local_PySpark__Assaadi_Yassine_2_notebook_052022.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/Yassaadi/Scaling_cnn_emr/blob/main/local_PySpark__Assaadi_Yassine_2_notebook_052022.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "#GOOGLE COLAB\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')\n",
        "%cd /content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire\n",
        "from IPython.display import Image"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "FRSGam0KpbIB",
        "outputId": "c0702b44-6e93-4585-d693-d3836b3a55e9",
        "collapsed": true
      },
      "id": "FRSGam0KpbIB",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/drive\n",
            "/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Dd6t0uFzuR4X"
      },
      "source": [
        "<a id='installing-spark'></a>\n",
        "### Installing Spark"
      ],
      "id": "Dd6t0uFzuR4X"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6apGVff5h4ca"
      },
      "source": [
        "Install Dependencies:\n",
        "\n",
        "\n",
        "1.   Java 8\n",
        "2.   Apache Spark with hadoop and\n",
        "3.   Findspark (used to locate the spark in the system)\n"
      ],
      "id": "6apGVff5h4ca"
    },
    {
      "cell_type": "code",
      "source": [
        "!pip install Pandas pillow pyarrow pyspark"
      ],
      "metadata": {
        "collapsed": true,
        "id": "4pgged2NUH9M",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "405a7ba1-1537-43b8-f91c-aa33899e115a"
      },
      "id": "4pgged2NUH9M",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Looking in indexes: https://pypi.org/simple, https://us-python.pkg.dev/colab-wheels/public/simple/\n",
            "Requirement already satisfied: Pandas in /usr/local/lib/python3.9/dist-packages (1.4.4)\n",
            "Requirement already satisfied: pillow in /usr/local/lib/python3.9/dist-packages (8.4.0)\n",
            "Requirement already satisfied: pyarrow in /usr/local/lib/python3.9/dist-packages (9.0.0)\n",
            "Collecting pyspark\n",
            "  Downloading pyspark-3.3.2.tar.gz (281.4 MB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m281.4/281.4 MB\u001b[0m \u001b[31m6.3 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25h  Preparing metadata (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "Requirement already satisfied: numpy>=1.18.5 in /usr/local/lib/python3.9/dist-packages (from Pandas) (1.22.4)\n",
            "Requirement already satisfied: pytz>=2020.1 in /usr/local/lib/python3.9/dist-packages (from Pandas) (2022.7.1)\n",
            "Requirement already satisfied: python-dateutil>=2.8.1 in /usr/local/lib/python3.9/dist-packages (from Pandas) (2.8.2)\n",
            "Collecting py4j==0.10.9.5\n",
            "  Downloading py4j-0.10.9.5-py2.py3-none-any.whl (199 kB)\n",
            "\u001b[2K     \u001b[90m━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━\u001b[0m \u001b[32m199.7/199.7 KB\u001b[0m \u001b[31m26.5 MB/s\u001b[0m eta \u001b[36m0:00:00\u001b[0m\n",
            "\u001b[?25hRequirement already satisfied: six>=1.5 in /usr/local/lib/python3.9/dist-packages (from python-dateutil>=2.8.1->Pandas) (1.16.0)\n",
            "Building wheels for collected packages: pyspark\n",
            "  Building wheel for pyspark (setup.py) ... \u001b[?25l\u001b[?25hdone\n",
            "  Created wheel for pyspark: filename=pyspark-3.3.2-py2.py3-none-any.whl size=281824028 sha256=904672271cb573a80c17542b8ca6e55e672d865aa500903129637dda750fd8c0\n",
            "  Stored in directory: /root/.cache/pip/wheels/6c/e3/9b/0525ce8a69478916513509d43693511463c6468db0de237c86\n",
            "Successfully built pyspark\n",
            "Installing collected packages: py4j, pyspark\n",
            "  Attempting uninstall: py4j\n",
            "    Found existing installation: py4j 0.10.9.7\n",
            "    Uninstalling py4j-0.10.9.7:\n",
            "      Successfully uninstalled py4j-0.10.9.7\n",
            "Successfully installed py4j-0.10.9.5 pyspark-3.3.2\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "id": "33a43845",
      "metadata": {
        "id": "33a43845"
      },
      "source": [
        "## 3.4 Import des librairies"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a5c0c74f",
      "metadata": {
        "id": "a5c0c74f"
      },
      "outputs": [],
      "source": [
        "import pandas as pd\n",
        "from PIL import Image\n",
        "import numpy as np\n",
        "import io\n",
        "import os\n",
        "\n",
        "import tensorflow as tf\n",
        "from tensorflow.keras.applications.mobilenet_v2 import MobileNetV2, preprocess_input\n",
        "from tensorflow.keras.preprocessing.image import img_to_array\n",
        "from tensorflow.keras import Model\n",
        "from pyspark.sql.functions import col, pandas_udf, PandasUDFType, element_at, split\n",
        "from pyspark.sql import SparkSession\n",
        "from pyspark.ml.feature import VectorAssembler, StandardScaler, PCA\n",
        "import matplotlib.pyplot as plt"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "!ls"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true,
        "id": "KbldLyyp6U--",
        "outputId": "94d5e05c-b040-4ccc-e3fc-7601fd8d2b48"
