analise-imagens-medicas.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# MBA FIAP Inteligência Artificial & Machine Learning\n",
    "\n",
    "## Visão Computacional: Análise de Imagens Médicas\n",
    "\n",
    "## 1. Introdução\n",
    "\n",
    "As tecnologias de imagens médicas estão cada vez mais integradas aos sitemas de visão computacional, incluindo as imagens de raio-x.\n",
    "\n",
    "Modelos de equipamentos modernos geram imagens digitais deste tipo de exame, proporcionando análises mais completas e menos _ad-hoc_, com isso algumas pré-análises podem ser realizadas por aplicações baseadas em inteligência artificial para confirmar ou sugerir diagnósticos ao profissional responsável pelo exame.\n",
    "\n",
    "No campo dos diagósticos por raios-x, a pnenumonia é uma das enfermidades onde seu uso é um dos mais aplicados para determinar o curso de tratamento.\n",
    "\n",
    "<p align=\"center\">\n",
    "    <img src=\"imagens/NORMAL2-IM-1422-0001.jpeg\">\n",
    "</p>\n",
    "\n",
    "## 2. Instruções\n",
    "\n",
    "Este projeto final tem como objetivo explorar os conhecimentos adquiridos nas aulas práticas.\n",
    "\n",
    "Por meio de uma trilha guiada, iremos constuir um modelo que seja capaz de classificar imagens de raio-x para determinar se a determinada pessoa está com alguma condição que necessita maiores cuidados.\n",
    "\n",
    "De acordo com as imagens disponíveis para o treinamento e validação, será de critério do grupo selecionar as quantidades ideais ou até mesmo pré-processar as imagens para obter o melhor resultado, nos principais indicadores de performance, como precisão, sensibilidade e pontuação F1.\n",
    "\n",
    "Este projeto poderá ser feita por grupos de até 4 pessoas.\n",
    "Caso este projeto seja substitutivo, deverá ser realizado por apenas uma pessoa.\n",
    "\n",
    "| Nome dos Integrantes     | RM            | Turma |\n",
    "| :----------------------- | :------------- | :-----: |\n",
    "| Fabio Rizzi              | RM 336656      | `11IA` |\n",
    "| Jurandir Ventura         | RM 336634      | `11IA` |\n",
    "| Marcelo Preto            | RM 336632      | `11IA` |\n",
    "| Thiago Alexandre         | RM 336583      | `11IA` |\n",
    "\n",
    "Por ser um projeto guiado, fique atento quando houver as marcações **Implementação** indica que é necessário realizar alguma implementação em Python no bloco a seguir onde há a inscrição ```##IMPLEMENTAR``` e **Resposta** indica que é esperado uma resposta objetiva relacionado a algum questionamento. \n",
    "\n",
    "**Cada grupo pode utilizar nas respostas objetivas quaisquer itens necessários que enriqueçam seu ponto vista, como gráficos, fotos e, até mesmo, trechos de código-fonte.**\n",
    "\n",
    "Pode-se utilizar quantos blocos forem necessários para realizar determinadas implementações ou utilizá-las para justificar as respostas. Não é obrigatório utilizar somente o bloco indicado.\n",
    "\n",
    "Ao final não se esqueça de subir os arquivos do projeto nas contas do GitHub de cada membro, ou subir na do representante do grupo e os membros realizarem o fork do projeto.\n",
    "\n",
    "A avaliação terá mais ênfase nos seguintes tópicos de desenvolvimento do projeto:\n",
    " \n",
    "1. __Pré-Processamento__\n",
    "2. __Classificação__\n",
    "3. __Performance__\n",
    "4. __Conclusões Finais__"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1 Detalhe do problema: a pneunomia\n",
    "\n",
    "Fonte: [artigo](https://drauziovarella.uol.com.br/doencas-e-sintomas/pneumonia) do Dr. Drauzio Varella.\n",
    "\n",
    "Pneumonias são infecções que se instalam nos pulmões, órgãos duplos localizados um de cada lado da caixa torácica. Podem acometer a região dos alvéolos pulmonares onde desembocam as ramificações terminais dos brônquios e, às vezes, os interstícios (espaço entre um alvéolo e outro).\n",
    "\n",
    "Basicamente, pneumonia é provocada pela penetração de um agente infeccioso ou irritante (bactérias, vírus, fungos e por reações alérgicas) no espaço alveolar, onde ocorre a troca gasosa. Esse local deve estar sempre muito limpo, livre de substâncias que possam impedir o contato do ar com o sangue.\n",
    "\n",
    "Exame clínico, auscultação dos pulmões e radiografias de tórax são recursos essenciais para o diagnóstico de pneumonia.\n",
    "\n",
    "<p align=\"center\">\n",
    "    <img src=\"imagens/pneumonia.jpeg\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2 Diagnóstico por raio-x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O exame de raio-x traz diferenças em cada tipo de diagnóstico, sendo considerado os seguintes grupos de análise: **normal** (ou controle) onde não há nenhuma condição de infeção, **bacterial pneumonia** (pneumonia bacteriana) que representa a condição de infecção bacteriana e **viral pneumonia** que indica a condição de infecção vira. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p align=\"center\">\n",
    "<img src=\"imagens/raiox.png\" height=\"60%\" width=\"60%\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "As imagens de controle não são mais brancas ao centro que é onde fica o coração. Já nas imagens com pneumonia é possível notar regiões brancas ao redor dos pulmões, que é como o exame identifica as secreções responsáveis pela infeçcão.\n",
    "\n",
    "Quando mais regiões brancas ao redor do pulmão mais severa é a inflamação e menos se observa dos detalhes dos pulmões, ficando um pouco esmaecido diante desta condição."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1 Problema\n",
    "\n",
    "Construir um classificador utilizando _transfer learning_ para identificar as seguintes classes: **controle**, **pneumonia bacteriana** e **pneumonia viral**.\n",
    "\n",
    "Para construir este classificador, utilize o dataset do [Kaggle Chest Ray Pneumonia](https://www.kaggle.com/paultimothymooney/chest-xray-pneumonia) e organize os dados de forma a separar em cada uma das classes que já estão definidas no diretório ```raiox```, sendo ```controle``` para as imagens normais (sem inflamação), ```bacteria``` para as imagens de pneumonia bacteriana e ```viral``` para as imagens de pneumonia viral.\n",
    "\n",
    "Determine a quantidade de imagens a serem treinadas e validadas. Utiliza pelo menos, 100 imagens para cada classe.\n",
    "\n",
    "Compare os resultados com pelo menos 3 classificadores, obtendo os valores de **precisão (precision)**, **sensibilidade (recall)** e **pontuação F1 (F1 Score)**. No guia abaixo, foi indicado os seguintes modelos: ResNet50, VGG16 e VGG19. \n",
    "\n",
    ">Importante: a escolha do número de imagens devem ser o suficiente para alcançar o valor de **precisão** mínima de 70%.\n",
    "\n",
    "A construção do modelo será utilizada o framework Keras."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Qual o número de imagens que foram selecionadas para cada classe?"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<span style='color:blue'>**Resposta:**</span>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "Para treino e validação, foram usadas as seguintes quantidades de imagens para cada classe:"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**viral**: 237 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**bacteria**: 3517 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**controle**: 1342 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n"
     ]
    },
    {
     "data": {
      "text/markdown": [
       "Para fins de demonstração, podemos ainda usar a quantidade de imagens abaixo, que está no diretório raiox/test:"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**viral**: 149 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**bacteria**: 242 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/markdown": [
       "**controle**: 234 imagens"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "from IPython.display import Markdown, display\n",
    "import os\n",
    "\n",
    "display(Markdown(\"<span style='color:blue'>**Resposta:**</span>\"))\n",
    "display(Markdown(\"Para treino e validação, foram usadas as seguintes quantidades de imagens para cada classe:\"))\n",
    "\n",
    "classes = ['viral','bacteria','controle']\n",
    "main_directory = ['raiox/train']\n",
    "\n",
    "def count_files(dir):\n",
    "    _, _, files = next(os.walk(dir))\n",
    "    return len(files)\n",
    "\n",
    "for classe in classes:\n",
    "    qtd = 0\n",
    "    for dir in main_directory:\n",
    "        qtd = qtd + count_files(dir + \"/\" + classe)\n",
    "    display(Markdown((\"**\" + classe + \"**: \" + str(qtd) + \" imagens\")))\n",
    "    \n",
    "\n",
    "play_directory = ['raiox/test']\n",
    "print()\n",
    "\n",
    "display(Markdown(\"Para fins de demonstração, podemos ainda usar a quantidade de imagens abaixo, que está no diretório raiox/test:\"))\n",
    "\n",
    "for classe in classes:\n",
    "    qtd = 0\n",
    "    for dir in play_directory:\n",
    "        qtd = qtd + count_files(dir + \"/\" + classe)\n",
    "    display(Markdown((\"**\" + classe + \"**: \" + str(qtd) + \" imagens\")))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Organização do diretório de imagens também pode ser consultado em:\n",
    "\n",
    "#### https://drive.google.com/drive/folders/1NtI4-0I_2YS1wBPcjvnHIhD1mEmYwSDJ?usp=sharing"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2 Componentes obrigatórios\n",
    "\n",
    "Este projeto requer a instalação dos seguintes componentes, via ```conda install```:\n",
    "\n",
    "* Keras\n",
    "* Tensorflow\n",
    "* Pillow\n",
    "* Matplotlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from PIL import Image\n",
    "\n",
    "%matplotlib inline\n",
    "\n",
    "import keras\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from keras.applications import ResNet50\n",
    "from keras.applications import VGG16, VGG19\n",
    "from keras.applications import Xception\n",
    "from keras.applications.xception import preprocess_input\n",
    "from keras.applications.resnet50 import preprocess_input\n",
    "from keras.applications.vgg16 import preprocess_input\n",
    "from keras import Model, layers\n",
    "from keras.models import load_model, model_from_json\n",
    "\n",
    "import keras.backend as K"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.3 Carregando imagens de treinamento e validação\n",
    "\n",
    "Selecione a melhor divisão entre dados de treinamento e validação. O número deverá ser representado em número fracionário, 5% equivale a 0.05, por exemplo."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "## IMPLEMENTE\n",
    "\n",
    "# Essa é a porção que será usada para validação. Tais imagens estão no mesmo diretório raiox/train/\n",
    "# A função ImageDataGenerator do Keras fará essa seleção de forma automática nas próximas linhas de código.\n",
    "# O diretório raiox/train foi construído com as 3 classes a serem analisadas localmente, mas podem ser acessados em:\n",
    "# https://drive.google.com/drive/folders/1NtI4-0I_2YS1wBPcjvnHIhD1mEmYwSDJ?usp=sharing\n",
    "\n",
    "divisao_treino_validacao = 0.35"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 3312 images belonging to 3 classes.\n"
     ]
    }
   ],
   "source": [
    "train_datagen = ImageDataGenerator(validation_split=divisao_treino_validacao)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "    \"raiox/train/\",\n",
    "    batch_size=32,\n",
    "    class_mode=\"categorical\",\n",
    "    color_mode=\"rgb\",\n",
    "    target_size=(224,224),\n",
    "    subset=\"training\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 1781 images belonging to 3 classes.\n"
     ]
    }
   ],
   "source": [
    "val_generator = train_datagen.flow_from_directory(\n",
    "    \"raiox/train/\",\n",
    "    batch_size=32,\n",
    "    class_mode=\"categorical\",\n",
    "    color_mode=\"rgb\",\n",
    "    target_size=(224,224),\n",
    "    subset=\"validation\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "({'bacteria': 0, 'controle': 1, 'viral': 2},\n",
       " {'bacteria': 0, 'controle': 1, 'viral': 2})"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train_generator.class_indices, val_generator.class_indices"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.4 Modelos de transfer learning\n",
    "\n",
    "O Keras já possui classes especializadas para os seguintes modelos de deep-learning treinados com o conjunto de dados [ImageNet](http://www.image-net.org/):\n",
    "  \n",
    "* Xception\n",
    "* VGG16\n",
    "* VGG19\n",
    "* ResNet50\n",
    "* InceptionV3\n",
    "* InceptionResNetV2\n",
    "* MobileNet\n",
    "* DenseNet\n",
    "* NASNet\n",
    "* MobileNetV2\n",
    "\n",
    "Mais detalhes, veja na [documentação do Keras](https://keras.io/applications/)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Para este estudo, vamos utilizar para avaliação as seguintes arquiteturas: RestNet50, VGG15 e VGG19."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.5 Indicadores de desempenho\n",
    "\n",
    "O Keras não possui os indicadores de desempenho como precisão, sensibilidade e pontuação f1 por padrão, portanto precisamos implementar externamente."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [],
   "source": [
    "def recall_score(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    possible_positives = K.sum(K.round(K.clip(y_true, 0, 1)))\n",
    "    recall = true_positives / (possible_positives + K.epsilon())\n",
    "    return recall\n",
    "\n",
    "def precision_score(y_true, y_pred):\n",
    "    true_positives = K.sum(K.round(K.clip(y_true * y_pred, 0, 1)))\n",
    "    predicted_positives = K.sum(K.round(K.clip(y_pred, 0, 1)))\n",
    "    precision = true_positives / (predicted_positives + K.epsilon())\n",
    "    return precision\n",
    "\n",
    "def f1_score(y_true, y_pred):\n",
    "    precision = precision_score(y_true, y_pred)\n",
    "    recall = recall_score(y_true, y_pred)\n",
    "    return 2*((precision*recall)/(precision+recall+K.epsilon()))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.5.1 Arquitetura ResNet50"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Explique como é constituída a arquitetura do ResNet50? *Utilize, se necessário, gráficos, projetos que utilizam essa arquitetura. Detalhe também sua topologia em camadas e mostre quais as situações essa arquitetura pode ter mais êxito e quais cenários não tem.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A ResNet50 é uma rede neural convolucional (CNN) que usa 50 camadas convolucionais, sendo distribuídas em 5 etapas, com  convolução da rede. É um modelo pré-treinado e possui a sequência de camadas descritas abaixo:\n",
