Collab 6 - Aprendizaje por Transferencia y puesta a punto de parámetros .ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://colab.research.google.com/github/joanby/tensorflow2/blob/master/Collab%206%20-%20Aprendizaje%20por%20Transferencia%20y%20puesta%20a%20punto%20de%20parámetros%20.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "19QTXXYFazRX"
   },
   "source": [
    "![alt text](https://live.staticflickr.com/4544/38228876666_3782386ca7_b.jpg)\n",
    "\n",
    "## Paso 1: Instalación de dependencias y configuración del entorno en GPU"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 579
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 45880,
     "status": "ok",
     "timestamp": 1558427766228,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "KmIh3NZAaEkC",
    "outputId": "259341e9-ac1b-473f-b55b-561eaa5c06e5"
   },
   "outputs": [],
   "source": [
    "#!pip install tensorflow-gpu==2.0.0.alpha0\n",
    "%tensorflow_version 2.x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 36
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 3898,
     "status": "ok",
     "timestamp": 1558427797046,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "iu2mbI2Ha8UW",
    "outputId": "384d426e-9b1c-470e-fd2e-4088988b50fc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Requirement already satisfied: tqdm in /usr/local/lib/python3.6/dist-packages (4.28.1)\n"
     ]
    }
   ],
   "source": [
    "!pip install tqdm"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "jQl1fF9Ca80N"
   },
   "source": [
    "### Descargar el dataset de Perros vs Gatos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 204
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 5647,
     "status": "ok",
     "timestamp": 1564667721078,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "mLWquLOxa-V5",
    "outputId": "213c361d-c6af-4710-ad24-8eb78011f60d"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "--2019-08-01 13:55:17--  https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip\n",
      "Resolving storage.googleapis.com (storage.googleapis.com)... 173.194.79.128, 2a00:1450:4013:c00::80\n",
      "Connecting to storage.googleapis.com (storage.googleapis.com)|173.194.79.128|:443... connected.\n",
      "HTTP request sent, awaiting response... 200 OK\n",
      "Length: 68606236 (65M) [application/zip]\n",
      "Saving to: ‘./cats_and_dogs_filtered.zip’\n",
      "\n",
      "./cats_and_dogs_fil 100%[===================>]  65.43M  34.4MB/s    in 1.9s    \n",
      "\n",
      "2019-08-01 13:55:19 (34.4 MB/s) - ‘./cats_and_dogs_filtered.zip’ saved [68606236/68606236]\n",
      "\n"
     ]
    }
   ],
   "source": [
    "!wget --no-check-certificate \\\n",
    "    https://storage.googleapis.com/mledu-datasets/cats_and_dogs_filtered.zip \\\n",
    "    -O ./cats_and_dogs_filtered.zip"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "5-tk5DfRjXN3"
   },
   "source": [
    "## Paso 2: Pre procesado del data set"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "LpbQm2i9jYrP"
   },
   "source": [
    "### Importar las dependencias del proyecto"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 3353,
     "status": "ok",
     "timestamp": 1564668175537,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "9fOTP2XZimCM",
    "outputId": "ad410d69-13b3-449b-8686-b573640c5f66"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.14.0'"
      ]
     },
     "execution_count": 2,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import zipfile\n",
    "import numpy as np\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "from tqdm import tqdm_notebook\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "\n",
    "%matplotlib inline\n",
    "tf.__version__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "MlsIOpGljdlS"
   },
   "source": [
    "### Descomprimir el dataset de Perros vs Gatos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "XsqDOC8WjiJ6"
   },
   "outputs": [],
   "source": [
    "dataset_path = \"./cats_and_dogs_filtered.zip\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jm5kaTO6jjHA"
   },
   "outputs": [],
   "source": [
    "zip_object = zipfile.ZipFile(file=dataset_path, mode=\"r\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "OI_eKR5-lhEy"
   },
   "outputs": [],
   "source": [
    "zip_object.extractall(\"./\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "lj0UvhtwlhNx"
   },
   "outputs": [],
   "source": [
    "zip_object.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "VSDt-MEVjlsA"
   },
   "source": [
    "### Configurar las rutas al dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "glNaVdNmjmfr"
   },
   "outputs": [],
   "source": [
    "dataset_path_new = \"./cats_and_dogs_filtered/\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "VhfMF891jt0P"
   },
   "outputs": [],
   "source": [
    "train_dir = os.path.join(dataset_path_new, \"train\")\n",
    "validation_dir = os.path.join(dataset_path_new, \"validation\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "-YLQb_WOmjwP"
   },
   "source": [
    "## Construir el Modelo"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "PoMFwC51mmNa"
   },
   "source": [
    "### Cargar un modelo pre entrenado (MobileNetV2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "owkKMJiFmZhj"
   },
   "outputs": [],
   "source": [
    "IMG_SHAPE = (128, 128, 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 139
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 13444,
     "status": "ok",
     "timestamp": 1564668821238,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "pEemsnZinInr",
    "outputId": "17182bb1-4d09-4344-b424-7295d40dd1fe"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING: Logging before flag parsing goes to stderr.\n",
      "W0801 14:13:28.105518 139643395622784 deprecation.py:506] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Downloading data from https://github.com/JonathanCMitchell/mobilenet_v2_keras/releases/download/v1.1/mobilenet_v2_weights_tf_dim_ordering_tf_kernels_1.0_128_no_top.h5\n",
      "9412608/9406464 [==============================] - 1s 0us/step\n"
     ]
    }
   ],
   "source": [
    "base_model = tf.keras.applications.MobileNetV2(input_shape=IMG_SHAPE, include_top=False, weights=\"imagenet\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1533,
     "status": "ok",
     "timestamp": 1564668827342,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "95RL7HQLoD8z",
    "outputId": "cf69d80a-a518-40f3-b621-a5009628e2b9"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"mobilenetv2_1.00_128\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 128, 128, 3) 0                                            \n",
      "__________________________________________________________________________________________________\n",
