of images $(A, P, N)$:  \n",
    "\n",
    "- A is an \"Anchor\" image--a picture of a person. \n",
    "- P is a \"Positive\" image--a picture of the same person as the Anchor image.\n",
    "- N is a \"Negative\" image--a picture of a different person than the Anchor image.\n",
    "\n",
    "These triplets are picked from our training dataset. We will write $(A^{(i)}, P^{(i)}, N^{(i)})$ to denote the $i$-th training example. \n",
    "\n",
    "You'd like to make sure that an image $A^{(i)}$ of an individual is closer to the Positive $P^{(i)}$ than to the Negative image $N^{(i)}$) by at least a margin $\\alpha$:-"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def triplet_loss(y_true, y_pred, alpha = 0.2):\n",
    "    \"\"\"\n",
    "    Implementation of the triplet loss as defined by formula (3)\n",
    "    \n",
    "    Arguments:\n",
    "    y_true -- true labels, required when you define a loss in Keras, you don't need it in this function.\n",
    "    y_pred -- python list containing three objects:\n",
    "            anchor -- the encodings for the anchor images, of shape (None, 128)\n",
    "            positive -- the encodings for the positive images, of shape (None, 128)\n",
    "            negative -- the encodings for the negative images, of shape (None, 128)\n",
    "    \n",
    "    Returns:\n",
    "    loss -- real number, value of the loss\n",
    "    \"\"\"\n",
    "    \n",
    "    anchor, positive, negative = y_pred[0], y_pred[1], y_pred[2]\n",
    "    \n",
    "    ### START CODE HERE ### (≈ 4 lines)\n",
    "    # Step 1: Compute the (encoding) distance between the anchor and the positive\n",
    "    pos_dist = tf.square(anchor-positive)\n",
    "\n",
    "    # Step 2: Compute the (encoding) distance between the anchor and the negative\n",
    "    neg_dist = tf.square(anchor-negative)\n",
    "\n",
    "    # Step 3: subtract the two previous distances and add alpha.\n",
    "    basic_loss = tf.reduce_sum(pos_dist-neg_dist)+alpha\n",
    "    \n",
    "    # Step 4: Take the maximum of basic_loss and 0.0. Sum over the training examples.\n",
    "    loss = tf.reduce_sum(tf.maximum(basic_loss,0.))\n",
    "    ### END CODE HERE ###\n",
    "    \n",
    "    return loss"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "loss = 350.026\n"
     ]
    }
   ],
   "source": [
    "with tf.Session() as test:\n",
    "    tf.set_random_seed(1)\n",
    "    y_true = (None, None, None)\n",
    "    y_pred = (tf.random_normal([3, 128], mean=6, stddev=0.1, seed = 1),\n",
    "              tf.random_normal([3, 128], mean=1, stddev=1, seed = 1),\n",
    "              tf.random_normal([3, 128], mean=3, stddev=4, seed = 1))\n",
    "    loss = triplet_loss(y_true, y_pred)\n",
    "    print(\"loss = \" + str(loss.eval()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2 - Loading the trained model\n",
    "\n",
    "FaceNet is trained by minimizing the triplet loss. But since training requires a lot of data and a lot of computation, we won't train it from scratch here. Instead, we load a previously trained model. Load a model using the following cell; this might take a couple of minutes to run. "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "FRmodel.compile(optimizer = 'adam', loss = triplet_loss, metrics = ['accuracy'])\n",
    "load_weights_from_FaceNet(FRmodel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let's now use this model to perform face verification and face recognition! "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3 - Applying the model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.1 - Face Verification\n",
    "\n",
    "Let's build a database containing one encoding vector for each person allowed to enter the happy house. To generate the encoding we use `img_to_encoding(image_path, model)` which basically runs the forward propagation of the model on the specified image. \n",
    "\n",
    "The following code build the database (represented as a python dictionary). This database maps each person's name to a 128-dimensional encoding of their face."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "database = {}\n",