    "\n",
    "* Primeiro com um convolução de tamanho 7 x 7, com de kernel 64 e saltos de 2, formando uma camada.\n",
    "\n",
    "* Na próxima convolução temos o tamanho 1 x 1 de 64 kernel, seguido de 3 x 3 com 64 e por último 1 x 1 de 256. Essas 3 camadas são repetidas 3 vezes e por isso formam 9 camadas nesta etapa.\n",
    "\n",
    "* Na próxima vemos um kernel de 1 x 1 de 128, depois 3 x 3 de 128 e por fim, 1 x 1 de 512, sendo repetido por 4 vezes e assim formam mais 12 camadas (3 * 4) neste passo.\n",
    "\n",
    "* Depois temos um kernel de 1 x 1 de 256, depois 3 x 3 de 256 e 1 x 1 de 1024, repetido 6 vezes, formando mais 18 camadas nesta etapa.\n",
    "\n",
    "* Na última camada de convolução temos kernel de 1 x 1 de 512, 3 x 3 de 512 e 1 x 1 de 2048, repetidos por 3 vezes e formando mais 9 camadas.\n",
    "\n",
    "* Por fim temos uma camada totalmente conectada no final, formando mais uma camada e assim totalizando as 50 camadas.\n",
    "\n",
    "1 + 9 + 12 + 18 + 9 + 1 = 50 camadas.\n",
    "\n",
    "A Arquitetura pode ser melhor entendida pela figura abaixo:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p align=\"center\">\n",
    "<img src=\"imagens/resnet50.jpg\" height=\"50%\" width=\"50%\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### PRÓS:\n",
    "A Resnet50 funciona bem para classificação de imagens e detecção de objetos. Tem a vantagem de usar o melhor resultado, mesmo executando uma camada de pior resultado posterior, com uso do método \"Skip Connection\", trazendo para a saída resultados de camadas anteriores, assim elimina o problema de redução ou desaparecimento de gradiente.\n",
    "\n",
    "### Utilizando-se da técnica de Transfer Learning, onde usamos um modelo pré-treinado, excluímos a camada densa original para usar às novas classes: bacteria, controle e viral e isso torna o processo de modelagem bem rápido, com pouca parametrização. "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### CONTRA:\n",
    "Por usar modelo pré-treinado, exige menos poder computacional ou parametrização, em contrapartida é limitado se for usado para classificação de imagens com grau de complexidade maior, ou para o uso de muitas classes."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Referências:\n",
    "\n",
    "#### https://github.com/priya-dwivedi/Deep-Learning/blob/master/resnet_keras/Residual_Networks_yourself.ipynb\n",
    "\n",
    "#### https://towardsdatascience.com/understanding-and-coding-a-resnet-in-keras-446d7ff84d33\n",
    "\n",
    "#### https://cv-tricks.com/keras/understand-implement-resnets/\n",
    "\n",
    "#### https://www.eenewsanalog.com/news/resnet-50-misleading-machine-learning-inference-benchmark-megapixel-images"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "A técnica de transfer learning consiste de utilizar o mesmo modelo e treiná-lo para outas imagens. Por tal motivo, excluímos a última camada para modelar com as classes que definimos, ou seja, **controle**, **bacteriana** e **viral**."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Informe a quantidade de classes a serem classificadas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Temos 3 classes, sendo elas => {'bacteria': 0, 'controle': 1, 'viral': 2}\n"
     ]
    }
   ],
   "source": [
    "## IMPLEMENTE\n",
    "# Foi realizada a modificação do dataset original:\n",
    "# O diretório com as imagens do raio-x \"NORMAL\" foi renomeado para \"controle\"\n",
    "# O diretório com as imagens do raio-x \"PNEUMONIA\" foi renomeado para \"bacteria\"\n",
    "# As imagens que possuíam 'virus' no nome do arquivo da pasta \"PNEUMONIA\" foram movidos a uma nova pasta chamada \"viral\"\n",
    "\n",
    "#Assim temos 3 classes: bacteria, controle, viral. Como descrito em:\n",
    "#train_generator.class_indices, val_generator.class_indices\n",
    "\n",
    "qtde_classes = len(train_generator.class_indices)\n",
    "\n",
    "print('Temos',qtde_classes,'classes, sendo elas =>', train_generator.class_indices)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "conv_base = ResNet50(include_top=False)\n",
    "\n",
    "for layer in conv_base.layers:\n",
    "     layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "tags": [
     "outputPrepend"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_1\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, None, None,  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "conv1_pad (ZeroPadding2D)       (None, None, None, 3 0           input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "conv1_conv (Conv2D)             (None, None, None, 6 9472        conv1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv1_bn (BatchNormalization)   (None, None, None, 6 256         conv1_conv[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv1_relu (Activation)         (None, None, None, 6 0           conv1_bn[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "pool1_pad (ZeroPadding2D)       (None, None, None, 6 0           conv1_relu[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "pool1_pool (MaxPooling2D)       (None, None, None, 6 0           pool1_pad[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_1_conv (Conv2D)    (None, None, None, 6 4160        pool1_pool[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_1_bn (BatchNormali (None, None, None, 6 256         conv2_block1_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_1_relu (Activation (None, None, None, 6 0           conv2_block1_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_2_conv (Conv2D)    (None, None, None, 6 36928       conv2_block1_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_2_bn (BatchNormali (None, None, None, 6 256         conv2_block1_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_2_relu (Activation (None, None, None, 6 0           conv2_block1_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_0_conv (Conv2D)    (None, None, None, 2 16640       pool1_pool[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_3_conv (Conv2D)    (None, None, None, 2 16640       conv2_block1_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_0_bn (BatchNormali (None, None, None, 2 1024        conv2_block1_0_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_3_bn (BatchNormali (None, None, None, 2 1024        conv2_block1_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_add (Add)          (None, None, None, 2 0           conv2_block1_0_bn[0][0]          \n",
      "                                                                 conv2_block1_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block1_out (Activation)   (None, None, None, 2 0           conv2_block1_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_1_conv (Conv2D)    (None, None, None, 6 16448       conv2_block1_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_1_bn (BatchNormali (None, None, None, 6 256         conv2_block2_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_1_relu (Activation (None, None, None, 6 0           conv2_block2_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_2_conv (Conv2D)    (None, None, None, 6 36928       conv2_block2_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_2_bn (BatchNormali (None, None, None, 6 256         conv2_block2_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_2_relu (Activation (None, None, None, 6 0           conv2_block2_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_3_conv (Conv2D)    (None, None, None, 2 16640       conv2_block2_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_3_bn (BatchNormali (None, None, None, 2 1024        conv2_block2_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_add (Add)          (None, None, None, 2 0           conv2_block1_out[0][0]           \n",
      "                                                                 conv2_block2_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block2_out (Activation)   (None, None, None, 2 0           conv2_block2_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_1_conv (Conv2D)    (None, None, None, 6 16448       conv2_block2_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_1_bn (BatchNormali (None, None, None, 6 256         conv2_block3_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_1_relu (Activation (None, None, None, 6 0           conv2_block3_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_2_conv (Conv2D)    (None, None, None, 6 36928       conv2_block3_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_2_bn (BatchNormali (None, None, None, 6 256         conv2_block3_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_2_relu (Activation (None, None, None, 6 0           conv2_block3_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_3_conv (Conv2D)    (None, None, None, 2 16640       conv2_block3_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_3_bn (BatchNormali (None, None, None, 2 1024        conv2_block3_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_add (Add)          (None, None, None, 2 0           conv2_block2_out[0][0]           \n",
      "                                                                 conv2_block3_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv2_block3_out (Activation)   (None, None, None, 2 0           conv2_block3_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_1_conv (Conv2D)    (None, None, None, 1 32896       conv2_block3_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_1_bn (BatchNormali (None, None, None, 1 512         conv3_block1_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_1_relu (Activation (None, None, None, 1 0           conv3_block1_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_2_conv (Conv2D)    (None, None, None, 1 147584      conv3_block1_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_2_bn (BatchNormali (None, None, None, 1 512         conv3_block1_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_2_relu (Activation (None, None, None, 1 0           conv3_block1_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_0_conv (Conv2D)    (None, None, None, 5 131584      conv2_block3_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_3_conv (Conv2D)    (None, None, None, 5 66048       conv3_block1_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_0_bn (BatchNormali (None, None, None, 5 2048        conv3_block1_0_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_3_bn (BatchNormali (None, None, None, 5 2048        conv3_block1_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_add (Add)          (None, None, None, 5 0           conv3_block1_0_bn[0][0]          \n",
      "                                                                 conv3_block1_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block1_out (Activation)   (None, None, None, 5 0           conv3_block1_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_1_conv (Conv2D)    (None, None, None, 1 65664       conv3_block1_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_1_bn (BatchNormali (None, None, None, 1 512         conv3_block2_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_1_relu (Activation (None, None, None, 1 0           conv3_block2_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_2_conv (Conv2D)    (None, None, None, 1 147584      conv3_block2_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_2_bn (BatchNormali (None, None, None, 1 512         conv3_block2_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_2_relu (Activation (None, None, None, 1 0           conv3_block2_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_3_conv (Conv2D)    (None, None, None, 5 66048       conv3_block2_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_3_bn (BatchNormali (None, None, None, 5 2048        conv3_block2_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_add (Add)          (None, None, None, 5 0           conv3_block1_out[0][0]           \n",
      "                                                                 conv3_block2_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block2_out (Activation)   (None, None, None, 5 0           conv3_block2_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_1_conv (Conv2D)    (None, None, None, 1 65664       conv3_block2_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_1_bn (BatchNormali (None, None, None, 1 512         conv3_block3_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_1_relu (Activation (None, None, None, 1 0           conv3_block3_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_2_conv (Conv2D)    (None, None, None, 1 147584      conv3_block3_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_2_bn (BatchNormali (None, None, None, 1 512         conv3_block3_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_2_relu (Activation (None, None, None, 1 0           conv3_block3_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_3_conv (Conv2D)    (None, None, None, 5 66048       conv3_block3_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_3_bn (BatchNormali (None, None, None, 5 2048        conv3_block3_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_add (Add)          (None, None, None, 5 0           conv3_block2_out[0][0]           \n",
      "                                                                 conv3_block3_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block3_out (Activation)   (None, None, None, 5 0           conv3_block3_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_1_conv (Conv2D)    (None, None, None, 1 65664       conv3_block3_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_1_bn (BatchNormali (None, None, None, 1 512         conv3_block4_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_1_relu (Activation (None, None, None, 1 0           conv3_block4_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_2_conv (Conv2D)    (None, None, None, 1 147584      conv3_block4_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_2_bn (BatchNormali (None, None, None, 1 512         conv3_block4_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_2_relu (Activation (None, None, None, 1 0           conv3_block4_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_3_conv (Conv2D)    (None, None, None, 5 66048       conv3_block4_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_3_bn (BatchNormali (None, None, None, 5 2048        conv3_block4_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_add (Add)          (None, None, None, 5 0           conv3_block3_out[0][0]           \n",
