16/dl_16.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Residual Networks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import h5py\n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from PIL import Image\n",
    "import urllib.request\n",
    "import zipfile"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_dataset = h5py.File('train_signs.h5', \"r\")\n",
    "x_train = np.array(train_dataset[\"train_set_x\"][:]) \n",
    "y_train = np.array(train_dataset[\"train_set_y\"][:])\n",
    "\n",
    "test_dataset = h5py.File('test_signs.h5', \"r\")\n",
    "x_test = np.array(test_dataset[\"test_set_x\"][:]) \n",
    "y_test = np.array(test_dataset[\"test_set_y\"][:])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train = np.array(x_train, dtype='float32')\n",
    "Y_train = np.array(np.reshape(np.eye(6)[y_train],(-1,6)), dtype='float32')\n",
    "\n",
    "X_test = np.array(x_test, dtype='float32')\n",
    "Y_test = np.array(np.reshape(np.eye(6)[y_test],(-1,6)), dtype='float32')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's a 2\n"
     ]
    },
    {
     "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": [
    "index = 6\n",
    "plt.imshow(np.array(X_train[index,:,:,:], dtype='int32'))\n",
    "print(\"It's a {}\".format(y_train[index]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Identity_Block(tf.keras.layers.Layer):\n",
    "    \n",
    "    def __init__(self, kernel_size, filters):\n",
    "        super().__init__()\n",
    "        self.f1 = tf.keras.layers.Conv2D(filters=filters[0], \n",
    "                                         kernel_size=(1,1), \n",
    "                                         strides=(1,1), \n",
    "                                         padding='valid', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f2 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f3 = tf.keras.layers.Activation(activation='relu')\n",
    "        self.f4 = tf.keras.layers.Conv2D(filters=filters[1], \n",
    "                                         kernel_size=kernel_size, \n",
    "                                         strides=(1,1), \n",
    "                                         padding='same', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f5 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f6 = tf.keras.layers.Activation(activation='relu')\n",
    "        self.f7 = tf.keras.layers.Conv2D(filters=filters[2], \n",
    "                                         kernel_size=(1,1), \n",
    "                                         strides=(1,1), \n",
    "                                         padding='valid', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f8 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f9 = tf.keras.layers.Add()\n",
    "        self.f10 = tf.keras.layers.Activation(activation='relu')\n",
    "        \n",
    "    def call(self, inputs):\n",
    "        a1 = self.f1(inputs)\n",
    "        a2 = self.f2(a1)\n",
    "        a3 = self.f3(a2)\n",
    "        a4 = self.f4(a3)\n",
    "        a5 = self.f5(a4)\n",
    "        a6 = self.f6(a5)\n",
    "        a7 = self.f7(a6)\n",
    "        a8 = self.f8(a7)\n",
    "        a9 = self.f9([a8,inputs])\n",
    "        y = self.f10(a9)\n",
    "        return y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Convolutional_Block(tf.keras.layers.Layer):\n",
    "    \n",
    "    def __init__(self, kernel_size, filters, strides):\n",
    "        super().__init__()\n",
    "        self.f1 = tf.keras.layers.Conv2D(filters=filters[0], \n",
    "                                         kernel_size=(1,1), \n",
    "                                         strides=strides, \n",
    "                                         padding='valid', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f2 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f3 = tf.keras.layers.Activation(activation='relu')\n",
    "        self.f4 = tf.keras.layers.Conv2D(filters=filters[1], \n",
    "                                         kernel_size=kernel_size, \n",
    "                                         strides=(1,1), \n",
    "                                         padding='same', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f5 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f6 = tf.keras.layers.Activation(activation='relu')\n",
    "        self.f7 = tf.keras.layers.Conv2D(filters=filters[2], \n",
    "                                         kernel_size=(1,1), \n",
    "                                         strides=(1,1), \n",
    "                                         padding='valid', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f8 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f9 = tf.keras.layers.Conv2D(filters=filters[2], \n",
    "                                         kernel_size=(1,1), \n",
    "                                         strides=strides, \n",
    "                                         padding='valid', \n",
    "                                         activation='linear', \n",
    "                                         kernel_initializer='glorot_uniform', \n",
    "                                         bias_initializer='zeros')\n",
    "        self.f10 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f11 = tf.keras.layers.Add()\n",
    "        self.f12 = tf.keras.layers.Activation(activation='relu')\n",
    "    \n",