      },
      "id": "KbldLyyp6U--",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "data\t\t\t\t  spark-3.1.1-bin-hadoop3.2.tgz.5\n",
            "img\t\t\t\t  spark-3.1.1-bin-hadoop3.2.tgz.6\n",
            "spark-3.1.1-bin-hadoop3.2\t  spark-3.1.1-bin-hadoop3.2.tgz.7\n",
            "spark-3.1.1-bin-hadoop3.2.tgz\t  spark-3.1.1-bin-hadoop3.2.tgz.8\n",
            "spark-3.1.1-bin-hadoop3.2.tgz.1   spark-3.1.1-bin-hadoop3.2.tgz.9\n",
            "spark-3.1.1-bin-hadoop3.2.tgz.10  spark-3.2.1-bin-hadoop3.2\n",
            "spark-3.1.1-bin-hadoop3.2.tgz.2   spark-3.2.1-bin-hadoop3.2.tgz\n",
            "spark-3.1.1-bin-hadoop3.2.tgz.3   YAS_P8_Notebook_Linux_EMR_PySpark_V2.0.ipynb\n",
            "spark-3.1.1-bin-hadoop3.2.tgz.4\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "id": "661ff67c",
      "metadata": {
        "id": "661ff67c"
      },
      "source": [
        "## 3.5 Définition des PATH pour charger les images <br /> et enregistrer les résultats\n",
        "\n",
        "Dans cette version locale nous partons du principe que les données <br />\n",
        "sont stockées dans le même répertoire que le notebook.<br />\n",
        "Nous n'utilisons qu'un extrait de **300 images** à traiter dans cette <br />\n",
        "première version en local.<br />\n",
        "L'extrait des images à charger est stockée dans le dossier **Test1**.<br />\n",
        "Nous enregistrerons le résultat de notre traitement <br />\n",
        "dans le dossier \"**Results_Local**\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "cde0aa67",
      "metadata": {
        "id": "cde0aa67",
        "outputId": "30dd48c4-ea47-472f-8911-6772c8532ddf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "PATH:        /content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire\n",
            "PATH_Data:   /content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1\n",
            "PATH_Result: /content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Results\n"
          ]
        }
      ],
      "source": [
        "PATH = os.getcwd() #Return current directory\n",
        "PATH_Data = PATH+'/data/Test1'\n",
        "PATH_Result = PATH+'/data/Results'\n",
        "print('PATH:        '+\\\n",
        "      PATH+'\\nPATH_Data:   '+\\\n",
        "      PATH_Data+'\\nPATH_Result: '+PATH_Result)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "da5e637a",
      "metadata": {
        "id": "da5e637a"
      },
      "source": [
        "## 3.6 Création de la SparkSession\n",
        "\n",
        "L’application Spark est contrôlée grâce à un processus de pilotage (driver process) appelé **SparkSession**. <br />\n",
        "<u>Une instance de **SparkSession** est la façon dont Spark exécute les fonctions définies par l’utilisateur <br />\n",
        "dans l’ensemble du cluster</u>. <u>Une SparkSession correspond toujours à une application Spark</u>.\n",
        "\n",
        "<u>Ici nous créons une session spark en spécifiant dans l'ordre</u> :\n",
        " 1. un **nom pour l'application**, qui sera affichée dans l'interface utilisateur Web Spark \"**P8**\"\n",
        " 2. que l'application doit s'exécuter **localement**. <br />\n",
        "   Nous ne définissons pas le nombre de cœurs à utiliser (comme .master('local[4]) pour 4 cœurs à utiliser), <br />\n",
        "   nous utiliserons donc tous les cœurs disponibles dans notre processeur.<br />\n",
        " 3. une option de configuration supplémentaire permettant d'utiliser le **format \"parquet\"** <br />\n",
        "   que nous utiliserons pour enregistrer et charger le résultat de notre travail.\n",
        " 4. vouloir **obtenir une session spark** existante ou si aucune n'existe, en créer une nouvelle"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "b7bea157",
      "metadata": {
        "id": "b7bea157"
      },
      "outputs": [],
      "source": [
        "spark = (SparkSession\n",
        "             .builder\n",
        "             .appName('P8')\n",
        "             .master(\"local[*]\")\n",
        "             .config(\"spark.sql.parquet.writeLegacyFormat\", 'true')\n",
        "             .getOrCreate()\n",
        ")\n",
        "\n",
        "#spark.conf.set(\"spark.sql.repl.eagerEval.enabled\", True) # Property used to format output tables better"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "5c8b53ac",
      "metadata": {
        "id": "5c8b53ac"
      },
      "source": [
        "<u>Nous créons également la variable \"**sc**\" qui est un **SparkContext** issue de la variable **spark**</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "14aeccb1",
      "metadata": {
        "id": "14aeccb1"
      },
      "outputs": [],
      "source": [
        "sc = spark.sparkContext"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "5a086010",
      "metadata": {
        "id": "5a086010"
      },
      "source": [
        "<u>Affichage des informations de Spark en cours d'execution</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e97bf13b",
      "metadata": {
        "id": "e97bf13b",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 219
        },
        "outputId": "59be8a3b-5555-4d53-8733-13e672bb306e"