      "Conv1_pad (ZeroPadding2D)       (None, 129, 129, 3)  0           input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "Conv1 (Conv2D)                  (None, 64, 64, 32)   864         Conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "bn_Conv1 (BatchNormalization)   (None, 64, 64, 32)   128         Conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "Conv1_relu (ReLU)               (None, 64, 64, 32)   0           bn_Conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise (Depthw (None, 64, 64, 32)   288         Conv1_relu[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise_BN (Bat (None, 64, 64, 32)   128         expanded_conv_depthwise[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise_relu (R (None, 64, 64, 32)   0           expanded_conv_depthwise_BN[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_project (Conv2D)  (None, 64, 64, 16)   512         expanded_conv_depthwise_relu[0][0\n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_project_BN (Batch (None, 64, 64, 16)   64          expanded_conv_project[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand (Conv2D)         (None, 64, 64, 96)   1536        expanded_conv_project_BN[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand_BN (BatchNormali (None, 64, 64, 96)   384         block_1_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand_relu (ReLU)      (None, 64, 64, 96)   0           block_1_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_1_pad (ZeroPadding2D)     (None, 65, 65, 96)   0           block_1_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise (DepthwiseCon (None, 32, 32, 96)   864         block_1_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise_BN (BatchNorm (None, 32, 32, 96)   384         block_1_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise_relu (ReLU)   (None, 32, 32, 96)   0           block_1_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_1_project (Conv2D)        (None, 32, 32, 24)   2304        block_1_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_1_project_BN (BatchNormal (None, 32, 32, 24)   96          block_1_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand (Conv2D)         (None, 32, 32, 144)  3456        block_1_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_2_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_2_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise (DepthwiseCon (None, 32, 32, 144)  1296        block_2_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise_BN (BatchNorm (None, 32, 32, 144)  576         block_2_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise_relu (ReLU)   (None, 32, 32, 144)  0           block_2_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_2_project (Conv2D)        (None, 32, 32, 24)   3456        block_2_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_2_project_BN (BatchNormal (None, 32, 32, 24)   96          block_2_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_2_add (Add)               (None, 32, 32, 24)   0           block_1_project_BN[0][0]         \n",
      "                                                                 block_2_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand (Conv2D)         (None, 32, 32, 144)  3456        block_2_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_3_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_3_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_3_pad (ZeroPadding2D)     (None, 33, 33, 144)  0           block_3_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise (DepthwiseCon (None, 16, 16, 144)  1296        block_3_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise_BN (BatchNorm (None, 16, 16, 144)  576         block_3_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise_relu (ReLU)   (None, 16, 16, 144)  0           block_3_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_3_project (Conv2D)        (None, 16, 16, 32)   4608        block_3_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_3_project_BN (BatchNormal (None, 16, 16, 32)   128         block_3_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand (Conv2D)         (None, 16, 16, 192)  6144        block_3_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_4_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_4_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_4_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_4_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_4_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_4_project (Conv2D)        (None, 16, 16, 32)   6144        block_4_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_4_project_BN (BatchNormal (None, 16, 16, 32)   128         block_4_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_4_add (Add)               (None, 16, 16, 32)   0           block_3_project_BN[0][0]         \n",
      "                                                                 block_4_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand (Conv2D)         (None, 16, 16, 192)  6144        block_4_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_5_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_5_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_5_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_5_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_5_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_5_project (Conv2D)        (None, 16, 16, 32)   6144        block_5_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_5_project_BN (BatchNormal (None, 16, 16, 32)   128         block_5_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_5_add (Add)               (None, 16, 16, 32)   0           block_4_add[0][0]                \n",
      "                                                                 block_5_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand (Conv2D)         (None, 16, 16, 192)  6144        block_5_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_6_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_6_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_6_pad (ZeroPadding2D)     (None, 17, 17, 192)  0           block_6_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise (DepthwiseCon (None, 8, 8, 192)    1728        block_6_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise_BN (BatchNorm (None, 8, 8, 192)    768         block_6_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise_relu (ReLU)   (None, 8, 8, 192)    0           block_6_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_6_project (Conv2D)        (None, 8, 8, 64)     12288       block_6_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_6_project_BN (BatchNormal (None, 8, 8, 64)     256         block_6_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand (Conv2D)         (None, 8, 8, 384)    24576       block_6_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_7_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_7_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_7_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_7_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_7_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_7_project (Conv2D)        (None, 8, 8, 64)     24576       block_7_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_7_project_BN (BatchNormal (None, 8, 8, 64)     256         block_7_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_7_add (Add)               (None, 8, 8, 64)     0           block_6_project_BN[0][0]         \n",