    "    def call(self, inputs):\n",
    "        a1 = self.f1(inputs)\n",
    "        a2 = self.f2(a1)\n",
    "        a3 = self.f3(a2)\n",
    "        a4 = self.f4(a3)\n",
    "        a5 = self.f5(a4)\n",
    "        a6 = self.f6(a5)\n",
    "        a7 = self.f7(a6)\n",
    "        a8 = self.f8(a7)\n",
    "        a9 = self.f9(inputs)\n",
    "        a10 = self.f10(a9)\n",
    "        a11 = self.f11([a8,a10])\n",
    "        y = self.f12(a11)\n",
    "        return y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class ResNet50(tf.keras.Model):\n",
    "    \n",
    "    def __init__(self, input_dim, output_dim):\n",
    "        super().__init__()\n",
    "        self.input_dim = input_dim\n",
    "        self.output_dim = output_dim\n",
    "        self.f1 = tf.keras.layers.Lambda(function=lambda x: x/255.)\n",
    "        self.f2 = tf.keras.layers.ZeroPadding2D(padding=(3,3))\n",
    "        self.f3 = tf.keras.layers.Conv2D(filters=64, \n",
    "                                    kernel_size=(7,7), \n",
    "                                    strides=(2,2), \n",
    "                                    padding='valid', \n",
    "                                    activation='linear', \n",
    "                                    kernel_initializer='glorot_uniform', \n",
    "                                    bias_initializer='zeros')\n",
    "        self.f4 = tf.keras.layers.BatchNormalization(axis=3)\n",
    "        self.f5 = tf.keras.layers.Activation(activation='relu')\n",
    "        self.f6 = tf.keras.layers.MaxPooling2D(pool_size=(3,3), \n",
    "                                          strides=(2,2), \n",
    "                                          padding='valid')\n",
    "        self.f7 = Convolutional_Block(kernel_size=(3,3), \n",
    "                                 filters=[64, 64, 256], \n",
    "                                 strides=(1,1))\n",
    "        self.f8 = Identity_Block(kernel_size=(3,3), \n",
    "                            filters=[64, 64, 256])\n",
    "        self.f9 = Identity_Block(kernel_size=(3,3), \n",
    "                            filters=[64, 64, 256])\n",
    "        self.f10 = Convolutional_Block(kernel_size=(3,3), \n",
    "                                  filters=[128, 128, 512],\n",
    "                                  strides=(2,2))\n",
    "        self.f11 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[128, 128, 512])\n",
    "        self.f12 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[128, 128, 512])\n",
    "        self.f13 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[128, 128, 512])\n",
    "        self.f14 = Convolutional_Block(kernel_size=(3,3), \n",
    "                                  filters=[256, 256, 1024],\n",
    "                                  strides=(2,2))\n",
    "        self.f15 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[256, 256, 1024])\n",
    "        self.f16 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[256, 256, 1024])\n",
    "        self.f17 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[256, 256, 1024])\n",
    "        self.f18 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[256, 256, 1024])\n",
    "        self.f19 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[256, 256, 1024])\n",
    "        self.f20 = Convolutional_Block(kernel_size=(3,3), \n",
    "                                  filters=[512, 512, 2048],\n",
    "                                  strides=(2,2))\n",
    "        self.f21 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[512, 512, 2048])\n",
    "        self.f22 = Identity_Block(kernel_size=(3,3), \n",
    "                             filters=[512, 512, 2048])\n",
    "        self.f23 = tf.keras.layers.AveragePooling2D(pool_size=(2,2), \n",
    "                                               strides=None, \n",
    "                                               padding='valid')\n",
    "        self.f24 = tf.keras.layers.Flatten()\n",
    "        self.f25 = tf.keras.layers.Dense(units=self.output_dim, \n",
    "                                    activation='softmax', \n",
    "                                    kernel_initializer='glorot_uniform', \n",
    "                                    bias_initializer='zeros')\n",
    "        \n",
    "    def call(self, inputs): \n",
    "        a1 = self.f1(inputs)\n",
    "        a2 = self.f2(a1)\n",
    "        a3 = self.f3(a2)\n",
    "        a4 = self.f4(a3)\n",
    "        a5 = self.f5(a4)\n",
    "        a6 = self.f6(a5)\n",
    "        a7 = self.f7(a6)\n",
    "        a8 = self.f8(a7)\n",
    "        a9 = self.f9(a8)\n",
    "        a10 = self.f10(a9)\n",
    "        a11 = self.f11(a10)\n",
    "        a12 = self.f12(a11)\n",
    "        a13 = self.f13(a12)\n",
    "        a14 = self.f14(a13)\n",
    "        a15 = self.f15(a14)\n",
    "        a16 = self.f16(a15)\n",
    "        a17 = self.f17(a16)\n",
    "        a18 = self.f18(a17)\n",
    "        a19 = self.f19(a18)\n",
    "        a20 = self.f20(a19)\n",
    "        a21 = self.f21(a20)\n",
    "        a22 = self.f22(a21)\n",
    "        a23 = self.f23(a22)\n",
    "        a24 = self.f24(a23)\n",
    "        y = self.f25(a24)\n",
    "        return y\n",
    "    \n",
    "    def fit(self, x, y, learning_rate=0.01, batch_size=64, epochs=100, beta_1=0.9, beta_2=0.999, epsilon=1e-07, verbose=True):\n",