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<pyspark.sql.session.SparkSession at 0x7fb50804d490>"
            ],
            "text/html": [
              "\n",
              "            <div>\n",
              "                <p><b>SparkSession - in-memory</b></p>\n",
              "                \n",
              "        <div>\n",
              "            <p><b>SparkContext</b></p>\n",
              "\n",
              "            <p><a href=\"http://3bf242038235:4040\">Spark UI</a></p>\n",
              "\n",
              "            <dl>\n",
              "              <dt>Version</dt>\n",
              "                <dd><code>v3.3.2</code></dd>\n",
              "              <dt>Master</dt>\n",
              "                <dd><code>local[*]</code></dd>\n",
              "              <dt>AppName</dt>\n",
              "                <dd><code>P8</code></dd>\n",
              "            </dl>\n",
              "        </div>\n",
              "        \n",
              "            </div>\n",
              "        "
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ],
      "source": [
        "spark"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "195a88b0",
      "metadata": {
        "id": "195a88b0"
      },
      "source": [
        "## 3.7 Traitement des données\n",
        "\n",
        "<u>Dans la suite de notre flux de travail, <br />\n",
        "nous allons successivement</u> :\n",
        "1. Préparer nos données\n",
        "    1. Importer les images dans un dataframe **pandas UDF**\n",
        "    2. Associer aux images leur **label**\n",
        "    3. Préprocesser en **redimensionnant nos images pour <br />\n",
        "       qu'elles soient compatibles avec notre modèle**\n",
        "2. Préparer notre modèle\n",
        "    1. Importer le modèle **MobileNetV2**\n",
        "    2. Créer un **nouveau modèle** dépourvu de la dernière couche de MobileNetV2\n",
        "3. Définir le processus de chargement des images et l'application <br />\n",
        "   de leur featurisation à travers l'utilisation de pandas UDF\n",
        "3. Exécuter les actions d'extraction de features\n",
        "4. Enregistrer le résultat de nos actions\n",
        "5. Tester le bon fonctionnement en chargeant les données enregistrées\n",
        "\n",
        "\n"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "386fe0bc",
      "metadata": {
        "id": "386fe0bc"
      },
      "source": [
        "### 3.7.1 Chargement des données\n",
        "\n",
        "Les images sont chargées au format binaire, ce qui offre, <br />\n",
        "plus de souplesse dans la façon de prétraiter les images.\n",
        "\n",
        "Avant de charger les images, nous spécifions que nous voulons charger <br />\n",
        "uniquement les fichiers dont l'extension est **jpg**.\n",
        "\n",
        "Nous indiquons également de charger tous les objets possibles contenus <br />\n",
        "dans les sous-dossiers du dossier communiqué."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e68e53b9",
      "metadata": {
        "id": "e68e53b9",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 201
        },
        "outputId": "cd5dd5d7-b7aa-4b0a-c4bf-7c6cb9f0e6ff"
      },
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-14-f83e1ebc5ce0>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      2\u001b[0m   \u001b[0;34m.\u001b[0m\u001b[0moption\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"pathGlobFilter\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"*.jpg\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      3\u001b[0m   \u001b[0;34m.\u001b[0m\u001b[0moption\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"recursiveFileLookup\"\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m\"true\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;31m \u001b[0m\u001b[0;31m\\\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 4\u001b[0;31m   \u001b[0;34m.\u001b[0m\u001b[0mload\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mPATH_Data\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m: name 'PATH_Data' is not defined"
          ]
        }
      ],
      "source": [
        "images = spark.read.format(\"binaryFile\") \\\n",
        "  .option(\"pathGlobFilter\", \"*.jpg\") \\\n",
        "  .option(\"recursiveFileLookup\", \"true\") \\\n",
        "  .load(PATH_Data)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "645faeaf",
      "metadata": {
        "id": "645faeaf"
      },
      "source": [
        "<u>Affichage des 5 premières images contenant</u> :\n",
        " - le path de l'image\n",
        " - la date et heure de sa dernière modification\n",
        " - sa longueur\n",
        " - son contenu encodé en valeur hexadécimal"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "863981e5",
      "metadata": {
        "id": "863981e5"
      },
      "source": [
        "<u>Je ne conserve que le **path** de l'image et j'ajoute <br />\n",
        "    une colonne contenant les **labels** de chaque image</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a08b0494",
      "metadata": {
        "id": "a08b0494",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "908846e9-d621-412b-ddca-de7de0c4d388"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "root\n",