      "                                                                 block_7_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand (Conv2D)         (None, 8, 8, 384)    24576       block_7_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_8_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_8_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_8_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_8_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_8_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_8_project (Conv2D)        (None, 8, 8, 64)     24576       block_8_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_8_project_BN (BatchNormal (None, 8, 8, 64)     256         block_8_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_8_add (Add)               (None, 8, 8, 64)     0           block_7_add[0][0]                \n",
      "                                                                 block_8_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand (Conv2D)         (None, 8, 8, 384)    24576       block_8_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_9_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_9_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_9_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_9_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_9_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_9_project (Conv2D)        (None, 8, 8, 64)     24576       block_9_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_9_project_BN (BatchNormal (None, 8, 8, 64)     256         block_9_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_9_add (Add)               (None, 8, 8, 64)     0           block_8_add[0][0]                \n",
      "                                                                 block_9_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand (Conv2D)        (None, 8, 8, 384)    24576       block_9_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand_BN (BatchNormal (None, 8, 8, 384)    1536        block_10_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand_relu (ReLU)     (None, 8, 8, 384)    0           block_10_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise (DepthwiseCo (None, 8, 8, 384)    3456        block_10_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise_BN (BatchNor (None, 8, 8, 384)    1536        block_10_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise_relu (ReLU)  (None, 8, 8, 384)    0           block_10_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_10_project (Conv2D)       (None, 8, 8, 96)     36864       block_10_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_10_project_BN (BatchNorma (None, 8, 8, 96)     384         block_10_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand (Conv2D)        (None, 8, 8, 576)    55296       block_10_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_11_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_11_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_11_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_11_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_11_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_11_project (Conv2D)       (None, 8, 8, 96)     55296       block_11_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_11_project_BN (BatchNorma (None, 8, 8, 96)     384         block_11_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_11_add (Add)              (None, 8, 8, 96)     0           block_10_project_BN[0][0]        \n",
      "                                                                 block_11_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand (Conv2D)        (None, 8, 8, 576)    55296       block_11_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_12_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_12_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_12_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_12_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_12_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_12_project (Conv2D)       (None, 8, 8, 96)     55296       block_12_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_12_project_BN (BatchNorma (None, 8, 8, 96)     384         block_12_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_12_add (Add)              (None, 8, 8, 96)     0           block_11_add[0][0]               \n",
      "                                                                 block_12_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand (Conv2D)        (None, 8, 8, 576)    55296       block_12_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_13_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_13_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_13_pad (ZeroPadding2D)    (None, 9, 9, 576)    0           block_13_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise (DepthwiseCo (None, 4, 4, 576)    5184        block_13_pad[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise_BN (BatchNor (None, 4, 4, 576)    2304        block_13_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise_relu (ReLU)  (None, 4, 4, 576)    0           block_13_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_13_project (Conv2D)       (None, 4, 4, 160)    92160       block_13_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_13_project_BN (BatchNorma (None, 4, 4, 160)    640         block_13_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand (Conv2D)        (None, 4, 4, 960)    153600      block_13_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_14_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_14_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_14_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_14_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_14_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_14_project (Conv2D)       (None, 4, 4, 160)    153600      block_14_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_14_project_BN (BatchNorma (None, 4, 4, 160)    640         block_14_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_14_add (Add)              (None, 4, 4, 160)    0           block_13_project_BN[0][0]        \n",