    "        loss = tf.keras.losses.CategoricalCrossentropy()\n",
    "        metric = tf.keras.metrics.CategoricalAccuracy()\n",
    "        optimizer = tf.keras.optimizers.Adam(learning_rate=learning_rate, beta_1=beta_1, beta_2=beta_2, epsilon=epsilon)\n",
    "        self.compile(loss=loss, metrics=[metric], optimizer=optimizer)\n",
    "        self.summary(input_dim=self.input_dim)\n",
    "        super().fit(x=x, y=y, batch_size=batch_size, epochs=epochs, verbose=verbose)\n",
    "        \n",
    "    def summary(self, input_dim):\n",
    "        x = tf.keras.Input(shape=input_dim)\n",
    "        y = self(x)\n",
    "        super().summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"res_net50\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "lambda (Lambda)              (None, 64, 64, 3)         0         \n",
      "_________________________________________________________________\n",
      "zero_padding2d (ZeroPadding2 (None, 70, 70, 3)         0         \n",
      "_________________________________________________________________\n",
      "conv2d (Conv2D)              (None, 32, 32, 64)        9472      \n",
      "_________________________________________________________________\n",
      "batch_normalization (BatchNo (None, 32, 32, 64)        256       \n",
      "_________________________________________________________________\n",
      "activation (Activation)      (None, 32, 32, 64)        0         \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 15, 15, 64)        0         \n",
      "_________________________________________________________________\n",
      "convolutional__block (Convol (None, 15, 15, 256)       76928     \n",
      "_________________________________________________________________\n",
      "identity__block (Identity_Bl (None, 15, 15, 256)       71552     \n",
      "_________________________________________________________________\n",
      "identity__block_1 (Identity_ (None, 15, 15, 256)       71552     \n",
      "_________________________________________________________________\n",
      "convolutional__block_1 (Conv (None, 8, 8, 512)         383232    \n",
      "_________________________________________________________________\n",
      "identity__block_2 (Identity_ (None, 8, 8, 512)         282368    \n",
      "_________________________________________________________________\n",
      "identity__block_3 (Identity_ (None, 8, 8, 512)         282368    \n",
      "_________________________________________________________________\n",
      "identity__block_4 (Identity_ (None, 8, 8, 512)         282368    \n",
      "_________________________________________________________________\n",
      "convolutional__block_2 (Conv (None, 4, 4, 1024)        1520128   \n",
      "_________________________________________________________________\n",
      "identity__block_5 (Identity_ (None, 4, 4, 1024)        1121792   \n",
      "_________________________________________________________________\n",
      "identity__block_6 (Identity_ (None, 4, 4, 1024)        1121792   \n",
      "_________________________________________________________________\n",
      "identity__block_7 (Identity_ (None, 4, 4, 1024)        1121792   \n",
      "_________________________________________________________________\n",
      "identity__block_8 (Identity_ (None, 4, 4, 1024)        1121792   \n",
      "_________________________________________________________________\n",
      "identity__block_9 (Identity_ (None, 4, 4, 1024)        1121792   \n",
      "_________________________________________________________________\n",
      "convolutional__block_3 (Conv (None, 2, 2, 2048)        6054912   \n",
      "_________________________________________________________________\n",
      "identity__block_10 (Identity (None, 2, 2, 2048)        4471808   \n",
      "_________________________________________________________________\n",
      "identity__block_11 (Identity (None, 2, 2, 2048)        4471808   \n",
      "_________________________________________________________________\n",
      "average_pooling2d (AveragePo (None, 1, 1, 2048)        0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 2048)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 6)                 12294     \n",
      "=================================================================\n",
      "Total params: 23,600,006\n",
      "Trainable params: 23,546,886\n",
      "Non-trainable params: 53,120\n",
      "_________________________________________________________________\n",
      "Train on 1080 samples\n",
      "Epoch 1/2\n",
      "1080/1080 [==============================] - 45s 42ms/sample - loss: 1.8651 - categorical_accuracy: 0.4472\n",
      "Epoch 2/2\n",
      "1080/1080 [==============================] - 39s 36ms/sample - loss: 0.5438 - categorical_accuracy: 0.7981\n"
     ]
    }
   ],
   "source": [
    "resnet50 = ResNet50(input_dim=X_train.shape[1:], \n",
    "                    output_dim=Y_train.shape[1])\n",
    "\n",