      "                                                                 conv3_block4_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv3_block4_out (Activation)   (None, None, None, 5 0           conv3_block4_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_1_conv (Conv2D)    (None, None, None, 2 131328      conv3_block4_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block1_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_1_relu (Activation (None, None, None, 2 0           conv4_block1_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block1_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block1_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_2_relu (Activation (None, None, None, 2 0           conv4_block1_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_0_conv (Conv2D)    (None, None, None, 1 525312      conv3_block4_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block1_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_0_bn (BatchNormali (None, None, None, 1 4096        conv4_block1_0_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block1_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_add (Add)          (None, None, None, 1 0           conv4_block1_0_bn[0][0]          \n",
      "                                                                 conv4_block1_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block1_out (Activation)   (None, None, None, 1 0           conv4_block1_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_1_conv (Conv2D)    (None, None, None, 2 262400      conv4_block1_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block2_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_1_relu (Activation (None, None, None, 2 0           conv4_block2_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block2_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block2_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_2_relu (Activation (None, None, None, 2 0           conv4_block2_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block2_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block2_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_add (Add)          (None, None, None, 1 0           conv4_block1_out[0][0]           \n",
      "                                                                 conv4_block2_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block2_out (Activation)   (None, None, None, 1 0           conv4_block2_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_1_conv (Conv2D)    (None, None, None, 2 262400      conv4_block2_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block3_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_1_relu (Activation (None, None, None, 2 0           conv4_block3_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block3_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block3_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_2_relu (Activation (None, None, None, 2 0           conv4_block3_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block3_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block3_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_add (Add)          (None, None, None, 1 0           conv4_block2_out[0][0]           \n",
      "                                                                 conv4_block3_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block3_out (Activation)   (None, None, None, 1 0           conv4_block3_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_1_conv (Conv2D)    (None, None, None, 2 262400      conv4_block3_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block4_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_1_relu (Activation (None, None, None, 2 0           conv4_block4_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block4_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block4_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_2_relu (Activation (None, None, None, 2 0           conv4_block4_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block4_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block4_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_add (Add)          (None, None, None, 1 0           conv4_block3_out[0][0]           \n",
      "                                                                 conv4_block4_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block4_out (Activation)   (None, None, None, 1 0           conv4_block4_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_1_conv (Conv2D)    (None, None, None, 2 262400      conv4_block4_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block5_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_1_relu (Activation (None, None, None, 2 0           conv4_block5_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block5_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block5_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_2_relu (Activation (None, None, None, 2 0           conv4_block5_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block5_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block5_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_add (Add)          (None, None, None, 1 0           conv4_block4_out[0][0]           \n",
      "                                                                 conv4_block5_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block5_out (Activation)   (None, None, None, 1 0           conv4_block5_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_1_conv (Conv2D)    (None, None, None, 2 262400      conv4_block5_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_1_bn (BatchNormali (None, None, None, 2 1024        conv4_block6_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_1_relu (Activation (None, None, None, 2 0           conv4_block6_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_2_conv (Conv2D)    (None, None, None, 2 590080      conv4_block6_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_2_bn (BatchNormali (None, None, None, 2 1024        conv4_block6_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_2_relu (Activation (None, None, None, 2 0           conv4_block6_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_3_conv (Conv2D)    (None, None, None, 1 263168      conv4_block6_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_3_bn (BatchNormali (None, None, None, 1 4096        conv4_block6_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_add (Add)          (None, None, None, 1 0           conv4_block5_out[0][0]           \n",
      "                                                                 conv4_block6_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv4_block6_out (Activation)   (None, None, None, 1 0           conv4_block6_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_1_conv (Conv2D)    (None, None, None, 5 524800      conv4_block6_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_1_bn (BatchNormali (None, None, None, 5 2048        conv5_block1_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_1_relu (Activation (None, None, None, 5 0           conv5_block1_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_2_conv (Conv2D)    (None, None, None, 5 2359808     conv5_block1_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_2_bn (BatchNormali (None, None, None, 5 2048        conv5_block1_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_2_relu (Activation (None, None, None, 5 0           conv5_block1_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_0_conv (Conv2D)    (None, None, None, 2 2099200     conv4_block6_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_3_conv (Conv2D)    (None, None, None, 2 1050624     conv5_block1_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_0_bn (BatchNormali (None, None, None, 2 8192        conv5_block1_0_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_3_bn (BatchNormali (None, None, None, 2 8192        conv5_block1_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_add (Add)          (None, None, None, 2 0           conv5_block1_0_bn[0][0]          \n",
      "                                                                 conv5_block1_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block1_out (Activation)   (None, None, None, 2 0           conv5_block1_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_1_conv (Conv2D)    (None, None, None, 5 1049088     conv5_block1_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_1_bn (BatchNormali (None, None, None, 5 2048        conv5_block2_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_1_relu (Activation (None, None, None, 5 0           conv5_block2_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_2_conv (Conv2D)    (None, None, None, 5 2359808     conv5_block2_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_2_bn (BatchNormali (None, None, None, 5 2048        conv5_block2_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_2_relu (Activation (None, None, None, 5 0           conv5_block2_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_3_conv (Conv2D)    (None, None, None, 2 1050624     conv5_block2_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_3_bn (BatchNormali (None, None, None, 2 8192        conv5_block2_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_add (Add)          (None, None, None, 2 0           conv5_block1_out[0][0]           \n",
      "                                                                 conv5_block2_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block2_out (Activation)   (None, None, None, 2 0           conv5_block2_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_1_conv (Conv2D)    (None, None, None, 5 1049088     conv5_block2_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_1_bn (BatchNormali (None, None, None, 5 2048        conv5_block3_1_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_1_relu (Activation (None, None, None, 5 0           conv5_block3_1_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_2_conv (Conv2D)    (None, None, None, 5 2359808     conv5_block3_1_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_2_bn (BatchNormali (None, None, None, 5 2048        conv5_block3_2_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_2_relu (Activation (None, None, None, 5 0           conv5_block3_2_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_3_conv (Conv2D)    (None, None, None, 2 1050624     conv5_block3_2_relu[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_3_bn (BatchNormali (None, None, None, 2 8192        conv5_block3_3_conv[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_add (Add)          (None, None, None, 2 0           conv5_block2_out[0][0]           \n",
      "                                                                 conv5_block3_3_bn[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "conv5_block3_out (Activation)   (None, None, None, 2 0           conv5_block3_add[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d (Globa (None, 2048)         0           conv5_block3_out[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "dense (Dense)                   (None, 128)          262272      global_average_pooling2d[0][0]   \n",
      "__________________________________________________________________________________________________\n",
      "dense_1 (Dense)                 (None, 3)            387         dense[0][0]                      \n",
      "==================================================================================================\n",
      "Total params: 23,850,371\n",
      "Trainable params: 262,659\n",
      "Non-trainable params: 23,587,712\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "x = conv_base.output\n",
    "x = layers.GlobalAveragePooling2D()(x)\n",
    "x = layers.Dense(128, activation='relu')(x) \n",
    "\n",
    "predictions = layers.Dense(qtde_classes, activation='softmax')(x)\n",
    "model = Model(conv_base.input, predictions)\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = keras.optimizers.Adam()\n",
    "model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=[precision_score, recall_score, f1_score])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "O número de épocas define quantas vezes o modelo irá treinar e validar o erro, assim ajustando os pesos para melhor convergência.\n",
    "Escolha o número adequado de épocas para alcançarmos pelo menos **70% de precisão de validação**."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "## IMPLEMENTE\n",
    "\n",
    "# A técnica de checkpoint permite salvar o modelo e para analisar como está sendo a melhoria do modelo em cada época\n",
    "# É possível salvar sempre o melhor modelo apenas. Se for pior que o carregado, então interrompe pelo Early Stopping.\n",
    "# Em função das regras do projeto, não foi usado, mas está aqui para evidenciar algumas coisas que aprendemos com o curso... rs\n",
    "# :-)\n",
    "\n",
    "'''\n",
    "from keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau\n",
    "\n",
    "checkpoint = ModelCheckpoint(filepath = 'modelo-resnet50.hdf5',\n",
    "                             monitor=\"precision_score\",\n",
    "                             mode=\"min\",\n",
    "                             save_best_only = True,\n",
    "                             verbose=1)\n",
    "\n",
    "earlystop = EarlyStopping(monitor = 'val_loss', \n",
    "                          min_delta = 0, \n",
    "                          patience = 10,\n",
    "                          verbose = 1,\n",
    "                          restore_best_weights = True)\n",
    "\n",
    "reduce_lr = ReduceLROnPlateau(monitor = 'val_loss',\n",
    "                              factor = 0.2,\n",
    "                              patience = 10,\n",
    "                              verbose = 1,\n",
    "                              min_delta = 0.0001)\n",
    "\n",
    "callbacks = [earlystop, checkpoint, reduce_lr]\n",
    "\n",
    "# Usamos os valores de checkpoint, early stopping e reduce_lr ao fazer o Fit do modelo\n",
    "# O Parâmetro preenchido abaixo é justamente callbacks.\n",
    "\n",
    "'''\n",
    "# Quantidade de épocas usada para todos os modelos:\n",
    "\n",