      "                                                                 block_14_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand (Conv2D)        (None, 4, 4, 960)    153600      block_14_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_15_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_15_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_15_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_15_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_15_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_15_project (Conv2D)       (None, 4, 4, 160)    153600      block_15_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_15_project_BN (BatchNorma (None, 4, 4, 160)    640         block_15_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_15_add (Add)              (None, 4, 4, 160)    0           block_14_add[0][0]               \n",
      "                                                                 block_15_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand (Conv2D)        (None, 4, 4, 960)    153600      block_15_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_16_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_16_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_16_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_16_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_16_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_16_project (Conv2D)       (None, 4, 4, 320)    307200      block_16_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_16_project_BN (BatchNorma (None, 4, 4, 320)    1280        block_16_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "Conv_1 (Conv2D)                 (None, 4, 4, 1280)   409600      block_16_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "Conv_1_bn (BatchNormalization)  (None, 4, 4, 1280)   5120        Conv_1[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "out_relu (ReLU)                 (None, 4, 4, 1280)   0           Conv_1_bn[0][0]                  \n",
      "==================================================================================================\n",
      "Total params: 2,257,984\n",
      "Trainable params: 2,223,872\n",
      "Non-trainable params: 34,112\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "base_model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "iRDmVCtFofpc"
   },
   "source": [
    "### Congelar el modelo base"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "kFlhPT4-oMxL"
   },
   "outputs": [],
   "source": [
    "base_model.trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "lZTcqkBipEEM"
   },
   "source": [
    "### Definir la cabecera personalizada para nuestra red neuronal"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 2493,
     "status": "ok",
     "timestamp": 1564669875898,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "7U82_j0oo3wT",
    "outputId": "4eba8396-f574-48b7-cd22-1b99b1d3aa4e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'out_relu/Relu6:0' shape=(?, 4, 4, 1280) dtype=float32>"
      ]
     },
     "execution_count": 13,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "base_model.output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vZlIXTCbpXEM"
   },
   "outputs": [],
   "source": [
    "global_average_layer = tf.keras.layers.GlobalAveragePooling2D()(base_model.output)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 36
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1221,
     "status": "ok",
     "timestamp": 1558433823416,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "al6yHSdEp2zL",
    "outputId": "57f9b2bc-78ce-4530-8a56-bcd5b482be6e"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<tf.Tensor 'global_average_pooling2d/Mean:0' shape=(None, 1280) dtype=float32>"
      ]
     },
     "execution_count": 15,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "global_average_layer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Q8K2WenOp4Ob"
   },
   "outputs": [],
   "source": [
    "prediction_layer = tf.keras.layers.Dense(units=1, activation='sigmoid')(global_average_layer)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "SwzYZTTwqT8l"
   },
   "source": [
    "### Definir el modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "qUHmyw4CqLTz"
   },
   "outputs": [],
   "source": [
    "model = tf.keras.models.Model(inputs=base_model.input, outputs=prediction_layer)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1980,
     "status": "ok",
     "timestamp": 1564669890500,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "E-V9XI7_rBTi",
    "outputId": "50780077-d4b1-43e5-e6f3-388f99c586a0"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 128, 128, 3) 0                                            \n",
      "__________________________________________________________________________________________________\n",
      "Conv1_pad (ZeroPadding2D)       (None, 129, 129, 3)  0           input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "Conv1 (Conv2D)                  (None, 64, 64, 32)   864         Conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "bn_Conv1 (BatchNormalization)   (None, 64, 64, 32)   128         Conv1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "Conv1_relu (ReLU)               (None, 64, 64, 32)   0           bn_Conv1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise (Depthw (None, 64, 64, 32)   288         Conv1_relu[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise_BN (Bat (None, 64, 64, 32)   128         expanded_conv_depthwise[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_depthwise_relu (R (None, 64, 64, 32)   0           expanded_conv_depthwise_BN[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_project (Conv2D)  (None, 64, 64, 16)   512         expanded_conv_depthwise_relu[0][0\n",
      "__________________________________________________________________________________________________\n",
