mnist_grad_cam.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Grad-CAM on MNIST\n",
    "\n",
    "2019-06-02\n",
    "\n",
    "This notebook demonstrates the use of Gradient Class-Activation-Map (Grad-CAM) on MNIST datset.\n",
    "\n",
    "## See:\n",
    "- https://github.com/fchollet/deep-learning-with-python-notebooks\n",
    "- [Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization (2017)](https://arxiv.org/abs/1610.02391)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'2.2.4'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import keras\n",
    "keras.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import matplotlib.image as mpimg\n",
    "import numpy as np\n",
    "import cv2\n",
    "\n",
    "from keras.models import Model\n",
    "from keras.models import load_model\n",
    "from train import prepare_mnist\n",
    "from keras import backend as K"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /home/zzandore/zzandore/anaconda3/lib/python3.7/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Colocations handled automatically by placer.\n",
      "WARNING:tensorflow:From /home/zzandore/zzandore/anaconda3/lib/python3.7/site-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
      "WARNING:tensorflow:From /home/zzandore/zzandore/anaconda3/lib/python3.7/site-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 28, 28, 1)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 26, 26, 32)        320       \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 24, 24, 64)        18496     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 12, 12, 64)        0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 12, 12, 64)        0         \n",
      "_________________________________________________________________\n",
      "flatten_1 (Flatten)          (None, 9216)              0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 128)               1179776   \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 128)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 10)                1290      \n",
      "=================================================================\n",
      "Total params: 1,199,882\n",
      "Trainable params: 1,199,882\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = load_model('model/mnist_2019-06-02')\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Check model activations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "x_train shape: (60000, 28, 28, 1)\n",
      "60000 train samples\n",
      "10000 test samples\n"
     ]
    }
   ],
   "source": [
    "(x_train, y_train), (x_test, y_test), input_shape = prepare_mnist()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5d584cef60>"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "x_train = x_train.squeeze()\n",
    "test_img = x_train[0]\n",
    "plt.imshow(test_img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.0"
      ]
     },
     "execution_count": 75,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x_train[0].max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Extract the outputs of the two conv layers\n",
    "layer_outputs = [layer.output for layer in model.layers[1:3]]\n",
    "# Create a model returning the layer_outputs for the model input\n",