    "qtde_epocas = 10"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-13-68139f1e6072>:5: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use Model.fit, which supports generators.\n"
     ]
    }
   ],
   "source": [
    "%%capture saida1 --no-stderr\n",
    "history = model.fit_generator(generator=train_generator, \n",
    "                              epochs=qtde_epocas, \n",
    "                              validation_steps=5, \n",
    "                              steps_per_epoch=5, \n",
    "                              validation_data=val_generator)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "5/5 [==============================] - 18s 4s/step - loss: 1.7573 - precision_score: 0.6205 - recall_score: 0.6000 - f1_score: 0.6096 - val_loss: 0.5973 - val_precision_score: 0.8975 - val_recall_score: 0.7125 - val_f1_score: 0.7934\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 16s 3s/step - loss: 0.6471 - precision_score: 0.7543 - recall_score: 0.7063 - f1_score: 0.7279 - val_loss: 0.6425 - val_precision_score: 0.7923 - val_recall_score: 0.7875 - val_f1_score: 0.7899\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.4555 - precision_score: 0.8272 - recall_score: 0.8000 - f1_score: 0.8133 - val_loss: 0.3181 - val_precision_score: 0.9052 - val_recall_score: 0.9000 - val_f1_score: 0.9026\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 18s 4s/step - loss: 0.4598 - precision_score: 0.8369 - recall_score: 0.8188 - f1_score: 0.8274 - val_loss: 0.3149 - val_precision_score: 0.9157 - val_recall_score: 0.8813 - val_f1_score: 0.8981\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.2615 - precision_score: 0.9429 - recall_score: 0.9125 - f1_score: 0.9271 - val_loss: 0.2760 - val_precision_score: 0.9242 - val_recall_score: 0.9125 - val_f1_score: 0.9183\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3545 - precision_score: 0.8998 - recall_score: 0.8938 - f1_score: 0.8967 - val_loss: 0.3459 - val_precision_score: 0.9030 - val_recall_score: 0.8813 - val_f1_score: 0.8918\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.4219 - precision_score: 0.8683 - recall_score: 0.8250 - f1_score: 0.8458 - val_loss: 0.3484 - val_precision_score: 0.8859 - val_recall_score: 0.8625 - val_f1_score: 0.8739\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3163 - precision_score: 0.8863 - recall_score: 0.8813 - f1_score: 0.8837 - val_loss: 0.2783 - val_precision_score: 0.9246 - val_recall_score: 0.9187 - val_f1_score: 0.9216\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.2865 - precision_score: 0.8979 - recall_score: 0.8875 - f1_score: 0.8925 - val_loss: 0.5745 - val_precision_score: 0.7980 - val_recall_score: 0.7875 - val_f1_score: 0.7927\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3036 - precision_score: 0.9123 - recall_score: 0.9062 - f1_score: 0.9092 - val_loss: 0.3520 - val_precision_score: 0.8869 - val_recall_score: 0.8813 - val_f1_score: 0.8840\n"
     ]
    }
   ],
   "source": [
    "saida1.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Um modelo que converge bem possui o gráfico de perda (*loss*) descendente e os gráfico de precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*) em acendente."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Exibindo dados de Precisão\n",
    "plt.plot(history.history['precision_score'])\n",
    "plt.plot(history.history['val_precision_score'])\n",
    "plt.title('model precision')\n",
    "plt.ylabel('precision')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Sensibilidade\n",
    "plt.plot(history.history['recall_score'])\n",
    "plt.plot(history.history['val_recall_score'])\n",
    "plt.title('model recall')\n",
    "plt.ylabel('recall')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de F1 Score\n",
    "plt.plot(history.history['f1_score'])\n",
    "plt.plot(history.history['val_f1_score'])\n",
    "plt.title('model f1_score')\n",
    "plt.ylabel('f1_score')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Perda\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Avalie os gráficos de perda (*loss*), precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*)  e explique o comportamento de ambos no que tange a convergência do modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testamos com 2 épocas, 4 épocas, 8 épocas e por fim 10 épocas. A que mostrou uma convergência mais consistente foi justamente esta última.\n",
    "\n",
    "### É possível verificar que existe uma ascendência dos valores de precisão, sensibilidade e pontuação f1 nos gráficos de treino e que tem uma proximidade com os dados de testes, além de percebermos uma estabilização da média de validação nas últimas épocas, ficando sempre acima de 70%. No caso do gráfico de perda, podemos perceber que é descendente, atingindo estabilidade nas últimas épocas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Quais são os valores de **precisão (precision)**, **sensibilidade (recall)** de validação? \n",
    "\n",
    "*Estes valores são exibidos durante o treinamento, utilize a última saída, exemplo:*\n",
    "\n",
    "```\n",
    "Epoch 10/10 [==============================] - 45s 9s/step - loss: 0.1234 - precision_score: 0.9742 - recall_score: 0.9683 - f1_score: 0.9712 - val_loss: 0.8819 - val_precision_score: 0.6912 - val_recall_score: 0.5649 - val_f1_score: 0.6216```\n",
    "\n",
    "No caso acima, o valor de precisão, sensibilidade e pontuação de validação são, respectivamente, 69,12%, 56,49% e 62,16%."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<span style='color:blue'>**Resposta:**</span>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precisão (precision): 88.69%\n",
      "Sensibilidade (recall): 88.13%\n",
      "Pontuação de Validação (f1 score): 88.4%\n",
      "\n",
      "Conforme a saída:\n",
      "Epoch 1/10\n",
      "5/5 [==============================] - 18s 4s/step - loss: 1.7573 - precision_score: 0.6205 - recall_score: 0.6000 - f1_score: 0.6096 - val_loss: 0.5973 - val_precision_score: 0.8975 - val_recall_score: 0.7125 - val_f1_score: 0.7934\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 16s 3s/step - loss: 0.6471 - precision_score: 0.7543 - recall_score: 0.7063 - f1_score: 0.7279 - val_loss: 0.6425 - val_precision_score: 0.7923 - val_recall_score: 0.7875 - val_f1_score: 0.7899\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.4555 - precision_score: 0.8272 - recall_score: 0.8000 - f1_score: 0.8133 - val_loss: 0.3181 - val_precision_score: 0.9052 - val_recall_score: 0.9000 - val_f1_score: 0.9026\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 18s 4s/step - loss: 0.4598 - precision_score: 0.8369 - recall_score: 0.8188 - f1_score: 0.8274 - val_loss: 0.3149 - val_precision_score: 0.9157 - val_recall_score: 0.8813 - val_f1_score: 0.8981\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.2615 - precision_score: 0.9429 - recall_score: 0.9125 - f1_score: 0.9271 - val_loss: 0.2760 - val_precision_score: 0.9242 - val_recall_score: 0.9125 - val_f1_score: 0.9183\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3545 - precision_score: 0.8998 - recall_score: 0.8938 - f1_score: 0.8967 - val_loss: 0.3459 - val_precision_score: 0.9030 - val_recall_score: 0.8813 - val_f1_score: 0.8918\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.4219 - precision_score: 0.8683 - recall_score: 0.8250 - f1_score: 0.8458 - val_loss: 0.3484 - val_precision_score: 0.8859 - val_recall_score: 0.8625 - val_f1_score: 0.8739\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3163 - precision_score: 0.8863 - recall_score: 0.8813 - f1_score: 0.8837 - val_loss: 0.2783 - val_precision_score: 0.9246 - val_recall_score: 0.9187 - val_f1_score: 0.9216\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.2865 - precision_score: 0.8979 - recall_score: 0.8875 - f1_score: 0.8925 - val_loss: 0.5745 - val_precision_score: 0.7980 - val_recall_score: 0.7875 - val_f1_score: 0.7927\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 17s 3s/step - loss: 0.3036 - precision_score: 0.9123 - recall_score: 0.9062 - f1_score: 0.9092 - val_loss: 0.3520 - val_precision_score: 0.8869 - val_recall_score: 0.8813 - val_f1_score: 0.8840\n"
     ]
    }
   ],
   "source": [
    "resnet50_val_precision_score = round(history.history['val_precision_score'][-1] * 100, 2)\n",
    "resnet50_val_recall_score = round(history.history['val_recall_score'][-1] * 100, 2)\n",
    "resnet50_val_f1_score = round(history.history['val_f1_score'][-1] * 100, 2)\n",
    "\n",
    "display(Markdown(\"<span style='color:blue'>**Resposta:**</span>\"))\n",
    "print(\"Precisão (precision): \" + str(resnet50_val_precision_score) + \"%\")\n",
    "print(\"Sensibilidade (recall): \" + str(resnet50_val_recall_score) + \"%\")\n",
    "print(\"Pontuação de Validação (f1 score): \" + str(resnet50_val_f1_score) + \"%\")\n",
    "print()\n",
    "print(\"Conforme a saída abaixo, onde é considerado apenas a última época:\")\n",
    "saida1.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.5.2 Arquitetura VGG16"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Explique como é constituída a arquitetura do VGG16? *Utilize, se necessário, gráficos, projetos que utilizam essa arquitetura. Detalhe também sua topologia em camadas e mostre quais as situações essa arquitetura pode ter mais êxito e quais cenários não tem.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A VGG16\n",
    "\n",
    "\n",
    "A arquitetura do VGG16 são pilhas de camadas convolucionais, onde utiliza filtros com um pequeno campo receptivo de 3x3 que é o menor tamanho para capturar a noção de esquerda / direita, cima / baixo, centro.\n",
    "\n",
    "As pilhas de camadas convolucionais são seguidas por três camadas totalmente conectadas, onde a configuração das mesmas são iguais em todas as redes.\n",
    "\n",
    "Na arquitetura VGG16 são 16 camadas de pesos.\n",
    "\n",
    "A Arquitetura pode ser melhor entendida pela figura abaixo:\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p align=\"center\">\n",
    "<img src=\"imagens/VGG16_Summary-Table.jpg\" height=\"50%\" width=\"50%\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prós:  Assim como no ResNet50, modelos pré-treinados como o VGG16, tem a vantagem de excluirmos a camada densa e assim usar as novas classes para Transfer Learning."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Contras: É um modelo pesado e exige mais poder computacional. Talvez mais interessante para se utilizar de menos épocas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Referências:\n",
    "#### https://engmrk.com/vgg16-implementation-using-keras/\n",
    "#### https://towardsdatascience.com/step-by-step-vgg16-implementation-in-keras-for-beginners-a833c686ae6c"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "conv_base = VGG16(include_top=False)\n",
    "\n",
    "for layer in conv_base.layers:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_3\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_6 (InputLayer)         [(None, None, None, 3)]   0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, None, None, 64)    1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, None, None, 64)    36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, None, None, 64)    0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, None, None, 128)   73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, None, None, 128)   147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, None, None, 128)   0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, None, None, 256)   295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, None, None, 256)   590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, None, None, 256)   590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, None, None, 256)   0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, None, None, 512)   1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, None, None, 512)   0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, None, None, 512)   0         \n",
      "_________________________________________________________________\n",
      "global_average_pooling2d_1 ( (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 128)               65664     \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 3)                 387       \n",
      "=================================================================\n",
      "Total params: 14,780,739\n",
      "Trainable params: 66,051\n",
      "Non-trainable params: 14,714,688\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "x = conv_base.output\n",
    "x = layers.GlobalAveragePooling2D()(x)\n",
    "x = layers.Dense(128, activation='relu')(x) \n",
    "\n",
    "predictions = layers.Dense(qtde_classes, activation='softmax')(x)\n",
    "model = Model(conv_base.input, predictions)\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = keras.optimizers.Adam()\n",
    "model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=[precision_score, recall_score, f1_score])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-20-68139f1e6072>:5: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use Model.fit, which supports generators.\n"
     ]
    }
   ],
   "source": [
    "%%capture saida2 --no-stderr\n",
    "history = model.fit_generator(generator=train_generator, \n",
    "                              epochs=qtde_epocas, \n",
    "                              validation_steps=5, \n",
    "                              steps_per_epoch=5, \n",
    "                              validation_data=val_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 2.5626 - precision_score: 0.4938 - recall_score: 0.4938 - f1_score: 0.4937 - val_loss: 1.7071 - val_precision_score: 0.6687 - val_recall_score: 0.6687 - val_f1_score: 0.6687\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 41s 8s/step - loss: 1.4395 - precision_score: 0.7437 - recall_score: 0.7437 - f1_score: 0.7437 - val_loss: 1.0912 - val_precision_score: 0.7937 - val_recall_score: 0.7937 - val_f1_score: 0.7937\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 42s 8s/step - loss: 1.1549 - precision_score: 0.8188 - recall_score: 0.8188 - f1_score: 0.8187 - val_loss: 1.3769 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 42s 8s/step - loss: 0.6415 - precision_score: 0.8052 - recall_score: 0.8000 - f1_score: 0.8026 - val_loss: 0.5911 - val_precision_score: 0.8158 - val_recall_score: 0.7812 - val_f1_score: 0.7979\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 44s 9s/step - loss: 0.6562 - precision_score: 0.8359 - recall_score: 0.8250 - f1_score: 0.8304 - val_loss: 0.5798 - val_precision_score: 0.8320 - val_recall_score: 0.8062 - val_f1_score: 0.8188\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.3915 - precision_score: 0.9036 - recall_score: 0.8813 - f1_score: 0.8921 - val_loss: 0.4674 - val_precision_score: 0.8861 - val_recall_score: 0.8813 - val_f1_score: 0.8836\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4117 - precision_score: 0.8938 - recall_score: 0.8938 - f1_score: 0.8937 - val_loss: 0.5985 - val_precision_score: 0.8371 - val_recall_score: 0.8313 - val_f1_score: 0.8341\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4715 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.5315 - val_precision_score: 0.8403 - val_recall_score: 0.8188 - val_f1_score: 0.8293\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.3804 - precision_score: 0.8650 - recall_score: 0.8500 - f1_score: 0.8573 - val_loss: 0.5929 - val_precision_score: 0.8348 - val_recall_score: 0.8188 - val_f1_score: 0.8266\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4168 - precision_score: 0.8855 - recall_score: 0.8750 - f1_score: 0.8802 - val_loss: 0.5149 - val_precision_score: 0.8554 - val_recall_score: 0.8500 - val_f1_score: 0.8527\n"