    "database[\"danielle\"] = img_to_encoding(\"images/danielle.png\", FRmodel)\n",
    "database[\"younes\"] = img_to_encoding(\"images/younes.jpg\", FRmodel)\n",
    "database[\"tian\"] = img_to_encoding(\"images/tian.jpg\", FRmodel)\n",
    "database[\"andrew\"] = img_to_encoding(\"images/andrew.jpg\", FRmodel)\n",
    "database[\"kian\"] = img_to_encoding(\"images/kian.jpg\", FRmodel)\n",
    "database[\"dan\"] = img_to_encoding(\"images/dan.jpg\", FRmodel)\n",
    "database[\"sebastiano\"] = img_to_encoding(\"images/sebastiano.jpg\", FRmodel)\n",
    "database[\"bertrand\"] = img_to_encoding(\"images/bertrand.jpg\", FRmodel)\n",
    "database[\"kevin\"] = img_to_encoding(\"images/kevin.jpg\", FRmodel)\n",
    "database[\"felix\"] = img_to_encoding(\"images/felix.jpg\", FRmodel)\n",
    "database[\"benoit\"] = img_to_encoding(\"images/benoit.jpg\", FRmodel)\n",
    "database[\"arnaud\"] = img_to_encoding(\"images/arnaud.jpg\", FRmodel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now, when someone shows up at your front door and swipes their ID card (thus giving you their name), you can look up their encoding in the database, and use it to check if the person standing at the front door matches the name on the ID.\n",
    "\n",
    "The verify() function which checks if the front-door camera picture (`image_path`) is actually the person called \"identity\". You will have to go through the following steps:\n",
    "1. Compute the encoding of the image from image_path\n",
    "2. Compute the distance about this encoding and the encoding of the identity image stored in the database\n",
    "3. Open the door if the distance is less than 0.7, else do not open.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def verify(image_path, identity, database, model):\n",
    "    \"\"\"\n",
    "    Function that verifies if the person on the \"image_path\" image is \"identity\".\n",
    "    \n",
    "    Arguments:\n",
    "    image_path -- path to an image\n",
    "    identity -- string, name of the person you'd like to verify the identity. Has to be a resident of the Happy house.\n",
    "    database -- python dictionary mapping names of allowed people's names (strings) to their encodings (vectors).\n",
    "    model -- your Inception model instance in Keras\n",
    "    \n",
    "    Returns:\n",
    "    dist -- distance between the image_path and the image of \"identity\" in the database.\n",
    "    door_open -- True, if the door should open. False otherwise.\n",
    "    \"\"\"\n",
    "    \n",
    "    ### START CODE HERE ###\n",
    "    \n",
    "    # Step 1: Compute the encoding for the image. Use img_to_encoding() see example above. (≈ 1 line)\n",
    "    encoding = img_to_encoding(image_path, FRmodel)\n",
    "    \n",
    "    # Step 2: Compute distance with identity's image (≈ 1 line)\n",
    "    dist = np.linalg.norm(encoding-database[identity])\n",
    "    \n",
    "    # Step 3: Open the door if dist < 0.7, else don't open (≈ 3 lines)\n",
    "    if dist<0.7:\n",
    "        print(\"It's \" + str(identity) + \", welcome home!\")\n",
    "        door_open = 1\n",
    "    else:\n",
    "        print(\"It's not \" + str(identity) + \", please go away\")\n",
    "        door_open = 0\n",
    "        \n",
    "    ### END CODE HERE ###\n",
    "        \n",
    "    return dist, door_open"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's younes, welcome home!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.65939283, 1)"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "verify(\"images/camera_0.jpg\", \"younes\", database, FRmodel)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's not kian, please go away\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.86224014, 0)"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "verify(\"images/camera_2.jpg\", \"kian\", database, FRmodel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 - Face Recognition\n",
    "\n",
    "I'll implement a face recognition system that takes as input an image, and figures out if it is one of the authorized persons (and if so, who). Unlike the previous face verification system, we will no longer get a person's name as another input. \n",