            " |-- path: string (nullable = true)\n",
            " |-- modificationTime: timestamp (nullable = true)\n",
            " |-- length: long (nullable = true)\n",
            " |-- content: binary (nullable = true)\n",
            " |-- label: string (nullable = true)\n",
            "\n",
            "None\n",
            "+--------------------------------------------------------------------------------------------------------------+---------------+\n",
            "|path                                                                                                          |label          |\n",
            "+--------------------------------------------------------------------------------------------------------------+---------------+\n",
            "|file:/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1/cabbage_white_1/r0_11.jpg|cabbage_white_1|\n",
            "|file:/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1/cabbage_white_1/r0_7.jpg |cabbage_white_1|\n",
            "|file:/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1/cabbage_white_1/r0_19.jpg|cabbage_white_1|\n",
            "|file:/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1/cabbage_white_1/r0_15.jpg|cabbage_white_1|\n",
            "|file:/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Test1/cabbage_white_1/r0_3.jpg |cabbage_white_1|\n",
            "+--------------------------------------------------------------------------------------------------------------+---------------+\n",
            "only showing top 5 rows\n",
            "\n",
            "None\n"
          ]
        }
      ],
      "source": [
        "images = images.withColumn('label', element_at(split(images['path'], '/'),-2))\n",
        "print(images.printSchema())\n",
        "print(images.select('path','label').show(5,False))"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "images.show()"
      ],
      "metadata": {
        "id": "v1SJpmO0sGMd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "fa169e3d-1248-4496-e4b0-2ffcaf1e5c8a"
      },
      "id": "v1SJpmO0sGMd",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "+--------------------+-------------------+------+--------------------+---------------+\n",
            "|                path|   modificationTime|length|             content|          label|\n",
            "+--------------------+-------------------+------+--------------------+---------------+\n",
            "|file:/content/dri...|2021-09-12 17:23:50| 99176|[FF D8 FF E0 00 1...|cabbage_white_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:50| 98391|[FF D8 FF E0 00 1...|cabbage_white_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:50| 97815|[FF D8 FF E0 00 1...|cabbage_white_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:50| 96546|[FF D8 FF E0 00 1...|cabbage_white_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:50| 96048|[FF D8 FF E0 00 1...|cabbage_white_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 77993|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 77311|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 76909|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 76395|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 75940|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 75424|[FF D8 FF E0 00 1...|         pear_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 65195|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 64938|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 64802|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 64734|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 64587|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 64575|[FF D8 FF E0 00 1...|     cucumber_3|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 56533|[FF D8 FF E0 00 1...|         pear_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:54| 56483|[FF D8 FF E0 00 1...|         pear_1|\n",
            "|file:/content/dri...|2021-09-12 17:23:52| 55913|[FF D8 FF E0 00 1...|     cucumber_1|\n",
            "+--------------------+-------------------+------+--------------------+---------------+\n",
            "only showing top 20 rows\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "id": "83d47705",
      "metadata": {
        "id": "83d47705"
      },
      "source": [
        "### 3.7.2 Préparation du modèle\n",
        "\n",
        "Je vais utiliser la technique du **transfert learning** pour extraire les features des images.<br />\n",
        "J'ai choisi d'utiliser le modèle **MobileNetV2** pour sa rapidité d'exécution comparée <br />\n",
        "à d'autres modèles comme *VGG16* par exemple.\n",
        "\n",
        "Pour en savoir plus sur la conception et le fonctionnement de MobileNetV2, <br />\n",
        "je vous invite à lire [cet article](https://towardsdatascience.com/review-mobilenetv2-light-weight-model-image-classification-8febb490e61c).\n",
        "\n",
        "<u>Voici le schéma de son architecture globale</u> : \n",
        "\n",
        "![Architecture de MobileNetV2](/content/drive/MyDrive/Colab\\Notebooks/P8-Fruits/P8_Mode_opératoire/img/mobilenetv2_architecture.png)\n",
        "\n",
        "Il existe une dernière couche qui sert à classer les images <br />\n",