    "resnet50.fit(x=X_train, y=Y_train, epochs=2, learning_rate=0.001, batch_size=32)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1080/1080 [==============================] - 5s 5ms/sample - loss: 7.5725 - categorical_accuracy: 0.1667\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[7.57253857365361, 0.16666667]"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50.evaluate(X_train, Y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120/120 [==============================] - 0s 4ms/sample - loss: 7.5745 - categorical_accuracy: 0.1667\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[7.574506187438965, 0.16666667]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50.evaluate(X_test, Y_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "-----"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "X_train_ = X_train/255.\n",
    "Y_train_ = Y_train\n",
    "\n",
    "X_test_ = X_test/255.\n",
    "Y_test_ = Y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "url = 'https://firebasestorage.googleapis.com/v0/b/z2tma61d2a74hya815w9x621uszb3a.appspot.com/o/ResNet50.h5.zip?alt=media&token=51524b8b-7f10-4c4e-993f-e9f30348a301'\n",
    "local_zip = 'ResNet50.h5.zip'\n",
    "urllib.request.urlretrieve(url, local_zip)\n",
    "zip_ref = zipfile.ZipFile(local_zip, 'r')\n",
    "zip_ref.extractall()\n",
    "zip_ref.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"ResNet50\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 64, 64, 3)]  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "zero_padding2d_1 (ZeroPadding2D (None, 70, 70, 3)    0           input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1 (Conv2D)                  (None, 32, 32, 64)   9472        zero_padding2d_1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "bn_conv1 (BatchNormalization)   (None, 32, 32, 64)   256         conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_4 (Activation)       (None, 32, 32, 64)   0           bn_conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2D)  (None, 15, 15, 64)   0           activation_4[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2a (Conv2D)         (None, 15, 15, 64)   4160        max_pooling2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_5 (Activation)       (None, 15, 15, 64)   0           bn2a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_5[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_6 (Activation)       (None, 15, 15, 64)   0           bn2a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_6[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "res2a_branch1 (Conv2D)          (None, 15, 15, 256)  16640       max_pooling2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn2a_branch1 (BatchNormalizatio (None, 15, 15, 256)  1024        res2a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_2 (Add)                     (None, 15, 15, 256)  0           bn2a_branch2c[0][0]              \n",
      "                                                                 bn2a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_7 (Activation)       (None, 15, 15, 256)  0           add_2[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2a (Conv2D)         (None, 15, 15, 64)   16448       activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_8 (Activation)       (None, 15, 15, 64)   0           bn2b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_8[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_9 (Activation)       (None, 15, 15, 64)   0           bn2b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2b_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "bn2b_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_3 (Add)                     (None, 15, 15, 256)  0           bn2b_branch2c[0][0]              \n",
      "                                                                 activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_10 (Activation)      (None, 15, 15, 256)  0           add_3[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2a (Conv2D)         (None, 15, 15, 64)   16448       activation_10[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2a (BatchNormalizati (None, 15, 15, 64)   256         res2c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_11 (Activation)      (None, 15, 15, 64)   0           bn2c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2b (Conv2D)         (None, 15, 15, 64)   36928       activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2b (BatchNormalizati (None, 15, 15, 64)   256         res2c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_12 (Activation)      (None, 15, 15, 64)   0           bn2c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res2c_branch2c (Conv2D)         (None, 15, 15, 256)  16640       activation_12[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn2c_branch2c (BatchNormalizati (None, 15, 15, 256)  1024        res2c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_4 (Add)                     (None, 15, 15, 256)  0           bn2c_branch2c[0][0]              \n",
      "                                                                 activation_10[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_13 (Activation)      (None, 15, 15, 256)  0           add_4[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2a (Conv2D)         (None, 8, 8, 128)    32896       activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_14 (Activation)      (None, 8, 8, 128)    0           bn3a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_14[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_15 (Activation)      (None, 8, 8, 128)    0           bn3a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3a_branch1 (Conv2D)          (None, 8, 8, 512)    131584      activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn3a_branch1 (BatchNormalizatio (None, 8, 8, 512)    2048        res3a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_5 (Add)                     (None, 8, 8, 512)    0           bn3a_branch2c[0][0]              \n",
