MainNotebook.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "MainNotebook",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": [],
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/syslogg/cervix-predict/blob/master/MainNotebook.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Gr2ANWAsD8Ix",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "692eae84-64b0-4fee-ee59-8cbbe92cbc07"
      },
      "source": [
        "\"\"\"\n",
        "Created on Aug\n",
        "\n",
        "@author: Hendy Rodrigues F.Silva\n",
        "\"\"\"\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "from keras.utils import to_categorical\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import os\n",
        "\n",
        "#Data Augument\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from PIL import Image\n",
        "from imutils import paths\n",
        "import cv2\n",
        "\n",
        "from sklearn.preprocessing import LabelBinarizer\n",
        "from sklearn.model_selection import train_test_split\n",
        "\n",
        "#Metrics\n",
        "from sklearn.metrics import classification_report\n",
        "\n",
        "#Cross Validator\n",
        "from sklearn.model_selection import KFold\n",
        "\n",
        "from keras.models import Sequential\n",
        "#import SpatialPyramidPooling\n",
        "\n",
        "# Layers\n",
        "from keras.layers import Dense\n",
        "from keras.layers import Flatten\n",
        "from keras.layers import Conv2D\n",
        "from keras.layers import MaxPool2D, Activation"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vZrvDdIsIfqj",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "9a9cbdb5-4d2a-43e7-e969-ba105f497a04"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "S6etf__hEJlY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "BATCH_SIZE_HERLEV = 917\n",
        "\n",
        "DIRECTORY_HERLEV = r'drive/My Drive/Colab Notebooks/tcc/dataset/herlev'\n",
        "\n",
        "LABELS_HERLEV = set([\n",
        "        'carcinoma_in_situ',\n",
        "        'light_dysplastic',\n",
        "        'moderate_dysplastic', # drive/'My Drive'/'Colab Notebooks'/tcc/\n",
        "        'normal_columnar',\n",
        "        'normal_intermediate',\n",
        "        'normal_superficiel',\n",
        "        'severe_dysplastic'])"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9d8oqTYBFPwZ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def load_images(path, labelsPath=[], target_resizer=(250, 250)):\n",
        "    imagePaths = list(paths.list_images(path))\n",
        "    data = []\n",
        "    labels = []\n",
        "\n",
        "    for imgPath in imagePaths:\n",
        "        label = imgPath.split(os.path.sep)[-2]\n",
        "\n",
        "        if label not in labelsPath:\n",
        "            continue\n",
        "\n",
        "        image = cv2.imread(imgPath)\n",
        "        image = cv2.resize(image,target_resizer)\n",
        "\n",
        "        data.append(image)\n",
        "        labels.append(label)\n",
        "\n",
        "    return np.array(data), labels"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6bLr41KAIV6u",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "\n",
        "X, Y_without_encoded = load_images(DIRECTORY_HERLEV, LABELS_HERLEV)\n",
        "X = X / 255\n"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Sz63WmgxIZBS",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Codificação One Hot\n",
        "lb = LabelBinarizer()\n",
        "Y = lb.fit_transform(Y_without_encoded)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Ja3GiFdzIauO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# Divisão de DataSets de Treino e Teste\n",
        "(X_train, X_test, Y_train, Y_test) = train_test_split(X, Y, test_size=0.25, stratify=Y, random_state=1)"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "P4YWwjl6PZxs",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "# -=== Trabalhar com Data Augmentation ===-\n",
        "dataGenerator = ImageDataGenerator(rotation_range=20, zoom_range=0.15,\n",
        "                        \twidth_shift_range=0.2, height_shift_range=0.2, shear_range=0.15,\n",
        "                        \thorizontal_flip=True, fill_mode=\"nearest\")"
      ],
      "execution_count": 0,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "uRY7XSYTPbAH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 273
        },
        "outputId": "704f86ad-e883-45c0-e64e-27f0812fcb0c"
      },
      "source": [
        "# 2. Montagem da Rede Neural Convolucional\n",
        "\n",
        "class NeuralNetwork:\n",
        "    @staticmethod\n",
        "    def build(num_filter, num_class):\n",
        "        cnn = Sequential()\n",
        "\n",