      "expanded_conv_project_BN (Batch (None, 64, 64, 16)   64          expanded_conv_project[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand (Conv2D)         (None, 64, 64, 96)   1536        expanded_conv_project_BN[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand_BN (BatchNormali (None, 64, 64, 96)   384         block_1_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_1_expand_relu (ReLU)      (None, 64, 64, 96)   0           block_1_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_1_pad (ZeroPadding2D)     (None, 65, 65, 96)   0           block_1_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise (DepthwiseCon (None, 32, 32, 96)   864         block_1_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise_BN (BatchNorm (None, 32, 32, 96)   384         block_1_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_1_depthwise_relu (ReLU)   (None, 32, 32, 96)   0           block_1_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_1_project (Conv2D)        (None, 32, 32, 24)   2304        block_1_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_1_project_BN (BatchNormal (None, 32, 32, 24)   96          block_1_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand (Conv2D)         (None, 32, 32, 144)  3456        block_1_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_2_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_2_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_2_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise (DepthwiseCon (None, 32, 32, 144)  1296        block_2_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise_BN (BatchNorm (None, 32, 32, 144)  576         block_2_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_2_depthwise_relu (ReLU)   (None, 32, 32, 144)  0           block_2_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_2_project (Conv2D)        (None, 32, 32, 24)   3456        block_2_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_2_project_BN (BatchNormal (None, 32, 32, 24)   96          block_2_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_2_add (Add)               (None, 32, 32, 24)   0           block_1_project_BN[0][0]         \n",
      "                                                                 block_2_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand (Conv2D)         (None, 32, 32, 144)  3456        block_2_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_3_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_3_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_3_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_3_pad (ZeroPadding2D)     (None, 33, 33, 144)  0           block_3_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise (DepthwiseCon (None, 16, 16, 144)  1296        block_3_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise_BN (BatchNorm (None, 16, 16, 144)  576         block_3_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_3_depthwise_relu (ReLU)   (None, 16, 16, 144)  0           block_3_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_3_project (Conv2D)        (None, 16, 16, 32)   4608        block_3_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_3_project_BN (BatchNormal (None, 16, 16, 32)   128         block_3_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand (Conv2D)         (None, 16, 16, 192)  6144        block_3_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_4_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_4_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_4_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_4_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_4_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_4_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_4_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_4_project (Conv2D)        (None, 16, 16, 32)   6144        block_4_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_4_project_BN (BatchNormal (None, 16, 16, 32)   128         block_4_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_4_add (Add)               (None, 16, 16, 32)   0           block_3_project_BN[0][0]         \n",
      "                                                                 block_4_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand (Conv2D)         (None, 16, 16, 192)  6144        block_4_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_5_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_5_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_5_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_5_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_5_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_5_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_5_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_5_project (Conv2D)        (None, 16, 16, 32)   6144        block_5_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_5_project_BN (BatchNormal (None, 16, 16, 32)   128         block_5_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_5_add (Add)               (None, 16, 16, 32)   0           block_4_add[0][0]                \n",
      "                                                                 block_5_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand (Conv2D)         (None, 16, 16, 192)  6144        block_5_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_6_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_6_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_6_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_6_pad (ZeroPadding2D)     (None, 17, 17, 192)  0           block_6_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise (DepthwiseCon (None, 8, 8, 192)    1728        block_6_pad[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise_BN (BatchNorm (None, 8, 8, 192)    768         block_6_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_6_depthwise_relu (ReLU)   (None, 8, 8, 192)    0           block_6_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_6_project (Conv2D)        (None, 8, 8, 64)     12288       block_6_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_6_project_BN (BatchNormal (None, 8, 8, 64)     256         block_6_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand (Conv2D)         (None, 8, 8, 384)    24576       block_6_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_7_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_7_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_7_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_7_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_7_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_7_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_7_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_7_project (Conv2D)        (None, 8, 8, 64)     24576       block_7_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_7_project_BN (BatchNormal (None, 8, 8, 64)     256         block_7_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_7_add (Add)               (None, 8, 8, 64)     0           block_6_project_BN[0][0]         \n",