        "selon 1000 catégories que nous ne voulons pas utiliser.<br />\n",
        "L'idée dans ce projet est de récupérer le **vecteur de caractéristiques <br />\n",
        "de dimensions (1,1,1280)** qui servira, plus tard, au travers d'un moteur <br />\n",
        "de classification à reconnaitre les différents fruits du jeu de données.\n",
        "\n",
        "Comme d'autres modèles similaires, **MobileNetV2**, lorsqu'on l'utilise <br />\n",
        "en incluant toutes ses couches, attend obligatoirement des images <br />\n",
        "de dimension (224,224,3). Nos images étant toutes de dimension (100,100,3), <br />\n",
        "nous devrons simplement les **redimensionner** avant de les confier au modèle.\n",
        "\n",
        "<u>Dans l'odre</u> :\n",
        " 1. Nous chargeons le modèle **MobileNetV2** avec les poids **précalculés** <br />\n",
        "    issus d'**imagenet** et en spécifiant le format de nos images en entrée\n",
        " 2. Nous créons un nouveau modèle avec:\n",
        "  - <u>en entrée</u> : l'entrée du modèle MobileNetV2\n",
        "  - <u>en sortie</u> : l'avant dernière couche du modèle MobileNetV2"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "9cdd9bdf",
      "metadata": {
        "id": "9cdd9bdf",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "0b76fa20-1b40-4c71-8caf-30d33708bcbc"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v2/mobilenet_v2_weights_tf_dim_ordering_tf_kernels_1.0_224.h5\n",
            "14536120/14536120 [==============================] - 2s 0us/step\n"
          ]
        }
      ],
      "source": [
        "model = MobileNetV2(weights='imagenet',\n",
        "                    include_top=True,\n",
        "                    input_shape=(224, 224, 3))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "99d6b68d",
      "metadata": {
        "id": "99d6b68d"
      },
      "outputs": [],
      "source": [
        "new_model = Model(inputs=model.input,\n",
        "                  outputs=model.layers[-2].output)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "7b197379",
      "metadata": {
        "id": "7b197379"
      },
      "source": [
        "Affichage du résumé de notre nouveau modèle où nous constatons <br />\n",
        "que <u>nous récupérons bien en sortie un vecteur de dimension (1, 1, 1280)</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e8207725",
      "metadata": {
        "id": "e8207725",
        "outputId": "1c037fcc-723d-4380-a2c3-db0e2cc00efb",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "collapsed": true
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            " Layer (type)                   Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            " input_1 (InputLayer)           [(None, 224, 224, 3  0           []                               \n",
            "                                )]                                                                \n",
            "                                                                                                  \n",
            " Conv1 (Conv2D)                 (None, 112, 112, 32  864         ['input_1[0][0]']                \n",
            "                                )                                                                 \n",
            "                                                                                                  \n",
            " bn_Conv1 (BatchNormalization)  (None, 112, 112, 32  128         ['Conv1[0][0]']                  \n",
            "                                )                                                                 \n",
            "                                                                                                  \n",
            " Conv1_relu (ReLU)              (None, 112, 112, 32  0           ['bn_Conv1[0][0]']               \n",
            "                                )                                                                 \n",
            "                                                                                                  \n",
            " expanded_conv_depthwise (Depth  (None, 112, 112, 32  288        ['Conv1_relu[0][0]']             \n",
            " wiseConv2D)                    )                                                                 \n",
            "                                                                                                  \n",
            " expanded_conv_depthwise_BN (Ba  (None, 112, 112, 32  128        ['expanded_conv_depthwise[0][0]']\n",
            " tchNormalization)              )                                                                 \n",
            "                                                                                                  \n",
            " expanded_conv_depthwise_relu (  (None, 112, 112, 32  0          ['expanded_conv_depthwise_BN[0][0\n",
            " ReLU)                          )                                ]']                              \n",
            "                                                                                                  \n",
            " expanded_conv_project (Conv2D)  (None, 112, 112, 16  512        ['expanded_conv_depthwise_relu[0]\n",