      "                                                                 bn3a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_16 (Activation)      (None, 8, 8, 512)    0           add_5[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_16[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_17 (Activation)      (None, 8, 8, 128)    0           bn3b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_17[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_18 (Activation)      (None, 8, 8, 128)    0           bn3b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3b_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_18[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3b_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_6 (Add)                     (None, 8, 8, 512)    0           bn3b_branch2c[0][0]              \n",
      "                                                                 activation_16[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_19 (Activation)      (None, 8, 8, 512)    0           add_6[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_19[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_20 (Activation)      (None, 8, 8, 128)    0           bn3c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_20[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_21 (Activation)      (None, 8, 8, 128)    0           bn3c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3c_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_21[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3c_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_7 (Add)                     (None, 8, 8, 512)    0           bn3c_branch2c[0][0]              \n",
      "                                                                 activation_19[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_22 (Activation)      (None, 8, 8, 512)    0           add_7[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2a (Conv2D)         (None, 8, 8, 128)    65664       activation_22[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2a (BatchNormalizati (None, 8, 8, 128)    512         res3d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_23 (Activation)      (None, 8, 8, 128)    0           bn3d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2b (Conv2D)         (None, 8, 8, 128)    147584      activation_23[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2b (BatchNormalizati (None, 8, 8, 128)    512         res3d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_24 (Activation)      (None, 8, 8, 128)    0           bn3d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res3d_branch2c (Conv2D)         (None, 8, 8, 512)    66048       activation_24[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn3d_branch2c (BatchNormalizati (None, 8, 8, 512)    2048        res3d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_8 (Add)                     (None, 8, 8, 512)    0           bn3d_branch2c[0][0]              \n",
      "                                                                 activation_22[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_25 (Activation)      (None, 8, 8, 512)    0           add_8[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2a (Conv2D)         (None, 4, 4, 256)    131328      activation_25[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_26 (Activation)      (None, 4, 4, 256)    0           bn4a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_26[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_27 (Activation)      (None, 4, 4, 256)    0           bn4a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_27[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4a_branch1 (Conv2D)          (None, 4, 4, 1024)   525312      activation_25[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn4a_branch1 (BatchNormalizatio (None, 4, 4, 1024)   4096        res4a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_9 (Add)                     (None, 4, 4, 1024)   0           bn4a_branch2c[0][0]              \n",
      "                                                                 bn4a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_28 (Activation)      (None, 4, 4, 1024)   0           add_9[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_28[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_29 (Activation)      (None, 4, 4, 256)    0           bn4b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_29[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_30 (Activation)      (None, 4, 4, 256)    0           bn4b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4b_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_30[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4b_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_10 (Add)                    (None, 4, 4, 1024)   0           bn4b_branch2c[0][0]              \n",