     ]
    }
   ],
   "source": [
    "saida2.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Um modelo que converge bem possui o gráfico de perda (*loss*) descendente e os gráfico de precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*) em acendente."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Exibindo dados de Precisão\n",
    "plt.plot(history.history['precision_score'])\n",
    "plt.plot(history.history['val_precision_score'])\n",
    "plt.title('model precision')\n",
    "plt.ylabel('precision')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Sensibilidade\n",
    "plt.plot(history.history['recall_score'])\n",
    "plt.plot(history.history['val_recall_score'])\n",
    "plt.title('model recall')\n",
    "plt.ylabel('recall')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de F1 Score\n",
    "plt.plot(history.history['f1_score'])\n",
    "plt.plot(history.history['val_f1_score'])\n",
    "plt.title('model f1_score')\n",
    "plt.ylabel('f1_score')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Perda\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Avalie os gráficos de perda (*loss*), precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*)  e explique o comportamento de ambos no que tange a convergência do modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Testamos com 2 épocas, 4 épocas, 8 épocas e por fim 10 épocas. A que mostrou uma convergência mais consistente foi justamente esta última.\n",
    "\n",
    "### É possível verificar que existe uma ascendência dos valores de precisão, sensibilidade e pontuação f1 nos gráficos de treino e que tem uma proximidade com os dados de testes, além de percebermos uma estabilização nas últimas épocas. Comparando com a ResNet50, a VGG16 só mostrou essa estabilização maior à partir da 6a época. No caso do gráfico de perda, podemos perceber que é descendente, atingindo estabilidade nas últimas épocas também."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Quais são os valores de **precisão (precision)**, **sensibilidade (recall)** de validação? \n",
    "\n",
    "*Estes valores são exibidos durante o treinamento, utilize a última saída, exemplo:*\n",
    "\n",
    "```\n",
    "Epoch 10/10 [==============================] - 45s 9s/step - loss: 0.1234 - precision_score: 0.9742 - recall_score: 0.9683 - f1_score: 0.9712 - val_loss: 0.8819 - val_precision_score: 0.6912 - val_recall_score: 0.5649 - val_f1_score: 0.6216```\n",
    "\n",
    "No caso acima, o valor de precisão, sensibilidade e pontuação de validação são, respectivamente, 69,12%, 56,49% e 62,16%."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<span style='color:blue'>**Resposta:**</span>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precisão (precision): 85.54%\n",
      "Sensibilidade (recall): 85.0%\n",
      "Pontuação de Validação (f1 score): 85.27%\n",
      "\n",
      "Conforme a saída abaixo, onde é considerado apenas a última época:\n",
      "Epoch 1/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 2.5626 - precision_score: 0.4938 - recall_score: 0.4938 - f1_score: 0.4937 - val_loss: 1.7071 - val_precision_score: 0.6687 - val_recall_score: 0.6687 - val_f1_score: 0.6687\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 41s 8s/step - loss: 1.4395 - precision_score: 0.7437 - recall_score: 0.7437 - f1_score: 0.7437 - val_loss: 1.0912 - val_precision_score: 0.7937 - val_recall_score: 0.7937 - val_f1_score: 0.7937\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 42s 8s/step - loss: 1.1549 - precision_score: 0.8188 - recall_score: 0.8188 - f1_score: 0.8187 - val_loss: 1.3769 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 42s 8s/step - loss: 0.6415 - precision_score: 0.8052 - recall_score: 0.8000 - f1_score: 0.8026 - val_loss: 0.5911 - val_precision_score: 0.8158 - val_recall_score: 0.7812 - val_f1_score: 0.7979\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 44s 9s/step - loss: 0.6562 - precision_score: 0.8359 - recall_score: 0.8250 - f1_score: 0.8304 - val_loss: 0.5798 - val_precision_score: 0.8320 - val_recall_score: 0.8062 - val_f1_score: 0.8188\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.3915 - precision_score: 0.9036 - recall_score: 0.8813 - f1_score: 0.8921 - val_loss: 0.4674 - val_precision_score: 0.8861 - val_recall_score: 0.8813 - val_f1_score: 0.8836\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4117 - precision_score: 0.8938 - recall_score: 0.8938 - f1_score: 0.8937 - val_loss: 0.5985 - val_precision_score: 0.8371 - val_recall_score: 0.8313 - val_f1_score: 0.8341\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4715 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.5315 - val_precision_score: 0.8403 - val_recall_score: 0.8188 - val_f1_score: 0.8293\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.3804 - precision_score: 0.8650 - recall_score: 0.8500 - f1_score: 0.8573 - val_loss: 0.5929 - val_precision_score: 0.8348 - val_recall_score: 0.8188 - val_f1_score: 0.8266\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 43s 9s/step - loss: 0.4168 - precision_score: 0.8855 - recall_score: 0.8750 - f1_score: 0.8802 - val_loss: 0.5149 - val_precision_score: 0.8554 - val_recall_score: 0.8500 - val_f1_score: 0.8527\n"
     ]
    }
   ],
   "source": [
    "vgg16_val_precision_score = round(history.history['val_precision_score'][-1] * 100, 2)\n",
    "vgg16_val_recall_score = round(history.history['val_recall_score'][-1] * 100, 2)\n",
    "vgg16_val_f1_score = round(history.history['val_f1_score'][-1] * 100, 2)\n",
    "\n",
    "display(Markdown(\"<span style='color:blue'>**Resposta:**</span>\"))\n",
    "print(\"Precisão (precision): \" + str(vgg16_val_precision_score) + \"%\")\n",
    "print(\"Sensibilidade (recall): \" + str(vgg16_val_recall_score) + \"%\")\n",
    "print(\"Pontuação de Validação (f1 score): \" + str(vgg16_val_f1_score) + \"%\")\n",
    "print()\n",
    "print(\"Conforme a saída abaixo, onde é considerado apenas a última época:\")\n",
    "saida2.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.5.3 Arquitetura VGG19"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Explique como é constituída a arquitetura do VGG19? *Utilize, se necessário, gráficos, projetos que utilizam essa arquitetura. Detalhe também sua topologia em camadas e mostre quais as situações essa arquitetura pode ter mais êxito e quais cenários não tem.*"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### A VGG19\n",
    "\n",
    "A arquitetura do VGG19 é semelhante a VGG16. No geral possui camadas convolucionais, onde utiliza filtros com um pequeno campo receptivo de 3x3 que é o menor tamanho para capturar a noção de esquerda / direita, cima / baixo, centro.\n",
    "\n",
    "VGG19 é uma variância do modelo VGG e consiste em 19 camadas (16 convolucionais, 3 totalmente conectadas, 5 de MaxPool e 1 SoftMax). Tem 19.6 bilhões de FLOPs.\n",
    "\n",
    "A Arquitetura pode ser melhor entendida pela figura abaixo:\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<p align=\"center\">\n",
    "<img src=\"imagens/VGG19-architecture.png\" height=\"30%\" width=\"30%\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prós:  Assim como nas anteriores ResNet50 e VGG16, modelos pré-treinados como o VGG19, tem a vantagem de excluirmos a camada densa e assim usar as novas classes para Transfer Learning."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Contras: Assim como a VGG16 é um modelo pesado e exige mais poder computacional. Talvez mais interessante para se utilizar com menos épocas."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Referências:\n",
    "#### https://iq.opengenus.org/vgg19-architecture/"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "conv_base = VGG19(include_top=False)\n",
    "\n",
    "for layer in conv_base.layers:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_7\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_8 (InputLayer)         [(None, None, None, 3)]   0         \n",
      "_________________________________________________________________\n",
      "block1_conv1 (Conv2D)        (None, None, None, 64)    1792      \n",
      "_________________________________________________________________\n",
      "block1_conv2 (Conv2D)        (None, None, None, 64)    36928     \n",
      "_________________________________________________________________\n",
      "block1_pool (MaxPooling2D)   (None, None, None, 64)    0         \n",
      "_________________________________________________________________\n",
      "block2_conv1 (Conv2D)        (None, None, None, 128)   73856     \n",
      "_________________________________________________________________\n",
      "block2_conv2 (Conv2D)        (None, None, None, 128)   147584    \n",
      "_________________________________________________________________\n",
      "block2_pool (MaxPooling2D)   (None, None, None, 128)   0         \n",
      "_________________________________________________________________\n",
      "block3_conv1 (Conv2D)        (None, None, None, 256)   295168    \n",
      "_________________________________________________________________\n",
      "block3_conv2 (Conv2D)        (None, None, None, 256)   590080    \n",
      "_________________________________________________________________\n",
      "block3_conv3 (Conv2D)        (None, None, None, 256)   590080    \n",
      "_________________________________________________________________\n",
      "block3_conv4 (Conv2D)        (None, None, None, 256)   590080    \n",
      "_________________________________________________________________\n",
      "block3_pool (MaxPooling2D)   (None, None, None, 256)   0         \n",
      "_________________________________________________________________\n",
      "block4_conv1 (Conv2D)        (None, None, None, 512)   1180160   \n",
      "_________________________________________________________________\n",
      "block4_conv2 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv3 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block4_conv4 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block4_pool (MaxPooling2D)   (None, None, None, 512)   0         \n",
      "_________________________________________________________________\n",
      "block5_conv1 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv2 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv3 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_conv4 (Conv2D)        (None, None, None, 512)   2359808   \n",
      "_________________________________________________________________\n",
      "block5_pool (MaxPooling2D)   (None, None, None, 512)   0         \n",
      "_________________________________________________________________\n",
      "global_average_pooling2d_3 ( (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 128)               65664     \n",
      "_________________________________________________________________\n",
      "dense_7 (Dense)              (None, 3)                 387       \n",
      "=================================================================\n",
      "Total params: 20,090,435\n",
      "Trainable params: 66,051\n",
      "Non-trainable params: 20,024,384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "x = conv_base.output\n",
    "x = layers.GlobalAveragePooling2D()(x)\n",
    "x = layers.Dense(128, activation='relu')(x) \n",
    "\n",
    "predictions = layers.Dense(qtde_classes, activation='softmax')(x)\n",
    "model = Model(conv_base.input, predictions)\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = keras.optimizers.Adam()\n",
    "model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=[precision_score, recall_score, f1_score])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [],
   "source": [
    "%%capture saida3 --no-stderr\n",
    "history = model.fit_generator(generator=train_generator, \n",
    "                              epochs=qtde_epocas, \n",
    "                              validation_steps=5, \n",
    "                              steps_per_epoch=5, \n",
    "                              validation_data=val_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 2.6404 - precision_score: 0.5777 - recall_score: 0.5625 - f1_score: 0.5696 - val_loss: 1.1742 - val_precision_score: 0.7625 - val_recall_score: 0.7625 - val_f1_score: 0.7625\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 1.0458 - precision_score: 0.7812 - recall_score: 0.7812 - f1_score: 0.7812 - val_loss: 0.7240 - val_precision_score: 0.8625 - val_recall_score: 0.8625 - val_f1_score: 0.8625\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5289 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.7872 - val_precision_score: 0.8625 - val_recall_score: 0.8625 - val_f1_score: 0.8625\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5357 - precision_score: 0.8554 - recall_score: 0.8500 - f1_score: 0.8527 - val_loss: 0.4540 - val_precision_score: 0.8540 - val_recall_score: 0.8438 - val_f1_score: 0.8488\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.4084 - precision_score: 0.8746 - recall_score: 0.8687 - f1_score: 0.8716 - val_loss: 0.4882 - val_precision_score: 0.8177 - val_recall_score: 0.7937 - val_f1_score: 0.8054\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5022 - precision_score: 0.8625 - recall_score: 0.8625 - f1_score: 0.8625 - val_loss: 0.4856 - val_precision_score: 0.8540 - val_recall_score: 0.8438 - val_f1_score: 0.8488\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5050 - precision_score: 0.8459 - recall_score: 0.8250 - f1_score: 0.8352 - val_loss: 0.4833 - val_precision_score: 0.8220 - val_recall_score: 0.8125 - val_f1_score: 0.8172\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.4104 - precision_score: 0.8475 - recall_score: 0.8375 - f1_score: 0.8423 - val_loss: 0.3776 - val_precision_score: 0.8990 - val_recall_score: 0.8938 - val_f1_score: 0.8963\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.2939 - precision_score: 0.9312 - recall_score: 0.9250 - f1_score: 0.9281 - val_loss: 0.4551 - val_precision_score: 0.8992 - val_recall_score: 0.8938 - val_f1_score: 0.8964\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.3807 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.4567 - val_precision_score: 0.8425 - val_recall_score: 0.8375 - val_f1_score: 0.8400\n"