            "                                )                                [0]']                            \n",
            "                                                                                                  \n",
            " expanded_conv_project_BN (Batc  (None, 112, 112, 16  64         ['expanded_conv_project[0][0]']  \n",
            " hNormalization)                )                                                                 \n",
            "                                                                                                  \n",
            " block_1_expand (Conv2D)        (None, 112, 112, 96  1536        ['expanded_conv_project_BN[0][0]'\n",
            "                                )                                ]                                \n",
            "                                                                                                  \n",
            " block_1_expand_BN (BatchNormal  (None, 112, 112, 96  384        ['block_1_expand[0][0]']         \n",
            " ization)                       )                                                                 \n",
            "                                                                                                  \n",
            " block_1_expand_relu (ReLU)     (None, 112, 112, 96  0           ['block_1_expand_BN[0][0]']      \n",
            "                                )                                                                 \n",
            "                                                                                                  \n",
            " block_1_pad (ZeroPadding2D)    (None, 113, 113, 96  0           ['block_1_expand_relu[0][0]']    \n",
            "                                )                                                                 \n",
            "                                                                                                  \n",
            " block_1_depthwise (DepthwiseCo  (None, 56, 56, 96)  864         ['block_1_pad[0][0]']            \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_1_depthwise_BN (BatchNor  (None, 56, 56, 96)  384         ['block_1_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_1_depthwise_relu (ReLU)  (None, 56, 56, 96)   0           ['block_1_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_1_project (Conv2D)       (None, 56, 56, 24)   2304        ['block_1_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_1_project_BN (BatchNorma  (None, 56, 56, 24)  96          ['block_1_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_2_expand (Conv2D)        (None, 56, 56, 144)  3456        ['block_1_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_2_expand_BN (BatchNormal  (None, 56, 56, 144)  576        ['block_2_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_2_expand_relu (ReLU)     (None, 56, 56, 144)  0           ['block_2_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_2_depthwise (DepthwiseCo  (None, 56, 56, 144)  1296       ['block_2_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_2_depthwise_BN (BatchNor  (None, 56, 56, 144)  576        ['block_2_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_2_depthwise_relu (ReLU)  (None, 56, 56, 144)  0           ['block_2_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_2_project (Conv2D)       (None, 56, 56, 24)   3456        ['block_2_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_2_project_BN (BatchNorma  (None, 56, 56, 24)  96          ['block_2_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_2_add (Add)              (None, 56, 56, 24)   0           ['block_1_project_BN[0][0]',     \n",
            "                                                                  'block_2_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_3_expand (Conv2D)        (None, 56, 56, 144)  3456        ['block_2_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_3_expand_BN (BatchNormal  (None, 56, 56, 144)  576        ['block_3_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_3_expand_relu (ReLU)     (None, 56, 56, 144)  0           ['block_3_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_3_pad (ZeroPadding2D)    (None, 57, 57, 144)  0           ['block_3_expand_relu[0][0]']    \n",
            "                                                                                                  \n",
            " block_3_depthwise (DepthwiseCo  (None, 28, 28, 144)  1296       ['block_3_pad[0][0]']            \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_3_depthwise_BN (BatchNor  (None, 28, 28, 144)  576        ['block_3_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_3_depthwise_relu (ReLU)  (None, 28, 28, 144)  0           ['block_3_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_3_project (Conv2D)       (None, 28, 28, 32)   4608        ['block_3_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_3_project_BN (BatchNorma  (None, 28, 28, 32)  128         ['block_3_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_4_expand (Conv2D)        (None, 28, 28, 192)  6144        ['block_3_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_4_expand_BN (BatchNormal  (None, 28, 28, 192)  