        "        # Passo 1: Criar Camada Convolucional\n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(7, 7), activation='relu',input_shape=(250, 250, 3)))\n",
        "\n",
        "        # Passo 2: Efetuar o MaxPooling\n",
        "        cnn.add(MaxPool2D())\n",
        "        \n",
        "        num_filter = num_filter * 2\n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(5, 5), activation='relu'))\n",
        "        cnn.add(MaxPool2D())\n",
        "        \n",
        "        num_filter = num_filter * 2\n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(3, 3), activation='relu'))\n",
        "        cnn.add(MaxPool2D())\n",
        "        \n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(3, 3), activation='relu'))\n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(3, 3), activation='relu'))\n",
        "        cnn.add(Conv2D(num_filter, kernel_size=(3, 3), activation='relu'))\n",
        "        cnn.add(MaxPool2D())\n",
        "        \n",
        "        # Passo 3: Flatten as saidas da camada convolucional\n",
        "        cnn.add(Flatten())\n",
        "\n",
        "        # Passo 4: Exibir saida com uma camada totalmente conectada (Fully-Connected)\n",
        "        cnn.add(Dense(num_class))\n",
        "        cnn.add(Activation('softmax'))\n",
        "        #cnn.add(SpatialPyramidPooling([1, 2, 4])) # Pesquisar sobre\n",
        "        \n",
        "        #https://machinelearningmastery.com/how-to-reduce-overfitting-with-dropout-regularization-in-keras/\n",
        "\n",
        "        return cnn\n",
        "\n",
        "model = NeuralNetwork.build(8,len(lb.classes_))\n",
        "model.compile(\n",
        "            loss='categorical_crossentropy', # Verificar cada parametro.\n",
        "            optimizer='adam',\n",
        "            metrics=['accuracy']\n",
        ")"
      ],
      "execution_count": 9,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING: Logging before flag parsing goes to stderr.\n",
            "W0902 23:37:16.073819 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:66: The name tf.get_default_graph is deprecated. Please use tf.compat.v1.get_default_graph instead.\n",
            "\n",
            "W0902 23:37:16.092803 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:541: The name tf.placeholder is deprecated. Please use tf.compat.v1.placeholder instead.\n",
            "\n",
            "W0902 23:37:16.096306 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4432: The name tf.random_uniform is deprecated. Please use tf.random.uniform instead.\n",
            "\n",
            "W0902 23:37:16.114850 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:4267: The name tf.nn.max_pool is deprecated. Please use tf.nn.max_pool2d instead.\n",
            "\n",
            "W0902 23:37:16.197931 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/optimizers.py:793: The name tf.train.Optimizer is deprecated. Please use tf.compat.v1.train.Optimizer instead.\n",
            "\n",
            "W0902 23:37:16.218064 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3576: The name tf.log is deprecated. Please use tf.math.log instead.\n",
            "\n"
          ],
          "name": "stderr"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "HHwRQDTGPcX6",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "4df9fa29-d8a3-4df4-f21a-293bfb1f8f83"
      },
      "source": [
        "# 3. Treinamento da Rede Neural\n",
        "\n",
        "STEPS_PER_EPOCH = len(X_train)\n",
        "EPOCHS = 30\n",
        "BATCH_SIZE=32\n",
        "\n",
        "result = model.fit_generator(dataGenerator.flow(X_train,Y_train, batch_size=BATCH_SIZE),\n",
        "                             validation_data=(X_test, Y_test), steps_per_epoch=STEPS_PER_EPOCH, epochs=EPOCHS)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "W0902 23:37:16.677542 139726810085248 deprecation.py:323] From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_grad.py:1250: 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",
            "W0902 23:37:16.754276 139726810085248 deprecation_wrapper.py:119] From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:1033: The name tf.assign_add is deprecated. Please use tf.compat.v1.assign_add instead.\n",
            "\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/30\n",
            "687/687 [==============================] - 256s 373ms/step - loss: 1.6876 - acc: 0.3116 - val_loss: 1.4294 - val_acc: 0.3783\n",
            "Epoch 2/30\n",
            "687/687 [==============================] - 251s 365ms/step - loss: 1.3431 - acc: 0.4580 - val_loss: 1.3055 - val_acc: 0.4435\n",
            "Epoch 3/30\n",
            "687/687 [==============================] - 251s 365ms/step - loss: 1.0966 - acc: 0.5536 - val_loss: 1.1633 - val_acc: 0.5348\n",
            "Epoch 4/30\n",
            "687/687 [==============================] - 252s 366ms/step - loss: 0.9810 - acc: 0.5985 - val_loss: 1.1916 - val_acc: 0.5174\n",