    "act_model = Model(inputs=model.input, outputs=layer_outputs)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_1 (InputLayer)         (None, 28, 28, 1)         0         \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 26, 26, 32)        320       \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 24, 24, 64)        18496     \n",
      "=================================================================\n",
      "Total params: 18,816\n",
      "Trainable params: 18,816\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "act_model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "activations = act_model.predict(test_img.reshape(1,test_img.shape[0],test_img.shape[1],1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1, 26, 26, 32)\n",
      "(1, 24, 24, 64)\n"
     ]
    }
   ],
   "source": [
    "[print(act.shape) for act in activations];"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5d58477208>"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQQAAAECCAYAAAAYUakXAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4zLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvnQurowAAEEZJREFUeJzt3X+MlVV+x/HPZ4aBwQHLD5WiooiwVtdGdGcVV9O4sbqUbqpuXFuTbtmuDaZdU203aY39Y/2nqdmou22auMHVLG3UViv+2KxRKdF1G8Xyo6yCrLJVXMABBLQDssAw8+0fczlOWOY5lzv33ucOvF8JmTvPOXOeLw/wybnPPZzHESEAkKS2sgsA0DoIBAAJgQAgIRAAJAQCgIRAAJCUFgi259t+2/YvbN9ZVh3VsL3J9pu219peVXY9Q9l+2PYO2+uGHJtie5ntjZWvk8us8bBhar3b9tbKtV1re0GZNVZqmmH7Jdtv2V5v+/bK8Za7rgW11nRdXcY6BNvtkt6RdI2kLZJWSro5It5qejFVsL1JUndE7Cy7liPZ/h1JeyX9S0RcWDn2HUm7I+KeSthOjoi/LbPOSl1Hq/VuSXsj4t4yaxvK9nRJ0yNije2JklZLul7S19Vi17Wg1ptUw3Uta4ZwqaRfRMS7EXFQ0r9Juq6kWka1iHhF0u4jDl8naUnl9RIN/gUp3TC1tpyI6ImINZXXeyRtkHSGWvC6FtRak7IC4QxJm4d8v0Uj+E00QUh60fZq24vKLqYK0yKip/J6m6RpZRZThdtsv1F5S1H6NHwo2zMlXSzpdbX4dT2iVqmG68pNxepcGRGXSPo9Sd+sTH1HhRh8T9jK69MfkHSupLmSeiTdV245n7I9QdKTku6IiN6hba12XY9Sa03XtaxA2CppxpDvz6wca0kRsbXydYekpzT4lqeVba+8tzz8HnNHyfUMKyK2R0R/RAxIelAtcm1td2jwH9gjEbG0crglr+vRaq31upYVCCslzbF9ju2xkv5I0rMl1VLIdlflZo1sd0m6VtK64p8q3bOSFlZeL5T0TIm1FDr8D6ziBrXAtbVtSQ9J2hAR9w9parnrOlyttV7XUj5lkKTKxyDfk9Qu6eGI+PtSCsmwPUuDswJJGiPp0Vaq1fZjkq6SdIqk7ZK+LelpSY9LOkvS+5JuiojSb+YNU+tVGpzWhqRNkm4d8j69FLavlPRTSW9KGqgcvkuD781b6roW1HqzariupQUCgNbDTUUACYEAICEQACQEAoCEQACQlBoIo2QZsCRqbRRqbYxaay17hjBqLrCotVGotTFGZSAAaCEjWphke76kf9TgasMfRMQ9Rf3HujM63ZW+74v96nBnzedvJmptDGptjCNr3R+f6GDsd+7nag6EWjY5ObltaszrmF/T+QDUbkXf8+od2JUNhJG8ZWCTE+A4M5JAGG2bnADIGNPoE1Q+/lgkSZ06qdGnAzACI5khVLXJSUQsjojuiOgeLTdkgBPVSAJh1GxyAqA6Nb9liIhDtm+T9II+3eRkfd0qA9B0I7qHEBHPSXquTrUAKBkrFQEkBAKAhEAAkBAIABICAUBCIABICAQACYEAICEQACQEAoCEQACQEAgAEgIBQEIgAEgIBAAJgQAgafgmq2iCtux2+1lx4EAdCpE8pvivVBw6VJfznGjcMbYp52GGACAhEAAkBAKAhEAAkBAIABICAUBCIABIWIcwAu4ovnwD+/Y1qZLWwTqD0Y0ZAoCEQACQEAgAEgIBQEIgAEgIBAAJgQAgIRAAJCxMGoY7x+U79fU1vhD8mvZTT832OTTn9GyfHZ/rKmxvu2ZXdowJ4w5m++z6yfTC9rN+/FF2jHjrf7N96mFEgWB7k6Q9kvolHYqI7noUBaAc9ZghfDEidtZhHAAl4x4CgGSkgRCSXrS92vaiehQEoDwjfctwZURstX2apGW2fx4RrwztUAmKRZLUqZNGeDoAjTSiGUJEbK183SHpKUmXHqXP4ojojojuDneO5HQAGqzmQLDdZXvi4deSrpW0rl6FAWi+kbxlmCbpKduHx3k0Ip6vS1UASlFzIETEu5IuqmMtraW/P9slPntucYfV6+tUTB24+OlO7Rd8JjvEjnlTsn32nFPcflJP/ilT0179v8L2X177G9kxfuvL72T7XNRVvPBo8pj8jlf/sWluts/4D6Owva03f57838b64GNHAAmBACAhEAAkBAKAhEAAkBAIABICAUDCBinD8Lj8BikfnT+xsH3id2Zkx3j/w8nZPhO79he27107NTvG9NeKn6gUbfn1AXuu/STb5+nLvl/YfvqY/HkuefyvCttnLc1/br+5Z062zwcHZhe2j+3Nf/p/2rZ8LW0fFG9u0r/74+wYzcIMAUBCIABICAQACYEAICEQACQEAoCEQACQEAgAEhYmDSMOHMj2mbipeMFQNYuO/qn737N9ZowpXriyYnZmVxJJ33/3hsL2vgn5BUN9O8Zn+/zZz/+4sP2Lv7kxO8aNX1xR2P768s9nx5j82Opsn2YZKLuAY8AMAUBCIABICAQACYEAICEQACQEAoCEQACQEAgAEhYmDSP6incYkqSOjR8Utp/+r2dnx/jr9bdk++yfXbwAatppxU86kqT9U4sXHp1+76vZMaZle0jtU4uf7vTf512SHePjOcVPCZ/4q4NVVIJaMEMAkBAIABICAUBCIABICAQACYEAICEQACSsQxiBgcwTd8Yv782OMfOnY7N9YtaZhe0b/6b4CVKS1D87v66iHvp37S5sb1u5NzvGlJX1qgbHKjtDsP2w7R221w05NsX2MtsbK1/zWwMBaHnVvGX4oaT5Rxy7U9LyiJgjaXnlewCjXDYQIuIVSUfOA6+TtKTyeomk6+tcF4AS1HpTcVpE9FReb1N1y9wBtLgR31SMiLAdw7XbXiRpkSR1qvg/rQAoV60zhO22p0tS5euO4TpGxOKI6I6I7g531ng6AM1QayA8K2lh5fVCSc/UpxwAZarmY8fHJL0m6TzbW2zfIukeSdfY3ijpdyvfAxjlsvcQIuLmYZqurnMtx53ozz+zJ35VvPmJJLX37itsH7duUnaM+77xUGH73y36RnaMUxa/lu2TE30j39zEHfnFXKgNS5cBJAQCgIRAAJAQCAASAgFAQiAASAgEAAkbpIwCAx9sK2yf8UJXdow/n/knhe1/8ZcvZsd4aenMbJ/+nbuyfUaqHmsZJNYzHA0zBAAJgQAgIRAAJAQCgIRAAJAQCAASAgFAQiAASFiYNArkNlpp27g5O8bMpecVtj9xxiXZMf555aPZPn+6+PbC9jP/4dXsGM2SW+B0Ii5cYoYAICEQACQEAoCEQACQEAgAEgIBQEIgAEgIBACJI4Z9cHPdndw2NeZ1zG/a+fCptinFT3f66OpZ2THGfb145yZJevnCpwvb79+dP89//uHnC9v717+dHaMejqeFSSv6nlfvwC7n+jFDAJAQCAASAgFAQiAASAgEAAmBACAhEAAkbJByghjY/XFh++TnNmTHONhzbrbPrBtvLWxf8Qf3Z8d48O4rCtvP+mp2iLqo5glRx9NaBamKGYLth23vsL1uyLG7bW+1vbbya0FjywTQDNW8ZfihpKMtL/xuRMyt/HquvmUBKEM2ECLiFUm7m1ALgJKN5KbibbbfqLylmFy3igCUptZAeEDSuZLmSuqRdN9wHW0vsr3K9qq+2F/j6QA0Q02BEBHbI6I/IgYkPSjp0oK+iyOiOyK6O9xZa50AmqCmQLA9fci3N0haN1xfAKNHdh2C7cckXSXpFNtbJH1b0lW250oKSZskFX/4DGBUYIMUSJLcOS7bZ2DPniZUIu396mWF7eM/7MuO0f7ymnqVU2i0LExigxQAx4xAAJAQCAASAgFAQiAASAgEAAmBACBhg5TjgMd2ZPsM7NtX2B578puBNMuEJ14vbD/4pe7sGGM/k9/Mxb17izuMzy+179/Sk+0zmjBDAJAQCAASAgFAQiAASAgEAAmBACAhEAAkBAKAhIVJJXNH/o8g+vsL2wc++aRe5bSE/qsuKWx/f0EV1+zkSdk+sx+aWNjevjL/NKvjDTMEAAmBACAhEAAkBAKAhEAAkBAIABICAUBCIABIWJg0ArlFRbldiiQp+lpnp6Kc9mmnZfvs+cI5he1bvpR/UtjNl60obL9wIP+0pGf+Z262z8FJxeOMH2jeU81aBTMEAAmBACAhEAAkBAKAhEAAkBAIABICAUByQq5DcHs+Bz2hK9unf+euepTTFB5T/Ee978vFm5JI0uYF+c/lv/K5lYXt85wf40ebLixsb3slv/nJ+T/enu0z8N7mbJ8TTfZfhu0Ztl+y/Zbt9bZvrxyfYnuZ7Y2Vr5MbXy6ARqrmLcMhSd+KiAskzZP0TdsXSLpT0vKImCNpeeV7AKNYNhAioici1lRe75G0QdIZkq6TtKTSbYmk6xtVJIDmOKabirZnSrpY0uuSpkXE4UffbpM0ra6VAWi6qgPB9gRJT0q6IyJ6h7ZFREg66t0i24tsr7K9qi/2j6hYAI1VVSDY7tBgGDwSEUsrh7fbnl5pny5px9F+NiIWR0R3RHR3uLMeNQNokGo+ZbCkhyRtiIj7hzQ9K2lh5fVCSc/UvzwAzVTNOoQrJH1N0pu211aO3SXpHkmP275F0vuSbmpMiQCaJRsIEfFfkjxM89X1LSevbXz+bUd/b29he/RVcaL9o+d+R1x+UbZPzxeKF1pNX/DL7BhfmfRBts+yzecVtvvl/HKVGS98WNg+8O7q7BgDJ+DmJvXA0mUACYEAICEQACQEAoCEQACQEAgAEgIBQEIgAEhG3Y5Jnjgh3ymzMKlZ2j9bvEhHkt67cWq2z/7phwrbL//tjdkxLu3aWdj+xNsXZ8fY9qOzsn3OXFa8i1S8t7awXZIG+op/v2gcZggAEgIBQEIgAEgIBAAJgQAgIRAAJAQCgGTUrUOIg/ndTXYuurywfeD3P8qOsXD2imyfa7o2FLb3xevZMTYdyq9DmNS2r7D9e1uuyY7x6E+uKGyf9dTB7BgdK3+W7TNQxZ8PWhczBAAJgQAgIRAAJAQCgIRAAJAQCAASAgFAQiAASEbdwiRPLH4CkSQd6hruQVODzp70cXaMSe3Fi4Ekadkn5xe2v7wzv0HKz97Jbzoy4e2OwvbT1hzIjnHeyvWF7bE/PwbPQjr+MUMAkBAIABICAUBCIABICAQACYEAICEQACSOaN6nyye3TY15HfObdj4Ag1b0Pa/egV3FC3RUxQzB9gzbL9l+y/Z627dXjt9te6vttZVfC+pROIDyVLNS8ZCkb0XEGtsTJa22vazS9t2IuLdx5QFopmwgRESPpJ7K6z22N0g6o9GFAWi+Y7qpaHumpIslHd499Dbbb9h+2PbkOtcGoMmqDgTbEyQ9KemOiOiV9ICkcyXN1eAM4r5hfm6R7VW2V/XF/jqUDKBRqgoE2x0aDINHImKpJEXE9ojoj4gBSQ9KuvRoPxsRiyOiOyK6O9xZr7oBNEA1nzJY0kOSNkTE/UOOTx/S7QZJ6+pfHoBmquZThiskfU3Sm7bXVo7dJelm23M1+N/kN0m6tSEVAmiapi5Msv2hpPeHHDpF0s6mFTAy1NoY1NoYR9Z6dkScmvuhpgbCr53cXhUR3aUVcAyotTGotTFqrZX/ywAgIRAAJGUHwuKSz38sqLUxqLUxaqq11HsIAFpL2TMEAC2EQACQEAgAEgIBQEIgAEj+HyQGmFrmi0DzAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 