768        ['block_4_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_4_expand_relu (ReLU)     (None, 28, 28, 192)  0           ['block_4_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_4_depthwise (DepthwiseCo  (None, 28, 28, 192)  1728       ['block_4_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_4_depthwise_BN (BatchNor  (None, 28, 28, 192)  768        ['block_4_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_4_depthwise_relu (ReLU)  (None, 28, 28, 192)  0           ['block_4_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_4_project (Conv2D)       (None, 28, 28, 32)   6144        ['block_4_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_4_project_BN (BatchNorma  (None, 28, 28, 32)  128         ['block_4_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_4_add (Add)              (None, 28, 28, 32)   0           ['block_3_project_BN[0][0]',     \n",
            "                                                                  'block_4_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_5_expand (Conv2D)        (None, 28, 28, 192)  6144        ['block_4_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_5_expand_BN (BatchNormal  (None, 28, 28, 192)  768        ['block_5_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_5_expand_relu (ReLU)     (None, 28, 28, 192)  0           ['block_5_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_5_depthwise (DepthwiseCo  (None, 28, 28, 192)  1728       ['block_5_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_5_depthwise_BN (BatchNor  (None, 28, 28, 192)  768        ['block_5_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_5_depthwise_relu (ReLU)  (None, 28, 28, 192)  0           ['block_5_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_5_project (Conv2D)       (None, 28, 28, 32)   6144        ['block_5_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_5_project_BN (BatchNorma  (None, 28, 28, 32)  128         ['block_5_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_5_add (Add)              (None, 28, 28, 32)   0           ['block_4_add[0][0]',            \n",
            "                                                                  'block_5_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_6_expand (Conv2D)        (None, 28, 28, 192)  6144        ['block_5_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_6_expand_BN (BatchNormal  (None, 28, 28, 192)  768        ['block_6_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_6_expand_relu (ReLU)     (None, 28, 28, 192)  0           ['block_6_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_6_pad (ZeroPadding2D)    (None, 29, 29, 192)  0           ['block_6_expand_relu[0][0]']    \n",
            "                                                                                                  \n",
            " block_6_depthwise (DepthwiseCo  (None, 14, 14, 192)  1728       ['block_6_pad[0][0]']            \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_6_depthwise_BN (BatchNor  (None, 14, 14, 192)  768        ['block_6_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_6_depthwise_relu (ReLU)  (None, 14, 14, 192)  0           ['block_6_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_6_project (Conv2D)       (None, 14, 14, 64)   12288       ['block_6_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_6_project_BN (BatchNorma  (None, 14, 14, 64)  256         ['block_6_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_7_expand (Conv2D)        (None, 14, 14, 384)  24576       ['block_6_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_7_expand_BN (BatchNormal  (None, 14, 14, 384)  1536       ['block_7_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_7_expand_relu (ReLU)     (None, 14, 14, 384)  0           ['block_7_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_7_depthwise (DepthwiseCo  (None, 14, 14, 384)  3456       ['block_7_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_7_depthwise_BN (BatchNor  (None, 14, 14, 384)  1536       ['block_7_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_7_depthwise_relu (ReLU)  (None, 14, 14, 384)  0           ['block_7_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_7_project (Conv2D)       (None, 14, 14, 64)   24576       ['block_7_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_7_project_BN (BatchNorma  (None, 14, 14, 64)  256         ['block_7_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_7_add (Add)              (None, 14, 14, 64)   0           ['block_6_project_BN[0][0]',     \n",
            "                                                                  'block_7_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_8_expand (Conv2D)        (None, 14, 14, 384)  24576       ['block_7_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_8_expand_BN (BatchNormal  (None, 14, 14, 384)  1536       ['block_8_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_8_expand_relu (ReLU)     (None, 14, 14, 384)  0           ['block_8_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_8_depthwise (DepthwiseCo  (None, 14, 14, 384)  3456       ['block_8_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_8_depthwise_BN (BatchNor  (None, 14, 14, 384)  1536       ['block_8_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_8_depthwise_relu (ReLU)  (None, 14, 14, 384)  0           ['block_8_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_8_project (Conv2D)       (None, 14, 14, 64)   24576       ['block_8_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_8_project_BN (BatchNorma  (None, 14, 14, 64)  256         ['block_8_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_8_add (Add)              (None, 14, 14, 64)   0           ['block_7_add[0][0]',            \n",