            "Epoch 5/30\n",
            "687/687 [==============================] - 251s 365ms/step - loss: 0.8860 - acc: 0.6375 - val_loss: 1.2663 - val_acc: 0.5000\n",
            "Epoch 6/30\n",
            "687/687 [==============================] - 249s 362ms/step - loss: 0.7924 - acc: 0.6809 - val_loss: 1.1941 - val_acc: 0.5391\n",
            "Epoch 7/30\n",
            "687/687 [==============================] - 248s 361ms/step - loss: 0.7344 - acc: 0.7098 - val_loss: 1.4739 - val_acc: 0.5478\n",
            "Epoch 8/30\n",
            "687/687 [==============================] - 248s 360ms/step - loss: 0.6681 - acc: 0.7356 - val_loss: 1.2741 - val_acc: 0.5522\n",
            "Epoch 9/30\n",
            "687/687 [==============================] - 246s 359ms/step - loss: 0.6200 - acc: 0.7588 - val_loss: 1.4242 - val_acc: 0.5609\n",
            "Epoch 10/30\n",
            "687/687 [==============================] - 247s 359ms/step - loss: 0.5780 - acc: 0.7764 - val_loss: 1.4734 - val_acc: 0.5348\n",
            "Epoch 11/30\n",
            "687/687 [==============================] - 247s 359ms/step - loss: 0.5346 - acc: 0.7940 - val_loss: 1.4356 - val_acc: 0.5304\n",
            "Epoch 12/30\n",
            "687/687 [==============================] - 246s 358ms/step - loss: 0.4992 - acc: 0.8080 - val_loss: 1.6481 - val_acc: 0.5609\n",
            "Epoch 13/30\n",
            "687/687 [==============================] - 246s 358ms/step - loss: 0.4587 - acc: 0.8211 - val_loss: 1.7651 - val_acc: 0.5435\n",
            "Epoch 14/30\n",
            "687/687 [==============================] - 246s 358ms/step - loss: 0.4528 - acc: 0.8306 - val_loss: 1.8580 - val_acc: 0.5435\n",
            "Epoch 15/30\n",
            "687/687 [==============================] - 244s 355ms/step - loss: 0.4316 - acc: 0.8380 - val_loss: 1.6717 - val_acc: 0.5348\n",
            "Epoch 16/30\n",
            "687/687 [==============================] - 244s 355ms/step - loss: 0.3965 - acc: 0.8482 - val_loss: 1.9305 - val_acc: 0.5348\n",
            "Epoch 17/30\n",
            "687/687 [==============================] - 246s 358ms/step - loss: 0.3795 - acc: 0.8602 - val_loss: 1.9635 - val_acc: 0.5565\n",
            "Epoch 18/30\n",
            "687/687 [==============================] - 249s 362ms/step - loss: 0.3629 - acc: 0.8605 - val_loss: 2.0359 - val_acc: 0.5043\n",
            "Epoch 19/30\n",
            "687/687 [==============================] - 246s 359ms/step - loss: 0.3409 - acc: 0.8740 - val_loss: 2.3329 - val_acc: 0.4870\n",
            "Epoch 20/30\n",
            "687/687 [==============================] - 245s 357ms/step - loss: 0.3334 - acc: 0.8768 - val_loss: 2.2134 - val_acc: 0.5130\n",
            "Epoch 21/30\n",
            "687/687 [==============================] - 246s 359ms/step - loss: 0.3223 - acc: 0.8812 - val_loss: 2.1326 - val_acc: 0.5087\n",
            "Epoch 22/30\n",
            "687/687 [==============================] - 255s 371ms/step - loss: 0.2990 - acc: 0.8918 - val_loss: 2.4762 - val_acc: 0.5217\n",
            "Epoch 23/30\n",
            "687/687 [==============================] - 258s 376ms/step - loss: 0.2933 - acc: 0.8919 - val_loss: 2.1452 - val_acc: 0.5609\n",
            "Epoch 24/30\n",
            "687/687 [==============================] - 258s 376ms/step - loss: 0.2870 - acc: 0.8944 - val_loss: 2.3770 - val_acc: 0.5043\n",
            "Epoch 25/30\n",
            "687/687 [==============================] - 257s 374ms/step - loss: 0.2508 - acc: 0.9086 - val_loss: 2.5653 - val_acc: 0.5522\n",
            "Epoch 26/30\n",
            "687/687 [==============================] - 254s 369ms/step - loss: 0.2624 - acc: 0.9071 - val_loss: 2.5798 - val_acc: 0.5304\n",
            "Epoch 27/30\n",
            "687/687 [==============================] - 256s 372ms/step - loss: 0.3062 - acc: 0.8959 - val_loss: 2.6480 - val_acc: 0.5174\n",
            "Epoch 28/30\n",
            "687/687 [==============================] - 257s 375ms/step - loss: 0.2527 - acc: 0.9087 - val_loss: 2.3572 - val_acc: 0.5391\n",
            "Epoch 29/30\n",
            "687/687 [==============================] - 253s 368ms/step - loss: 0.2405 - acc: 0.9147 - val_loss: 2.5566 - val_acc: 0.5478\n",
            "Epoch 30/30\n",
            "687/687 [==============================] - 251s 366ms/step - loss: 0.2378 - acc: 0.9147 - val_loss: 2.5892 - val_acc: 0.5522\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "AcSgBB28PxWR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 271
        },
        "outputId": "43b19f00-14a7-4dff-b464-3c2d7add6057"
      },
      "source": [
        "# 4. Predizer modelo de aprendizado\n",
        "\n",
        "predictions = model.predict(X_test, batch_size=32)\n",
        "print(classification_report(Y_test.argmax(axis=1),\n",