      "                                                                 block_7_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand (Conv2D)         (None, 8, 8, 384)    24576       block_7_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_8_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_8_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_8_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_8_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_8_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_8_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_8_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_8_project (Conv2D)        (None, 8, 8, 64)     24576       block_8_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_8_project_BN (BatchNormal (None, 8, 8, 64)     256         block_8_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_8_add (Add)               (None, 8, 8, 64)     0           block_7_add[0][0]                \n",
      "                                                                 block_8_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand (Conv2D)         (None, 8, 8, 384)    24576       block_8_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_9_expand[0][0]             \n",
      "__________________________________________________________________________________________________\n",
      "block_9_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_9_expand_BN[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_9_expand_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_9_depthwise[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "block_9_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_9_depthwise_BN[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_9_project (Conv2D)        (None, 8, 8, 64)     24576       block_9_depthwise_relu[0][0]     \n",
      "__________________________________________________________________________________________________\n",
      "block_9_project_BN (BatchNormal (None, 8, 8, 64)     256         block_9_project[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_9_add (Add)               (None, 8, 8, 64)     0           block_8_add[0][0]                \n",
      "                                                                 block_9_project_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand (Conv2D)        (None, 8, 8, 384)    24576       block_9_add[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand_BN (BatchNormal (None, 8, 8, 384)    1536        block_10_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_10_expand_relu (ReLU)     (None, 8, 8, 384)    0           block_10_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise (DepthwiseCo (None, 8, 8, 384)    3456        block_10_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise_BN (BatchNor (None, 8, 8, 384)    1536        block_10_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_10_depthwise_relu (ReLU)  (None, 8, 8, 384)    0           block_10_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_10_project (Conv2D)       (None, 8, 8, 96)     36864       block_10_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_10_project_BN (BatchNorma (None, 8, 8, 96)     384         block_10_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand (Conv2D)        (None, 8, 8, 576)    55296       block_10_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_11_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_11_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_11_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_11_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_11_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_11_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_11_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_11_project (Conv2D)       (None, 8, 8, 96)     55296       block_11_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_11_project_BN (BatchNorma (None, 8, 8, 96)     384         block_11_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_11_add (Add)              (None, 8, 8, 96)     0           block_10_project_BN[0][0]        \n",
      "                                                                 block_11_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand (Conv2D)        (None, 8, 8, 576)    55296       block_11_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_12_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_12_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_12_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_12_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_12_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_12_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_12_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_12_project (Conv2D)       (None, 8, 8, 96)     55296       block_12_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_12_project_BN (BatchNorma (None, 8, 8, 96)     384         block_12_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_12_add (Add)              (None, 8, 8, 96)     0           block_11_add[0][0]               \n",
      "                                                                 block_12_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand (Conv2D)        (None, 8, 8, 576)    55296       block_12_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_13_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_13_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_13_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_13_pad (ZeroPadding2D)    (None, 9, 9, 576)    0           block_13_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise (DepthwiseCo (None, 4, 4, 576)    5184        block_13_pad[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise_BN (BatchNor (None, 4, 4, 576)    2304        block_13_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_13_depthwise_relu (ReLU)  (None, 4, 4, 576)    0           block_13_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_13_project (Conv2D)       (None, 4, 4, 160)    92160       block_13_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_13_project_BN (BatchNorma (None, 4, 4, 160)    640         block_13_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand (Conv2D)        (None, 4, 4, 960)    153600      block_13_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_14_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_14_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_14_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_14_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_14_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_14_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_14_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_14_project (Conv2D)       (None, 4, 4, 160)    153600      block_14_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_14_project_BN (BatchNorma (None, 4, 4, 160)    640         block_14_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_14_add (Add)              (None, 4, 4, 160)    0           block_13_project_BN[0][0]        \n",