288x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.matshow(activations[0][0,:,:,31], cmap='viridis')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Make grid plots\n",
    "def plot_conv_layer(which_conv):\n",
    "    # Plot conv layer activation\n",
    "    # which_conv = 0 # 0=first, 1=second\n",
    "    img_per_row = 10\n",
    "    n_feat = activations[which_conv].shape[-1] # number of feature maps\n",
    "    size = activations[which_conv].shape[1] # image feature (e.g., 26)\n",
    "    n_col = np.round(n_feat/img_per_row).astype('int')\n",
    "    display_grid = np.zeros((size*n_col, img_per_row*size))\n",
    "    # Add each filter into the grids\n",
    "    for col in range(n_col):\n",
    "        for row in range(img_per_row):\n",
    "            if col*img_per_row+row < size:\n",
    "                img = activations[which_conv][0,:,:,col*img_per_row+row]\n",
    "                # Post-process images\n",
    "                img -= img.mean()\n",
    "                img -= img.std()\n",
    "                img *= 64\n",
    "                img += 128\n",
    "                img = np.clip(img, 0, 255).astype('uint8')\n",
    "                display_grid[col*size:(col+1)*size,\n",
    "                             row*size:(row+1)*size] = img\n",
    "    scale = 1/size\n",
    "    fig, ax = plt.subplots(figsize=(scale*display_grid.shape[1], scale*display_grid.shape[0]))\n",
    "    ax.set_title(f'ConvLayer = {which_conv+1}')\n",
    "    ax.grid(False)\n",
    "    ax.imshow(display_grid, aspect='auto', cmap='viridis')\n",
    "    return fig, ax"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<Figure size 720x216 with 1 Axes>,\n",
       " <matplotlib.axes._subplots.AxesSubplot at 0x7f5d5842ebe0>)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x216 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_conv_layer(0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(<Figure size 720x432 with 1 Axes>,\n",
       " <matplotlib.axes._subplots.AxesSubplot at 0x7f5d583a1e80>)"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_conv_layer(1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Visualize Grad-CAM\n",
    "\n",
    "Grad-CAM visualizes input features that are most likely to generate the target."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "which_number = 6\n",
    "# Convert ont-hot vector to integer\n",
    "y_test_int = np.where(y_test==1)[1]\n",
    "# Randomly select one example\n",
    "rand_idx = np.random.choice(np.where(y_test_int == which_number)[0])\n",
    "test_input = x_test[rand_idx].reshape(1,x_test.shape[1],x_test.shape[2],1)\n",
    "# Visualize test_input\n",
    "plt.imshow(test_input.squeeze());"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Get output vector\n",
    "output_vector = model.output[:, which_number]\n",
    "# Get the last convolutional layer\n",
    "last_conv_layer = model.layers[2]\n",