      "                                                                 activation_28[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_31 (Activation)      (None, 4, 4, 1024)   0           add_10[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_31[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_32 (Activation)      (None, 4, 4, 256)    0           bn4c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_32[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_33 (Activation)      (None, 4, 4, 256)    0           bn4c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4c_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_33[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4c_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_11 (Add)                    (None, 4, 4, 1024)   0           bn4c_branch2c[0][0]              \n",
      "                                                                 activation_31[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_34 (Activation)      (None, 4, 4, 1024)   0           add_11[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_34[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4d_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_35 (Activation)      (None, 4, 4, 256)    0           bn4d_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_35[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4d_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_36 (Activation)      (None, 4, 4, 256)    0           bn4d_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4d_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_36[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4d_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4d_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_12 (Add)                    (None, 4, 4, 1024)   0           bn4d_branch2c[0][0]              \n",
      "                                                                 activation_34[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_37 (Activation)      (None, 4, 4, 1024)   0           add_12[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_37[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4e_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_38 (Activation)      (None, 4, 4, 256)    0           bn4e_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_38[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4e_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_39 (Activation)      (None, 4, 4, 256)    0           bn4e_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4e_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_39[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4e_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4e_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_13 (Add)                    (None, 4, 4, 1024)   0           bn4e_branch2c[0][0]              \n",
      "                                                                 activation_37[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_40 (Activation)      (None, 4, 4, 1024)   0           add_13[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2a (Conv2D)         (None, 4, 4, 256)    262400      activation_40[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2a (BatchNormalizati (None, 4, 4, 256)    1024        res4f_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_41 (Activation)      (None, 4, 4, 256)    0           bn4f_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2b (Conv2D)         (None, 4, 4, 256)    590080      activation_41[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2b (BatchNormalizati (None, 4, 4, 256)    1024        res4f_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_42 (Activation)      (None, 4, 4, 256)    0           bn4f_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res4f_branch2c (Conv2D)         (None, 4, 4, 1024)   263168      activation_42[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn4f_branch2c (BatchNormalizati (None, 4, 4, 1024)   4096        res4f_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_14 (Add)                    (None, 4, 4, 1024)   0           bn4f_branch2c[0][0]              \n",
      "                                                                 activation_40[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_43 (Activation)      (None, 4, 4, 1024)   0           add_14[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2a (Conv2D)         (None, 2, 2, 512)    524800      activation_43[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5a_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_44 (Activation)      (None, 2, 2, 512)    0           bn5a_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_44[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5a_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_45 (Activation)      (None, 2, 2, 512)    0           bn5a_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_45[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5a_branch1 (Conv2D)          (None, 2, 2, 2048)   2099200     activation_43[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5a_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "bn5a_branch1 (BatchNormalizatio (None, 2, 2, 2048)   8192        res5a_branch1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_15 (Add)                    (None, 2, 2, 2048)   0           bn5a_branch2c[0][0]              \n",