      "                                                                 block_14_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand (Conv2D)        (None, 4, 4, 960)    153600      block_14_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_15_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_15_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_15_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_15_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_15_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_15_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_15_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_15_project (Conv2D)       (None, 4, 4, 160)    153600      block_15_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_15_project_BN (BatchNorma (None, 4, 4, 160)    640         block_15_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block_15_add (Add)              (None, 4, 4, 160)    0           block_14_add[0][0]               \n",
      "                                                                 block_15_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand (Conv2D)        (None, 4, 4, 960)    153600      block_15_add[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_16_expand[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block_16_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_16_expand_BN[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_16_expand_relu[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_16_depthwise[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block_16_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_16_depthwise_BN[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block_16_project (Conv2D)       (None, 4, 4, 320)    307200      block_16_depthwise_relu[0][0]    \n",
      "__________________________________________________________________________________________________\n",
      "block_16_project_BN (BatchNorma (None, 4, 4, 320)    1280        block_16_project[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "Conv_1 (Conv2D)                 (None, 4, 4, 1280)   409600      block_16_project_BN[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "Conv_1_bn (BatchNormalization)  (None, 4, 4, 1280)   5120        Conv_1[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "out_relu (ReLU)                 (None, 4, 4, 1280)   0           Conv_1_bn[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d (Globa (None, 1280)         0           out_relu[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 1)            1281        global_average_pooling2d[0][0]   \n",
      "==================================================================================================\n",
      "Total params: 2,259,265\n",
      "Trainable params: 1,281\n",
      "Non-trainable params: 2,257,984\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "qBrj4jkqrVfy"
   },
   "source": [
    "### Compilar el modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 88
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1569,
     "status": "ok",
     "timestamp": 1564670316266,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "Qt1ZOPKlsO87",
    "outputId": "33c4983f-bcc3-41b6-f249-72a92854ef1b"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "W0801 14:38:35.068260 139643395622784 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/nn_impl.py:180: add_dispatch_support.<locals>.wrapper (from tensorflow.python.ops.array_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.where in 2.0, which has the same broadcast rule as np.where\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer=tf.keras.optimizers.RMSprop(lr=0.0001), loss=\"binary_crossentropy\", metrics=[\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "YJRlLGaotJ_J"
   },
   "source": [
    "### Crear generadores de datos\n",
    "\n",
    "Redimensionar imágenes\n",
    "\n",
    "    Las grandes arquitecturas pre-entrenadas solamente soportan cierto tipo de tamaños de imágenes.\n",
    "    \n",
    "Por ejemplo: MobileNet (la arquitectura que nosotros usamos) soporta: (96, 96), (128, 128), (160, 160), (192, 192), (224, 224)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "7QV5h3-OtC97"
   },
   "outputs": [],
   "source": [
    "data_gen_train = ImageDataGenerator(rescale=1/255.)\n",
    "data_gen_valid = ImageDataGenerator(rescale=1/255.)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1833,
     "status": "ok",
     "timestamp": 1564670793400,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "7Ms7kEkAutJb",
    "outputId": "0bc118b0-0585-4c9c-c259-5754267a6191"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "train_generator = data_gen_train.flow_from_directory(train_dir, target_size=(128,128), batch_size=128, class_mode=\"binary\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 892,
     "status": "ok",
     "timestamp": 1564670804035,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "WbJL_jtwvgGk",
    "outputId": "f0ed7ef6-f295-488a-bdbc-209c13cb754c"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1000 images belonging to 2 classes.\n"
     ]
    }
   ],
   "source": [
    "valid_generator = data_gen_valid.flow_from_directory(validation_dir, target_size=(128,128), batch_size=128, class_mode=\"binary\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "eSzbQEcpv4C-"
   },
   "source": [
    "### Entrenar el modelo "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 306
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 57303,
     "status": "ok",
     "timestamp": 1564671171572,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "O6jUOnNBvuYD",
    "outputId": "c0742a56-7355-41fe-9885-f37d7cd7ab29"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/8\n",
      "16/16 [==============================] - 9s 560ms/step - loss: 0.4902 - acc: 0.7640 - val_loss: 0.4279 - val_acc: 0.8090\n",
      "Epoch 2/8\n",
      "16/16 [==============================] - 7s 412ms/step - loss: 0.4526 - acc: 0.7955 - val_loss: 0.4030 - val_acc: 0.8210\n",
      "Epoch 3/8\n",
      "16/16 [==============================] - 6s 405ms/step - loss: 0.4297 - acc: 0.8100 - val_loss: 0.3745 - val_acc: 0.8320\n",
      "Epoch 4/8\n",
      "16/16 [==============================] - 7s 412ms/step - loss: 0.4036 - acc: 0.8270 - val_loss: 0.3584 - val_acc: 0.8410\n",
      "Epoch 5/8\n",
      "16/16 [==============================] - 7s 415ms/step - loss: 0.3779 - acc: 0.8455 - val_loss: 0.3405 - val_acc: 0.8530\n",
      "Epoch 6/8\n",
      "16/16 [==============================] - 7s 419ms/step - loss: 0.3560 - acc: 0.8590 - val_loss: 0.3235 - val_acc: 0.8650\n",