    "# Get the gradient of the given number with regard to the output feature map of the last conv layer\n",
    "grads = K.gradients(output_vector, last_conv_layer.output)[0] # (None,24,24,64)\n",
    "# Get the mean intensity of the gradient over each feature map (64)\n",
    "pooled_grads = K.mean(grads, axis=(0, 1, 2)) # 64\n",
    "# Compute gradient given an inputw\n",
    "iterate = K.function([model.input], [pooled_grads, last_conv_layer.output[0]])\n",
    "pooled_grads_value, conv_layer_output_value = iterate([test_input]) # 64, (24,24,64)\n",
    "# Multiply each channel in the feature map array\n",
    "# by 'how important this channel is' with regard to the given number\n",
    "for i in range(len(pooled_grads_value)):\n",
    "    conv_layer_output_value[:,:,i] *= pooled_grads_value[i]\n",
    "# Calculate channel-wise mean for the heatmap activation\n",
    "heatmap = np.mean(conv_layer_output_value, axis=-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5cd472b7b8>"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 288x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "heatmap = np.maximum(heatmap, 0) # => max(heatmap, 0)\n",
    "heatmap /= np.max(heatmap)\n",
    "plt.matshow(heatmap)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Add 3 channels for coloring the heatmap\n",
    "img_size = (500, 500)\n",
    "# Prepare image\n",
    "test_input_resized = cv2.resize(test_input.squeeze(), img_size) # (500, 500)\n",
    "test_input_resized = cv2.cvtColor(test_input_resized, cv2.COLOR_GRAY2BGR) # (500, 500, 3)\n",
    "test_input_resized *= 255\n",
    "# Make heatmap\n",
    "heatmap_resized = cv2.resize(heatmap, img_size) # (500, 500)\n",
    "heatmap_resized = np.uint8(255*heatmap_resized)\n",
    "heatmap_resized = cv2.applyColorMap(heatmap_resized, cv2.COLORMAP_JET)\n",
    "# Apply heatmap on the original image\n",
    "superimposed_img = heatmap_resized*0.5 + test_input_resized\n",
    "# Save\n",
    "cv2.imwrite('outputs/test.jpg', superimposed_img);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.image.AxesImage at 0x7f5cd46f4c50>"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "img = mpimg.imread('outputs/test.jpg')\n",
    "plt.imshow(img)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [],
   "source": [
    "# # Generate input JPEG files for testing\n",
    "# for number in range(10):\n",
    "#     # Convert ont-hot vector to integer\n",
    "#     y_test_int = np.where(y_test==1)[1]\n",
    "#     # Randomly select one example\n",
    "#     rand_idx = np.random.choice(np.where(y_test_int == number)[0])\n",
    "#     test_input = x_test[rand_idx].reshape(1,x_test.shape[1],x_test.shape[2],1)\n",
    "#     # Resize\n",
    "#     img_size = (500, 500)\n",
    "#     # Prepare image\n",
    "#     test_input_resized = cv2.resize(test_input.squeeze(), img_size) # (500, 500)\n",
    "#     test_input_resized = cv2.cvtColor(test_input_resized, cv2.COLOR_GRAY2BGR) # (500, 500, 3)\n",
    "#     test_input_resized /= test_input_resized.max()\n",
    "#     test_input_resized *= 255\n",
    "#     cv2.imwrite(f'examples/{number:02}.jpg', test_input_resized);"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}