    "\n",
    " `who_is_it()`. You will have to go through the following steps:\n",
    "1. Compute the target encoding of the image from image_path\n",
    "2. Find the encoding from the database that has smallest distance with the target encoding. \n",
    "    - Initialize the `min_dist` variable to a large enough number (100). It will help you keep track of what is the closest encoding to the input's encoding.\n",
    "    - Loop over the database dictionary's names and encodings. To loop use `for (name, db_enc) in database.items()`.\n",
    "        - Compute L2 distance between the target \"encoding\" and the current \"encoding\" from the database.\n",
    "        - If this distance is less than the min_dist, then set min_dist to dist, and identity to name."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def who_is_it(image_path, database, model):\n",
    "    \"\"\"\n",
    "    Implements face recognition for the happy house by finding who is the person on the image_path image.\n",
    "    \n",
    "    Arguments:\n",
    "    image_path -- path to an image\n",
    "    database -- database containing image encodings along with the name of the person on the image\n",
    "    model -- your Inception model instance in Keras\n",
    "    \n",
    "    Returns:\n",
    "    min_dist -- the minimum distance between image_path encoding and the encodings from the database\n",
    "    identity -- string, the name prediction for the person on image_path\n",
    "    \"\"\"\n",
    "    \n",
    "    ### START CODE HERE ### \n",
    "    \n",
    "    ## Step 1: Compute the target \"encoding\" for the image. Use img_to_encoding() see example above. ## (≈ 1 line)\n",
    "    encoding = img_to_encoding(image_path, model)\n",
    "    \n",
    "    ## Step 2: Find the closest encoding ##\n",
    "    \n",
    "    # Initialize \"min_dist\" to a large value, say 100 (≈1 line)\n",
    "    min_dist = 100\n",
    "    \n",
    "    # Loop over the database dictionary's names and encodings.\n",
    "    for (name, db_enc) in database.items():\n",
    "        \n",
    "        # Compute L2 distance between the target \"encoding\" and the current \"emb\" from the database. (≈ 1 line)\n",
    "        dist = np.linalg.norm(encoding-database[name])\n",
    "\n",
    "        # If this distance is less than the min_dist, then set min_dist to dist, and identity to name. (≈ 3 lines)\n",
    "        if dist<min_dist:\n",
    "            min_dist = dist\n",
    "            identity = name\n",
    "\n",
    "    ### END CODE HERE ###\n",
    "    \n",
    "    if min_dist > 0.7:\n",
    "        print(\"Not in the database.\")\n",
    "    else:\n",
    "        print (\"it's \" + str(identity) + \", the distance is \" + str(min_dist))\n",
    "        \n",
    "    return min_dist, identity"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "it's younes, the distance is 0.659393\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(0.65939283, 'younes')"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "who_is_it(\"images/camera_0.jpg\", database, FRmodel)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### References:\n",
    "\n",
    "- Florian Schroff, Dmitry Kalenichenko, James Philbin (2015). [FaceNet: A Unified Embedding for Face Recognition and Clustering](https://arxiv.org/pdf/1503.03832.pdf)\n",
    "- Yaniv Taigman, Ming Yang, Marc'Aurelio Ranzato, Lior Wolf (2014). [DeepFace: Closing the gap to human-level performance in face verification](https://research.fb.com/wp-content/uploads/2016/11/deepface-closing-the-gap-to-human-level-performance-in-face-verification.pdf) \n",
    "- The pretrained model we use is inspired by Victor Sy Wang's implementation and was loaded using his code: https://github.com/iwantooxxoox/Keras-OpenFace.\n",
    "- Our implementation also took a lot of inspiration from the official FaceNet github repository: https://github.com/davidsandberg/facenet \n"
   ]
  }
 ],
 "metadata": {
  "coursera": {
   "course_slug": "convolutional-neural-networks",
   "graded_item_id": "IaknP",
   "launcher_item_id": "5UMr4"
  },
  "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}