      "Epoch 7/8\n",
      "16/16 [==============================] - 7s 407ms/step - loss: 0.3439 - acc: 0.8690 - val_loss: 0.3107 - val_acc: 0.8710\n",
      "Epoch 8/8\n",
      "16/16 [==============================] - 7s 407ms/step - loss: 0.3278 - acc: 0.8715 - val_loss: 0.2936 - val_acc: 0.8810\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f00e61d1e10>"
      ]
     },
     "execution_count": 23,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit_generator(train_generator, epochs=8, validation_data=valid_generator)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "t_Dx30HVxWr2"
   },
   "source": [
    "### Evaluar el modelo de aprendizaje por transferencia"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "Bsc0GzhCxNrz"
   },
   "outputs": [],
   "source": [
    "valid_loss, valid_accuracy = model.evaluate_generator(valid_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 1337,
     "status": "ok",
     "timestamp": 1564671308577,
     "user": {
      "displayName": "Juan Gabriel Gomila Salas",
      "photoUrl": "https://lh4.googleusercontent.com/-RJ38mzk8Fog/AAAAAAAAAAI/AAAAAAAAA_g/VV8PcNMAUrk/s64/photo.jpg",
      "userId": "03106941341701838274"
     },
     "user_tz": -120
    },
    "id": "6XOGP0tExZvv",
    "outputId": "abd7ca7e-d353-4826-b977-5e1813ddc34f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Accuracy after transfer learning: 0.8809999823570251\n"
     ]
    }
   ],
   "source": [
    "print(\"Accuracy after transfer learning: {}\".format(valid_accuracy))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "3mmrlX05yhkU"
   },
   "source": [
    "## Puesta a punto de parámetros\n",
    "\n",
    "\n",
    "Un par de cosas:\n",
    "\n",
    "- NUNCA HAY QUE USAR la puesta a punto (fine tuning) de parámetros en toda la red neuronal: con algunas de las capas superiores (las finales) es más que suficiente suficiente. En la mayoría de casos, son las más especializadas. El objetivo del fine tuning es adaptar esa parte específica de la red neuronal para nuestro nuevo dataset específico.\n",
    "- Empezar con la puesta a punto DESPUÉS de haber finalizado la fase de aprendizaje por transferencia. Si intentamos hacer el Fine tuning inmediatamente, los gradientes serán muy diferentes entre nuestra cabecera personalizada de la red neuronal y las nuevas capas no congeladas del modelo base."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "Vg-Na6Wmypx5"
   },
   "source": [
    "### Descongelar unas cuantas capas superiores del modelo"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "p9CHIFobyTVe"
   },
   "outputs": [],
   "source": [
    "base_model.trainable = True"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 36
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 525,
     "status": "ok",
     "timestamp": 1558433978710,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "j_PpMM9xyuUW",
    "outputId": "5c55b9f0-232b-4467-9800-8461dc50cb42"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Number of layersin the base model: 155\n"
     ]
    }
   ],
   "source": [
    "print(\"Number of layersin the base model: {}\".format(len(base_model.layers)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "kgijBsNS04AN"
   },
   "outputs": [],
   "source": [
    "fine_tune_at = 100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "j8hnc35D1EOc"
   },
   "outputs": [],
   "source": [
    "for layer in base_model.layers[:fine_tune_at]:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "TIjXrxRa1alF"
   },
   "source": [
    "### Compilar el modelo para la puesta a punto"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "sOM615rt1POD"
   },
   "outputs": [],
   "source": [
    "model.compile(optimizer=tf.keras.optimizers.RMSprop(lr=0.0001),\n",
    "              loss='binary_crossentropy',\n",
    "              metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "CD57phTH2QAI"
   },
   "source": [
    "### Puesta a punto"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 223
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 54761,
     "status": "ok",
     "timestamp": 1558434128850,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "kPRnLqEu2G5z",
    "outputId": "4595cbfd-7b6e-4774-c37f-39c6836d7f8b"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/5\n",
      "16/16 [==============================] - 26s 2s/step - loss: 4.4638e-04 - accuracy: 1.0000 - val_loss: 0.4165 - val_accuracy: 0.9390\n",
      "Epoch 2/5\n",
      "16/16 [==============================] - 7s 454ms/step - loss: 1.3087e-04 - accuracy: 1.0000 - val_loss: 0.5169 - val_accuracy: 0.9290\n",
      "Epoch 3/5\n",
      "16/16 [==============================] - 7s 428ms/step - loss: 6.1403e-05 - accuracy: 1.0000 - val_loss: 0.4853 - val_accuracy: 0.9370\n",
      "Epoch 4/5\n",
      "16/16 [==============================] - 7s 449ms/step - loss: 2.1485e-05 - accuracy: 1.0000 - val_loss: 0.5115 - val_accuracy: 0.9340\n",
      "Epoch 5/5\n",
      "16/16 [==============================] - 7s 440ms/step - loss: 8.7441e-06 - accuracy: 1.0000 - val_loss: 0.5752 - val_accuracy: 0.9340\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7f4ce90229b0>"
      ]
     },
     "execution_count": 34,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model.fit_generator(train_generator,  \n",
    "                    epochs=5, \n",
    "                    validation_data=valid_generator)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "rutvwJBp6s2x"
   },
   "source": [
    "### Evaluar el modelo re calibrado"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "FPTyisdx33u3"
   },
   "outputs": [],
   "source": [
    "valid_loss, valid_accuracy = model.evaluate_generator(valid_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 36
    },
    "colab_type": "code",
    "executionInfo": {
     "elapsed": 735,
     "status": "ok",
     "timestamp": 1558434407744,
     "user": {
      "displayName": "TechGuy Luka",
      "photoUrl": "https://lh3.googleusercontent.com/-v_8ixmE_SJc/AAAAAAAAAAI/AAAAAAAAAEY/QGGmNZURhdc/s64/photo.jpg",
      "userId": "12375685325186592450"
     },
     "user_tz": -120
    },
    "id": "jijEA5n26yQu",
    "outputId": "48734072-4fce-4947-e00f-280b2510a7cf"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Validation accuracy after fine tuning: 0.9340000152587891\n"
     ]
    }
   ],
   "source": [
    "print(\"Validation accuracy after fine tuning: {}\".format(valid_accuracy))"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Collab 6 - Aprendizaje por Transferencia y puesta a punto de parámetros .ipynb",
   "provenance": [
    {
     "file_id": "1xrPfD-JzssQIwg_ofgq0aVbGfoOuac0z",
     "timestamp": 1564666215162
    }
   ]
  },
  "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}