     ]
    }
   ],
   "source": [
    "saida3.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Um modelo que converge bem possui o gráfico de perda (*loss*) descendente e os gráfico de precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*) em acendente."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Exibindo dados de Precisão\n",
    "plt.plot(history.history['precision_score'])\n",
    "plt.plot(history.history['val_precision_score'])\n",
    "plt.title('model precision')\n",
    "plt.ylabel('precision')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Sensibilidade\n",
    "plt.plot(history.history['recall_score'])\n",
    "plt.plot(history.history['val_recall_score'])\n",
    "plt.title('model recall')\n",
    "plt.ylabel('recall')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de F1 Score\n",
    "plt.plot(history.history['f1_score'])\n",
    "plt.plot(history.history['val_f1_score'])\n",
    "plt.title('model f1_score')\n",
    "plt.ylabel('f1_score')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Perda\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Avalie os gráficos de perda (*loss*), precisão (*precision*), sensibilidade (*recall*) e pontuação f1 (*f1 score*)  e explique o comportamento de ambos no que tange a convergência do modelo."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Para este caso, testamos com 10 épocas diretamente. Mostrou também uma convergência mais consistente até a 8a época, quando passou ter uma variação maior, mas manteve-se com precisão e f1 acima de 70%.\n",
    "\n",
    "### Comparando com a ResNet50 e a VGG16, a VGG19 mostrou um erro maior na validação, comparando os gráficos de Treino e Teste, mas assim como os demais, também mostra uma tendência de ascendência para precisão, sensibilidade e pontuação f1 e descendência para o gráfico de perda."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Pergunta**: Quais são os valores de **precisão (precision)**, **sensibilidade (recall)** de validação? \n",
    "\n",
    "*Estes valores são exibidos durante o treinamento, utilize a última saída, exemplo:*\n",
    "\n",
    "```\n",
    "Epoch 10/10 [==============================] - 45s 9s/step - loss: 0.1234 - precision_score: 0.9742 - recall_score: 0.9683 - f1_score: 0.9712 - val_loss: 0.8819 - val_precision_score: 0.6912 - val_recall_score: 0.5649 - val_f1_score: 0.6216```\n",
    "\n",
    "No caso acima, o valor de precisão, sensibilidade e pontuação de validação são, respectivamente, 69,12%, 56,49% e 62,16%."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<span style='color:blue'>**Resposta:**</span>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precisão (precision): 84.25%\n",
      "Sensibilidade (recall): 83.75%\n",
      "Pontuação de Validação (f1 score): 84.0%\n",
      "\n",
      "Conforme a saída abaixo, onde é considerado apenas a última época:\n",
      "Epoch 1/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 2.6404 - precision_score: 0.5777 - recall_score: 0.5625 - f1_score: 0.5696 - val_loss: 1.1742 - val_precision_score: 0.7625 - val_recall_score: 0.7625 - val_f1_score: 0.7625\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 1.0458 - precision_score: 0.7812 - recall_score: 0.7812 - f1_score: 0.7812 - val_loss: 0.7240 - val_precision_score: 0.8625 - val_recall_score: 0.8625 - val_f1_score: 0.8625\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5289 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.7872 - val_precision_score: 0.8625 - val_recall_score: 0.8625 - val_f1_score: 0.8625\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5357 - precision_score: 0.8554 - recall_score: 0.8500 - f1_score: 0.8527 - val_loss: 0.4540 - val_precision_score: 0.8540 - val_recall_score: 0.8438 - val_f1_score: 0.8488\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.4084 - precision_score: 0.8746 - recall_score: 0.8687 - f1_score: 0.8716 - val_loss: 0.4882 - val_precision_score: 0.8177 - val_recall_score: 0.7937 - val_f1_score: 0.8054\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5022 - precision_score: 0.8625 - recall_score: 0.8625 - f1_score: 0.8625 - val_loss: 0.4856 - val_precision_score: 0.8540 - val_recall_score: 0.8438 - val_f1_score: 0.8488\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.5050 - precision_score: 0.8459 - recall_score: 0.8250 - f1_score: 0.8352 - val_loss: 0.4833 - val_precision_score: 0.8220 - val_recall_score: 0.8125 - val_f1_score: 0.8172\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.4104 - precision_score: 0.8475 - recall_score: 0.8375 - f1_score: 0.8423 - val_loss: 0.3776 - val_precision_score: 0.8990 - val_recall_score: 0.8938 - val_f1_score: 0.8963\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.2939 - precision_score: 0.9312 - recall_score: 0.9250 - f1_score: 0.9281 - val_loss: 0.4551 - val_precision_score: 0.8992 - val_recall_score: 0.8938 - val_f1_score: 0.8964\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 53s 11s/step - loss: 0.3807 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 0.4567 - val_precision_score: 0.8425 - val_recall_score: 0.8375 - val_f1_score: 0.8400\n"
     ]
    }
   ],
   "source": [
    "vgg19_val_precision_score = round(history.history['val_precision_score'][-1] * 100, 2)\n",
    "vgg19_val_recall_score = round(history.history['val_recall_score'][-1] * 100, 2)\n",
    "vgg19_val_f1_score = round(history.history['val_f1_score'][-1] * 100, 2)\n",
    "\n",
    "display(Markdown(\"<span style='color:blue'>**Resposta:**</span>\"))\n",
    "print(\"Precisão (precision): \" + str(vgg19_val_precision_score) + \"%\")\n",
    "print(\"Sensibilidade (recall): \" + str(vgg19_val_recall_score) + \"%\")\n",
    "print(\"Pontuação de Validação (f1 score): \" + str(vgg19_val_f1_score) + \"%\")\n",
    "print()\n",
    "print(\"Conforme a saída abaixo, onde é considerado apenas a última época:\")\n",
    "saida3.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### Apesar do valor de f1-score já ser dado pela função, vamos usar o cálculo do código abaixo, para demonstrar os valores no item 4.6, conforme fórmula descrita neste item, sendo dado por:\n",
    "_2 * (Precisão * Sensibilidade) / (Precisão + Sensibilidade)._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "\n",
    "import pandas as pd\n",
    "\n",
    "resnet50_f1_score_calc = round((2*resnet50_val_precision_score*resnet50_val_recall_score)/(resnet50_val_precision_score+resnet50_val_recall_score), 2)\n",
    "\n",
    "vgg16_f1_score_calc = round((2*vgg16_val_precision_score*vgg16_val_recall_score)/(vgg16_val_precision_score+vgg16_val_recall_score),2)\n",
    "\n",
    "vgg19_f1_score_calc = round((2*vgg19_val_precision_score*vgg19_val_recall_score)/(vgg19_val_precision_score+vgg19_val_recall_score),2)\n",
    "\n",
    "Comparativo_Arquiteturas = pd.DataFrame(data=(['ResNet50', str(resnet50_val_precision_score) + ' %', str(resnet50_val_recall_score) + ' %', str(resnet50_val_f1_score) + ' %', str(resnet50_f1_score_calc) + ' %'],\n",
    "                                       ['VGG16', str(vgg16_val_precision_score) + ' %', str(vgg16_val_recall_score) + ' %', str(vgg16_val_f1_score) + ' %', str(vgg16_f1_score_calc) + ' %'],\n",
    "                                       ['VGG19', str(vgg19_val_precision_score) + ' %', str(vgg19_val_recall_score) + ' %', str(vgg19_val_f1_score) + ' %', str(vgg19_f1_score_calc) + ' %']),\n",
    "                                        columns=['Modelo', 'Precisão (Precision)', 'Sensibilidade (Recall)', 'F1-Score', 'F1-Score Calculado'])\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.6 Compartivo de arquiteturas\n",
    "\n",
    "Preencha a tabela abaixo com os valores dos indicadores de performance apresentados.\n",
    "\n",
    "_O cálculo do F1-Score é dado por 2 * (Precisão * Sensibilidade) / (Precisão + Sensibilidade)._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Modelo</th>\n",
       "      <th>Precisão (Precision)</th>\n",
       "      <th>Sensibilidade (Recall)</th>\n",
       "      <th>F1-Score</th>\n",
       "      <th>F1-Score Calculado</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ResNet50</td>\n",
       "      <td>88.69 %</td>\n",
       "      <td>88.13 %</td>\n",
       "      <td>88.4 %</td>\n",
       "      <td>88.41 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>VGG16</td>\n",
       "      <td>85.54 %</td>\n",
       "      <td>85.0 %</td>\n",
       "      <td>85.27 %</td>\n",
       "      <td>85.27 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VGG19</td>\n",
       "      <td>84.25 %</td>\n",
       "      <td>83.75 %</td>\n",
       "      <td>84.0 %</td>\n",
       "      <td>84.0 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Modelo Precisão (Precision) Sensibilidade (Recall) F1-Score  \\\n",
       "0  ResNet50              88.69 %                88.13 %   88.4 %   \n",
       "1     VGG16              85.54 %                 85.0 %  85.27 %   \n",
       "2     VGG19              84.25 %                83.75 %   84.0 %   \n",
       "\n",
       "  F1-Score Calculado  \n",
       "0            88.41 %  \n",
       "1            85.27 %  \n",
       "2             84.0 %  "
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Comparativo_Arquiteturas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.7 Conclusões\n",
    "\n",
    "Analise os resultados da tabela de indicadores do comparativo de arquiteturas e explique os principais motivos pelos quais cada modelo obteve cada resultado."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conclusão\n",
    "\n",
    "Nos 3 modelos propostos, o que demonstrou um melhor resultado e consistência for o ResNet50. Apesar de todos terem apresentado resultados acima de 70%, rodamos várias vezes o ResNet50 e o resultado de precisão, sensibilidade e f1-score do modelo foram sempre muito parecidos, diferente do VGG, que várias vezes não conseguímos concluir as 10 épocas, com estouro de thread do computador.\n",
    "\n",
    "Mantivemos a quantidade original de imagens, que é bem grande por sinal, mas alteramos a validação para 35%. Após algumas tentativas e fechando todos os aplicativos do computador (i7 + 16 GB RAM), conseguimos rodar com êxito todos eles e com precisão acima dos 70%.\n",
    "\n",
    "Por ser um modelo mais leve, para esse caso do raio-x a proposição é usar o ResNet50."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.8 Abordagem Extra\n",
    "\n",
    "Considerando os outros classificadores, escolha outro que ainda não foi utilizado, implemente abaixo. Ao final compare os resultados e explique os resultados.\n",
    "\n",
    "_Não se esquece de utilizar as importações adequadas para cada modelo.\n",
    "A forma de implementação deve respeitar as mesmas condições como valor de split e quantidade de imagens para poder comparar os modelos._"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "#IMPLEMENTE\n",
    "from keras.applications import xception\n",
    "from keras.applications.xception import preprocess_input\n",
    "\n",
    "from keras import Model, layers\n",
    "from keras.models import load_model, model_from_json\n",
    "\n",
    "import keras.backend as K"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Vamos usar o modelo Xception, pois tem um depth bem maior que o VGG16 e VGG19, mas com um tamanho bem menor, menor inclusive que o Resnet50.\n",
    "\n",
    "### Conforme figura abaixo:\n",
    "\n",
    "<p align=\"center\">\n",
    "<img src=\"imagens/keras_modelos.png\" height=\"20%\" width=\"20%\">\n",
    "</p>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "tags": []
   },
   "outputs": [],
   "source": [
    "conv_base = Xception(include_top=False)\n",
    "\n",
    "for layer in conv_base.layers:\n",
    "    layer.trainable = False"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {
    "tags": [
     "outputPrepend"
    ]
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"functional_3\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            [(None, None, None,  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv1 (Conv2D)           (None, None, None, 3 864         input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv1_bn (BatchNormaliza (None, None, None, 3 128         block1_conv1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv1_act (Activation)   (None, None, None, 3 0           block1_conv1_bn[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv2 (Conv2D)           (None, None, None, 6 18432       block1_conv1_act[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv2_bn (BatchNormaliza (None, None, None, 6 256         block1_conv2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "block1_conv2_act (Activation)   (None, None, None, 6 0           block1_conv2_bn[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block2_sepconv1 (SeparableConv2 (None, None, None, 1 8768        block1_conv2_act[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block2_sepconv1_bn (BatchNormal (None, None, None, 1 512         block2_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block2_sepconv2_act (Activation (None, None, None, 1 0           block2_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block2_sepconv2 (SeparableConv2 (None, None, None, 1 17536       block2_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block2_sepconv2_bn (BatchNormal (None, None, None, 1 512         block2_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_4 (Conv2D)               (None, None, None, 1 8192        block1_conv2_act[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block2_pool (MaxPooling2D)      (None, None, None, 1 0           block2_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_4 (BatchNor (None, None, None, 1 512         conv2d_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "add_12 (Add)                    (None, None, None, 1 0           block2_pool[0][0]                \n",