        "\tpredictions.argmax(axis=1), target_names=lb.classes_))"
      ],
      "execution_count": 11,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "                     precision    recall  f1-score   support\n",
            "\n",
            "  carcinoma_in_situ       0.67      0.47      0.55        38\n",
            "   light_dysplastic       0.53      0.74      0.62        46\n",
            "moderate_dysplastic       0.36      0.32      0.34        37\n",
            "    normal_columnar       0.48      0.52      0.50        25\n",
            "normal_intermediate       0.82      0.82      0.82        17\n",
            " normal_superficiel       0.84      0.89      0.86        18\n",
            "  severe_dysplastic       0.47      0.41      0.43        49\n",
            "\n",
            "           accuracy                           0.55       230\n",
            "          macro avg       0.60      0.60      0.59       230\n",
            "       weighted avg       0.55      0.55      0.55       230\n",
            "\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "O7RX7dXuPxxY",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 299
        },
        "outputId": "ad496c21-3f63-4c43-cc0b-da9ccaf6bb5c"
      },
      "source": [
        "# Grafico com Histórico de acuracia por Epoch\n",
        "\n",
        "N = EPOCHS\n",
        "plt.style.use(\"ggplot\")\n",
        "plt.figure()\n",
        "plt.plot(np.arange(0, N), result.history[\"loss\"], label=\"train_loss\")\n",
        "plt.plot(np.arange(0, N), result.history[\"val_loss\"], label=\"val_loss\")\n",
        "plt.plot(np.arange(0, N), result.history[\"acc\"], label=\"train_acc\")\n",
        "plt.plot(np.arange(0, N), result.history[\"val_acc\"], label=\"val_acc\")\n",
        "plt.title(\"Treinamento: Loss e Accuracy do Dataset HERLEV\")\n",
        "plt.xlabel(\"Epoch #\")\n",
        "plt.ylabel(\"Loss/Accuracy\")\n",
        "plt.legend(loc=\"lower left\")\n",
        "plt.show()"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAZIAAAEaCAYAAAA7YdFPAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "D7iJCylAP37K",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 55
        },
        "outputId": "dbfe4665-24e5-459e-c1d3-4480e0614aa7"
      },
      "source": [
        "#plt.savefig(args[\"plot\"])\n",
        "\n",
        "\n",
        "# Leitura do banco de imagem\n",
        "\n",
        "# -=== Estudo sobre o Banco de Imagem Herlev ===-\n",
        "\n",
        "\n",
        "\n",
        "# - Quantidade de imagens para cada valor:\n",
        "\n",
        "# 1. Montagem do Dataset\n",
        "\n",
        "\n",
        "\"\"\"\n",
        "train_datagen = ImageDataGenerator(\n",
        "                    rescale=1./255,\n",
        "                    shear_range=0.2,\n",
        "                    zoom_range=0.2,\n",
        "                    horizontal_flip=True,\n",
        "                    validation_split=0.2)\n",
        "\n",
        "valid_datagen = ImageDataGenerator(rescale=1./255, validation_split=0.2)\n",
        "\n",
        "train_generator = train_datagen.flow_from_directory(\n",
        "    directory=r'./dataset/herlev',\n",
        "    target_size=(250, 250),\n",
        "    batch_size=32,\n",
        "    color_mode='rgb',\n",
        "    class_mode='categorical',\n",
        "    seed=1,\n",
        "    shuffle=True,\n",
        "    subset='training'\n",
        ")\n",
        "\n",
        "validator_generator = valid_datagen.flow_from_directory(\n",
        "    directory=r'./dataset/herlev',\n",
        "    target_size=(250, 250),\n",
        "    batch_size=32,\n",
        "    color_mode='rgb',\n",
        "    class_mode='categorical',\n",
        "    seed=1,\n",
        "    shuffle=True,\n",
        "    subset='validation'\n",
        ")\n",
        "\"\"\"\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "\n",
        "# 3. Treinamento Rede Neural\n",
        "\n",
        "\"\"\"\n",
        "model.fit_generator(\n",
        "        train_generator,\n",
        "        steps_per_epoch=len(train_generator),\n",
        "        epochs=10,\n",
        "        validation_data=validator_generator,\n",
        "        validation_steps=len(validator_generator)\n",
        ")\n",
        "\"\"\"\n",
        "\n",
        "# 4. Prever Dados\n",
        "\n",
        "# model.predict()\n",
        "\n",
        "# 5. Postar Resultados"
      ],
      "execution_count": 13,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'\\nmodel.fit_generator(\\n        train_generator,\\n        steps_per_epoch=len(train_generator),\\n        epochs=10,\\n        validation_data=validator_generator,\\n        validation_steps=len(validator_generator)\\n)\\n'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 13
        }
      ]
    }
  ]
}