      "                                                                 bn5a_branch1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_46 (Activation)      (None, 2, 2, 2048)   0           add_15[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2a (Conv2D)         (None, 2, 2, 512)    1049088     activation_46[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5b_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_47 (Activation)      (None, 2, 2, 512)    0           bn5b_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_47[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5b_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_48 (Activation)      (None, 2, 2, 512)    0           bn5b_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5b_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_48[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5b_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5b_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_16 (Add)                    (None, 2, 2, 2048)   0           bn5b_branch2c[0][0]              \n",
      "                                                                 activation_46[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_49 (Activation)      (None, 2, 2, 2048)   0           add_16[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2a (Conv2D)         (None, 2, 2, 512)    1049088     activation_49[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2a (BatchNormalizati (None, 2, 2, 512)    2048        res5c_branch2a[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_50 (Activation)      (None, 2, 2, 512)    0           bn5c_branch2a[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2b (Conv2D)         (None, 2, 2, 512)    2359808     activation_50[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2b (BatchNormalizati (None, 2, 2, 512)    2048        res5c_branch2b[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "activation_51 (Activation)      (None, 2, 2, 512)    0           bn5c_branch2b[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "res5c_branch2c (Conv2D)         (None, 2, 2, 2048)   1050624     activation_51[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "bn5c_branch2c (BatchNormalizati (None, 2, 2, 2048)   8192        res5c_branch2c[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "add_17 (Add)                    (None, 2, 2, 2048)   0           bn5c_branch2c[0][0]              \n",
      "                                                                 activation_49[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_52 (Activation)      (None, 2, 2, 2048)   0           add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "avg_pool (AveragePooling2D)     (None, 1, 1, 2048)   0           activation_52[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "flatten_1 (Flatten)             (None, 2048)         0           avg_pool[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "fc6 (Dense)                     (None, 6)            12294       flatten_1[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 23,600,006\n",
      "Trainable params: 23,546,886\n",
      "Non-trainable params: 53,120\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "resnet50_ = tf.keras.models.load_model(filepath='ResNet50.h5')\n",
    "resnet50_.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1080/1080 [==============================] - 5s 5ms/sample - loss: 0.3063 - accuracy: 0.9056\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.3063314074719394, 0.90555555]"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50_.evaluate(X_train_, Y_train_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120/120 [==============================] - 0s 3ms/sample - loss: 0.5302 - accuracy: 0.8667\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "[0.5301782925923665, 0.8666667]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "resnet50_.evaluate(X_test_, Y_test_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's a 2\n"
     ]
    },
    {
     "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": [
    "index = 6\n",
    "plt.imshow(np.array(X_train[index,:,:,:], dtype='int32'))\n",
    "print(\"It's a {}\".format(np.argmax(resnet50_.predict(X_train[[index],:,:,:]/255.))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's a 3\n"
     ]
    },
    {
     "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 = np.array(plt.imread(\"thumbs_up.jpg\"))\n",
    "img_ = np.array(Image.fromarray(img).resize((64,64)))\n",
    "plt.imshow(img_)\n",
    "img__ = np.expand_dims(img_, axis=0)\n",
    "print(\"It's a {}\".format(np.argmax(resnet50_.predict(img__/255.))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's a 4\n"
     ]
    },
    {
     "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 = np.array(plt.imread(\"two_fingers.jpg\"))\n",
    "img_ = np.array(Image.fromarray(img).resize((64,64)))\n",
    "plt.imshow(img_)\n",
    "img__ = np.expand_dims(img_, axis=0)\n",
    "print(\"It's a {}\".format(np.argmax(resnet50_.predict(img__/255.))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It's a 5\n"
     ]
    },
    {
     "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 = np.array(plt.imread(\"three_fingers.jpg\"))\n",
    "img_ = np.array(Image.fromarray(img).resize((64,64)))\n",
    "plt.imshow(img_)\n",
    "img__ = np.expand_dims(img_, axis=0)\n",
    "print(\"It's a {}\".format(np.argmax(resnet50_.predict(img__/255.))))"
   ]
  }
 ],
 "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": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}