      "                                                                 batch_normalization_4[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv1_act (Activation (None, None, None, 1 0           add_12[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv1 (SeparableConv2 (None, None, None, 2 33920       block3_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv1_bn (BatchNormal (None, None, None, 2 1024        block3_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv2_act (Activation (None, None, None, 2 0           block3_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv2 (SeparableConv2 (None, None, None, 2 67840       block3_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block3_sepconv2_bn (BatchNormal (None, None, None, 2 1024        block3_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_5 (Conv2D)               (None, None, None, 2 32768       add_12[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block3_pool (MaxPooling2D)      (None, None, None, 2 0           block3_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_5 (BatchNor (None, None, None, 2 1024        conv2d_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "add_13 (Add)                    (None, None, None, 2 0           block3_pool[0][0]                \n",
      "                                                                 batch_normalization_5[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv1_act (Activation (None, None, None, 2 0           add_13[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv1 (SeparableConv2 (None, None, None, 7 188672      block4_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block4_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv2_act (Activation (None, None, None, 7 0           block4_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block4_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block4_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block4_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_6 (Conv2D)               (None, None, None, 7 186368      add_13[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block4_pool (MaxPooling2D)      (None, None, None, 7 0           block4_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_6 (BatchNor (None, None, None, 7 2912        conv2d_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "add_14 (Add)                    (None, None, None, 7 0           block4_pool[0][0]                \n",
      "                                                                 batch_normalization_6[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv1_act (Activation (None, None, None, 7 0           add_14[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv1 (SeparableConv2 (None, None, None, 7 536536      block5_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block5_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv2_act (Activation (None, None, None, 7 0           block5_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block5_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block5_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv3_act (Activation (None, None, None, 7 0           block5_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv3 (SeparableConv2 (None, None, None, 7 536536      block5_sepconv3_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block5_sepconv3_bn (BatchNormal (None, None, None, 7 2912        block5_sepconv3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_15 (Add)                    (None, None, None, 7 0           block5_sepconv3_bn[0][0]         \n",
      "                                                                 add_14[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv1_act (Activation (None, None, None, 7 0           add_15[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv1 (SeparableConv2 (None, None, None, 7 536536      block6_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block6_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv2_act (Activation (None, None, None, 7 0           block6_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block6_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block6_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv3_act (Activation (None, None, None, 7 0           block6_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv3 (SeparableConv2 (None, None, None, 7 536536      block6_sepconv3_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block6_sepconv3_bn (BatchNormal (None, None, None, 7 2912        block6_sepconv3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_16 (Add)                    (None, None, None, 7 0           block6_sepconv3_bn[0][0]         \n",
      "                                                                 add_15[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv1_act (Activation (None, None, None, 7 0           add_16[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv1 (SeparableConv2 (None, None, None, 7 536536      block7_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block7_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv2_act (Activation (None, None, None, 7 0           block7_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block7_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block7_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv3_act (Activation (None, None, None, 7 0           block7_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv3 (SeparableConv2 (None, None, None, 7 536536      block7_sepconv3_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block7_sepconv3_bn (BatchNormal (None, None, None, 7 2912        block7_sepconv3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_17 (Add)                    (None, None, None, 7 0           block7_sepconv3_bn[0][0]         \n",
      "                                                                 add_16[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv1_act (Activation (None, None, None, 7 0           add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv1 (SeparableConv2 (None, None, None, 7 536536      block8_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block8_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv2_act (Activation (None, None, None, 7 0           block8_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block8_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block8_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv3_act (Activation (None, None, None, 7 0           block8_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv3 (SeparableConv2 (None, None, None, 7 536536      block8_sepconv3_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block8_sepconv3_bn (BatchNormal (None, None, None, 7 2912        block8_sepconv3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_18 (Add)                    (None, None, None, 7 0           block8_sepconv3_bn[0][0]         \n",
      "                                                                 add_17[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv1_act (Activation (None, None, None, 7 0           add_18[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv1 (SeparableConv2 (None, None, None, 7 536536      block9_sepconv1_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv1_bn (BatchNormal (None, None, None, 7 2912        block9_sepconv1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv2_act (Activation (None, None, None, 7 0           block9_sepconv1_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv2 (SeparableConv2 (None, None, None, 7 536536      block9_sepconv2_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv2_bn (BatchNormal (None, None, None, 7 2912        block9_sepconv2[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv3_act (Activation (None, None, None, 7 0           block9_sepconv2_bn[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv3 (SeparableConv2 (None, None, None, 7 536536      block9_sepconv3_act[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block9_sepconv3_bn (BatchNormal (None, None, None, 7 2912        block9_sepconv3[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_19 (Add)                    (None, None, None, 7 0           block9_sepconv3_bn[0][0]         \n",
      "                                                                 add_18[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv1_act (Activatio (None, None, None, 7 0           add_19[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv1 (SeparableConv (None, None, None, 7 536536      block10_sepconv1_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv1_bn (BatchNorma (None, None, None, 7 2912        block10_sepconv1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv2_act (Activatio (None, None, None, 7 0           block10_sepconv1_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv2 (SeparableConv (None, None, None, 7 536536      block10_sepconv2_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv2_bn (BatchNorma (None, None, None, 7 2912        block10_sepconv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv3_act (Activatio (None, None, None, 7 0           block10_sepconv2_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv3 (SeparableConv (None, None, None, 7 536536      block10_sepconv3_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block10_sepconv3_bn (BatchNorma (None, None, None, 7 2912        block10_sepconv3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "add_20 (Add)                    (None, None, None, 7 0           block10_sepconv3_bn[0][0]        \n",
      "                                                                 add_19[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv1_act (Activatio (None, None, None, 7 0           add_20[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv1 (SeparableConv (None, None, None, 7 536536      block11_sepconv1_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv1_bn (BatchNorma (None, None, None, 7 2912        block11_sepconv1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv2_act (Activatio (None, None, None, 7 0           block11_sepconv1_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv2 (SeparableConv (None, None, None, 7 536536      block11_sepconv2_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv2_bn (BatchNorma (None, None, None, 7 2912        block11_sepconv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv3_act (Activatio (None, None, None, 7 0           block11_sepconv2_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv3 (SeparableConv (None, None, None, 7 536536      block11_sepconv3_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block11_sepconv3_bn (BatchNorma (None, None, None, 7 2912        block11_sepconv3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "add_21 (Add)                    (None, None, None, 7 0           block11_sepconv3_bn[0][0]        \n",
      "                                                                 add_20[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv1_act (Activatio (None, None, None, 7 0           add_21[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv1 (SeparableConv (None, None, None, 7 536536      block12_sepconv1_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv1_bn (BatchNorma (None, None, None, 7 2912        block12_sepconv1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv2_act (Activatio (None, None, None, 7 0           block12_sepconv1_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv2 (SeparableConv (None, None, None, 7 536536      block12_sepconv2_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv2_bn (BatchNorma (None, None, None, 7 2912        block12_sepconv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv3_act (Activatio (None, None, None, 7 0           block12_sepconv2_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv3 (SeparableConv (None, None, None, 7 536536      block12_sepconv3_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block12_sepconv3_bn (BatchNorma (None, None, None, 7 2912        block12_sepconv3[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "add_22 (Add)                    (None, None, None, 7 0           block12_sepconv3_bn[0][0]        \n",
      "                                                                 add_21[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv1_act (Activatio (None, None, None, 7 0           add_22[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv1 (SeparableConv (None, None, None, 7 536536      block13_sepconv1_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv1_bn (BatchNorma (None, None, None, 7 2912        block13_sepconv1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv2_act (Activatio (None, None, None, 7 0           block13_sepconv1_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv2 (SeparableConv (None, None, None, 1 752024      block13_sepconv2_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block13_sepconv2_bn (BatchNorma (None, None, None, 1 4096        block13_sepconv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_7 (Conv2D)               (None, None, None, 1 745472      add_22[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block13_pool (MaxPooling2D)     (None, None, None, 1 0           block13_sepconv2_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_7 (BatchNor (None, None, None, 1 4096        conv2d_7[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "add_23 (Add)                    (None, None, None, 1 0           block13_pool[0][0]               \n",
      "                                                                 batch_normalization_7[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv1 (SeparableConv (None, None, None, 1 1582080     add_23[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv1_bn (BatchNorma (None, None, None, 1 6144        block14_sepconv1[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv1_act (Activatio (None, None, None, 1 0           block14_sepconv1_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv2 (SeparableConv (None, None, None, 2 3159552     block14_sepconv1_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv2_bn (BatchNorma (None, None, None, 2 8192        block14_sepconv2[0][0]           \n",
      "__________________________________________________________________________________________________\n",
      "block14_sepconv2_act (Activatio (None, None, None, 2 0           block14_sepconv2_bn[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "global_average_pooling2d_1 (Glo (None, 2048)         0           block14_sepconv2_act[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "dense_2 (Dense)                 (None, 128)          262272      global_average_pooling2d_1[0][0] \n",
      "__________________________________________________________________________________________________\n",
      "dense_3 (Dense)                 (None, 3)            387         dense_2[0][0]                    \n",
      "==================================================================================================\n",
      "Total params: 21,124,139\n",
      "Trainable params: 262,659\n",
      "Non-trainable params: 20,861,480\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "x = conv_base.output\n",
    "x = layers.GlobalAveragePooling2D()(x)\n",
    "x = layers.Dense(128, activation='relu')(x) \n",
    "\n",
    "predictions = layers.Dense(qtde_classes, activation='softmax')(x)\n",
    "model = Model(conv_base.input, predictions)\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "optimizer = keras.optimizers.Adam()\n",
    "model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=[precision_score, recall_score, f1_score])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From <ipython-input-18-68139f1e6072>:5: Model.fit_generator (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Please use Model.fit, which supports generators.\n"
     ]
    }
   ],
   "source": [
    "%%capture saida4 --no-stderr\n",
    "history = model.fit_generator(generator=train_generator, \n",
    "                              epochs=qtde_epocas, \n",
    "                              validation_steps=5, \n",
    "                              steps_per_epoch=5, \n",
    "                              validation_data=val_generator)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/10\n",
      "5/5 [==============================] - 22s 4s/step - loss: 34.2798 - precision_score: 0.4563 - recall_score: 0.4563 - f1_score: 0.4562 - val_loss: 15.4155 - val_precision_score: 0.7500 - val_recall_score: 0.7500 - val_f1_score: 0.7500\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 15.1647 - precision_score: 0.7063 - recall_score: 0.7063 - f1_score: 0.7062 - val_loss: 9.7982 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 13.6672 - precision_score: 0.6625 - recall_score: 0.6625 - f1_score: 0.6625 - val_loss: 7.0599 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 6.7960 - precision_score: 0.8000 - recall_score: 0.8000 - f1_score: 0.8000 - val_loss: 10.1895 - val_precision_score: 0.7000 - val_recall_score: 0.7000 - val_f1_score: 0.7000\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 21s 4s/step - loss: 9.3786 - precision_score: 0.7688 - recall_score: 0.7688 - f1_score: 0.7687 - val_loss: 5.4264 - val_precision_score: 0.8438 - val_recall_score: 0.8438 - val_f1_score: 0.8437\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 6.9110 - precision_score: 0.7625 - recall_score: 0.7625 - f1_score: 0.7625 - val_loss: 5.3213 - val_precision_score: 0.7688 - val_recall_score: 0.7688 - val_f1_score: 0.7687\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 4.4100 - precision_score: 0.7875 - recall_score: 0.7875 - f1_score: 0.7875 - val_loss: 3.1550 - val_precision_score: 0.7875 - val_recall_score: 0.7875 - val_f1_score: 0.7875\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 1.8632 - precision_score: 0.8000 - recall_score: 0.8000 - f1_score: 0.8000 - val_loss: 5.0161 - val_precision_score: 0.2688 - val_recall_score: 0.2688 - val_f1_score: 0.2687\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 2.1266 - precision_score: 0.7046 - recall_score: 0.7000 - f1_score: 0.7023 - val_loss: 1.3176 - val_precision_score: 0.8687 - val_recall_score: 0.8687 - val_f1_score: 0.8687\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 1.3689 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 1.5367 - val_precision_score: 0.9000 - val_recall_score: 0.9000 - val_f1_score: 0.9000\n"
     ]
    }
   ],
   "source": [
    "saida4.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Exibindo dados de Precisão\n",
    "plt.plot(history.history['precision_score'])\n",
    "plt.plot(history.history['val_precision_score'])\n",
    "plt.title('model precision')\n",
    "plt.ylabel('precision')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Sensibilidade\n",
    "plt.plot(history.history['recall_score'])\n",
    "plt.plot(history.history['val_recall_score'])\n",
    "plt.title('model recall')\n",
    "plt.ylabel('recall')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de F1 Score\n",
    "plt.plot(history.history['f1_score'])\n",
    "plt.plot(history.history['val_f1_score'])\n",
    "plt.title('model f1_score')\n",
    "plt.ylabel('f1_score')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()\n",
    "\n",
    "# Exibindo dados de Perda\n",
    "plt.plot(history.history['loss'])\n",
    "plt.plot(history.history['val_loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "data": {
      "text/markdown": [
       "<span style='color:blue'>**Resposta:**</span>"
      ],
      "text/plain": [
       "<IPython.core.display.Markdown object>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Precisão (precision): 90.0%\n",
      "Sensibilidade (recall): 90.0%\n",
      "Pontuação de Validação (f1 score): 90.0%\n",
      "\n",
      "Conforme a saída abaixo, onde é considerado apenas a última época:\n",
      "Epoch 1/10\n",
      "5/5 [==============================] - 22s 4s/step - loss: 34.2798 - precision_score: 0.4563 - recall_score: 0.4563 - f1_score: 0.4562 - val_loss: 15.4155 - val_precision_score: 0.7500 - val_recall_score: 0.7500 - val_f1_score: 0.7500\n",
      "Epoch 2/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 15.1647 - precision_score: 0.7063 - recall_score: 0.7063 - f1_score: 0.7062 - val_loss: 9.7982 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 3/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 13.6672 - precision_score: 0.6625 - recall_score: 0.6625 - f1_score: 0.6625 - val_loss: 7.0599 - val_precision_score: 0.7375 - val_recall_score: 0.7375 - val_f1_score: 0.7375\n",
      "Epoch 4/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 6.7960 - precision_score: 0.8000 - recall_score: 0.8000 - f1_score: 0.8000 - val_loss: 10.1895 - val_precision_score: 0.7000 - val_recall_score: 0.7000 - val_f1_score: 0.7000\n",
      "Epoch 5/10\n",
      "5/5 [==============================] - 21s 4s/step - loss: 9.3786 - precision_score: 0.7688 - recall_score: 0.7688 - f1_score: 0.7687 - val_loss: 5.4264 - val_precision_score: 0.8438 - val_recall_score: 0.8438 - val_f1_score: 0.8437\n",
      "Epoch 6/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 6.9110 - precision_score: 0.7625 - recall_score: 0.7625 - f1_score: 0.7625 - val_loss: 5.3213 - val_precision_score: 0.7688 - val_recall_score: 0.7688 - val_f1_score: 0.7687\n",
      "Epoch 7/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 4.4100 - precision_score: 0.7875 - recall_score: 0.7875 - f1_score: 0.7875 - val_loss: 3.1550 - val_precision_score: 0.7875 - val_recall_score: 0.7875 - val_f1_score: 0.7875\n",
      "Epoch 8/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 1.8632 - precision_score: 0.8000 - recall_score: 0.8000 - f1_score: 0.8000 - val_loss: 5.0161 - val_precision_score: 0.2688 - val_recall_score: 0.2688 - val_f1_score: 0.2687\n",
      "Epoch 9/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 2.1266 - precision_score: 0.7046 - recall_score: 0.7000 - f1_score: 0.7023 - val_loss: 1.3176 - val_precision_score: 0.8687 - val_recall_score: 0.8687 - val_f1_score: 0.8687\n",
      "Epoch 10/10\n",
      "5/5 [==============================] - 20s 4s/step - loss: 1.3689 - precision_score: 0.8813 - recall_score: 0.8813 - f1_score: 0.8812 - val_loss: 1.5367 - val_precision_score: 0.9000 - val_recall_score: 0.9000 - val_f1_score: 0.9000\n"
     ]
    }
   ],
   "source": [
    "xception_val_precision_score = round(history.history['val_precision_score'][-1] * 100, 2)\n",
    "xception_val_recall_score = round(history.history['val_recall_score'][-1] * 100, 2)\n",
    "xception_val_f1_score = round(history.history['val_f1_score'][-1] * 100, 2)\n",
    "\n",
    "display(Markdown(\"<span style='color:blue'>**Resposta:**</span>\"))\n",
    "print(\"Precisão (precision): \" + str(xception_val_precision_score) + \"%\")\n",
    "print(\"Sensibilidade (recall): \" + str(xception_val_recall_score) + \"%\")\n",
    "print(\"Pontuação de Validação (f1 score): \" + str(xception_val_f1_score) + \"%\")\n",
    "print()\n",
    "print(\"Conforme a saída abaixo, onde é considerado apenas a última época:\")\n",
    "saida4.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.8.1 Conclusões sobre a abordagem extra\n",
    "\n",
    "Como seu modelo performou em comparação com os demais modelos anteriores? Justifique sua resposta levando em consideração a arquitetura respectiva."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "**Resposta**:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Modelo</th>\n",
       "      <th>Precisão (Precision)</th>\n",
       "      <th>Sensibilidade (Recall)</th>\n",
       "      <th>F1-Score</th>\n",
       "      <th>F1-Score Calculado</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>ResNet50</td>\n",
       "      <td>88.69 %</td>\n",
       "      <td>88.13 %</td>\n",
       "      <td>88.4 %</td>\n",
       "      <td>88.41 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>VGG16</td>\n",
       "      <td>85.54 %</td>\n",
       "      <td>85.0 %</td>\n",
       "      <td>85.27 %</td>\n",
       "      <td>85.27 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>VGG19</td>\n",
       "      <td>84.25 %</td>\n",
       "      <td>83.75 %</td>\n",
       "      <td>84.0 %</td>\n",
       "      <td>84.0 %</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Xception</td>\n",
       "      <td>90.0 %</td>\n",
       "      <td>90.0 %</td>\n",
       "      <td>90.0 %</td>\n",
       "      <td>90.0 %</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "     Modelo Precisão (Precision) Sensibilidade (Recall) F1-Score  \\\n",
       "0  ResNet50              88.69 %                88.13 %   88.4 %   \n",
       "1     VGG16              85.54 %                 85.0 %  85.27 %   \n",
       "2     VGG19              84.25 %                83.75 %   84.0 %   \n",
       "3  Xception               90.0 %                 90.0 %   90.0 %   \n",
       "\n",
       "  F1-Score Calculado  \n",
       "0            88.41 %  \n",
       "1            85.27 %  \n",
       "2             84.0 %  \n",
       "3             90.0 %  "
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "xception_f1_score_calc = round((2*xception_val_precision_score*xception_val_recall_score)/(xception_val_precision_score+xception_val_recall_score),2)\n",
    "\n",
    "mymodel_results = {'Modelo':'Xception',\n",
    "                   'Precisão (Precision)': str(xception_val_precision_score) + ' %',\n",
    "                   'Sensibilidade (Recall)': str(xception_val_recall_score) + ' %',\n",
    "                   'F1-Score': str(xception_val_f1_score) + ' %',\n",
    "                   'F1-Score Calculado': str(xception_f1_score_calc) + ' %'}\n",
    "                            \n",
    "if len(Comparativo_Arquiteturas) < 4:\n",
    "    Comparativo_Arquiteturas = Comparativo_Arquiteturas.append(mymodel_results, ignore_index=True)\n",
    "\n",
    "Comparativo_Arquiteturas"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Como observado na tabela acima, a performance do modelo Xception teve a melhor precisão e F1-Score.\n",
    "\n",
    "O Xception é semelhante ao ResNet50 em tamanho, só que possui uma profundidade elevada, o que pode explicar sua precisão levemente melhor. Para esta aplicação, com os modelos usados, podemos sugerir o uso do Xception em função dos resultados apresentados."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Abaixo fizemos alguns testes, para salvar e carregar o modelo, mas em função de não sabermos se isso estaria infringindo as regras do trabalho :-), acabamos deixando de fora os resultados, mas serve de consulta e até de revisão do Github no futuro."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Testes\n",
    "\n",
    "'''\n",
    "model.save('modelos/modelo-xception.h5')\n",
    "model.save_weights('modelos/modelo-xception_pesos.h5')\n",
    "\n",
    "with open('modelos/architecture-raiox.json', 'w') as f:\n",
    "    f.write(model.to_json())'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# carregando o modelo no formato HDf5\n",
    "\n",
    "'''\n",
    "model = load_model('modelos/modelo-xception.h5')\n",
    "\n",
    "# arquitetura das camadas em JSSON e pesos treinados em HDF5\n",
    "with open('modelos/architecture-raiox.json') as f:\n",
    "    model = model_from_json(f.read())\n",
    "    \n",
    "model.load_weights('modelos/modelo-xception_pesos.h5')'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Teste simples do modelo Xception\n",
    "'''\n",
    "validation_img_paths = [\"raiox/test/bacteria/person101_bacteria_484.jpg\",\n",
    "                        \"raiox/test/bacteria/person109_bacteria_519.jpg\",\n",
    "                        \"raiox/test/controle/IM-0011-0001-0002.jpg\",\n",
    "                        \"raiox/test/viral/person15_virus_46.jpeg\"]\n",
    "\n",
    "img_list = [Image.open(img_path) for img_path in validation_img_paths]'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Convertendo as imagens, em função de não usar o ImageDataGenerator\n",
    "\n",
    "'''\n",
    "validation_batch = np.stack([preprocess_input(np.array(img.resize((224,224)).convert(\"RGB\"))) for img in img_list])\n",
    "\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "pred_probs = model.predict(validation_batch)\n",
    "pred_probs\n",
    "\n",
    "'''"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "'''\n",
    "\n",
    "plt.figure(figsize=(20,20))\n",
    "\n",
    "for i, img in enumerate(img_list):\n",
    "    plt.subplot(2,2,i+1)\n",
    "    plt.imshow(img)\n",
    "    plt.title(\"{:.0f}% Bacteria, {:.0f}% Controle, {:.0f}% Viral\".format(100*pred_probs[i,0], 100*pred_probs[i,1], 100*pred_probs[i,2]))\n",
    "    '''"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}