            "                                                                  'block_8_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_9_expand (Conv2D)        (None, 14, 14, 384)  24576       ['block_8_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_9_expand_BN (BatchNormal  (None, 14, 14, 384)  1536       ['block_9_expand[0][0]']         \n",
            " ization)                                                                                         \n",
            "                                                                                                  \n",
            " block_9_expand_relu (ReLU)     (None, 14, 14, 384)  0           ['block_9_expand_BN[0][0]']      \n",
            "                                                                                                  \n",
            " block_9_depthwise (DepthwiseCo  (None, 14, 14, 384)  3456       ['block_9_expand_relu[0][0]']    \n",
            " nv2D)                                                                                            \n",
            "                                                                                                  \n",
            " block_9_depthwise_BN (BatchNor  (None, 14, 14, 384)  1536       ['block_9_depthwise[0][0]']      \n",
            " malization)                                                                                      \n",
            "                                                                                                  \n",
            " block_9_depthwise_relu (ReLU)  (None, 14, 14, 384)  0           ['block_9_depthwise_BN[0][0]']   \n",
            "                                                                                                  \n",
            " block_9_project (Conv2D)       (None, 14, 14, 64)   24576       ['block_9_depthwise_relu[0][0]'] \n",
            "                                                                                                  \n",
            " block_9_project_BN (BatchNorma  (None, 14, 14, 64)  256         ['block_9_project[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_9_add (Add)              (None, 14, 14, 64)   0           ['block_8_add[0][0]',            \n",
            "                                                                  'block_9_project_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_10_expand (Conv2D)       (None, 14, 14, 384)  24576       ['block_9_add[0][0]']            \n",
            "                                                                                                  \n",
            " block_10_expand_BN (BatchNorma  (None, 14, 14, 384)  1536       ['block_10_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_10_expand_relu (ReLU)    (None, 14, 14, 384)  0           ['block_10_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_10_depthwise (DepthwiseC  (None, 14, 14, 384)  3456       ['block_10_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_10_depthwise_BN (BatchNo  (None, 14, 14, 384)  1536       ['block_10_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_10_depthwise_relu (ReLU)  (None, 14, 14, 384)  0          ['block_10_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_10_project (Conv2D)      (None, 14, 14, 96)   36864       ['block_10_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_10_project_BN (BatchNorm  (None, 14, 14, 96)  384         ['block_10_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_11_expand (Conv2D)       (None, 14, 14, 576)  55296       ['block_10_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_11_expand_BN (BatchNorma  (None, 14, 14, 576)  2304       ['block_11_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_11_expand_relu (ReLU)    (None, 14, 14, 576)  0           ['block_11_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_11_depthwise (DepthwiseC  (None, 14, 14, 576)  5184       ['block_11_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_11_depthwise_BN (BatchNo  (None, 14, 14, 576)  2304       ['block_11_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_11_depthwise_relu (ReLU)  (None, 14, 14, 576)  0          ['block_11_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_11_project (Conv2D)      (None, 14, 14, 96)   55296       ['block_11_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_11_project_BN (BatchNorm  (None, 14, 14, 96)  384         ['block_11_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_11_add (Add)             (None, 14, 14, 96)   0           ['block_10_project_BN[0][0]',    \n",
            "                                                                  'block_11_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_12_expand (Conv2D)       (None, 14, 14, 576)  55296       ['block_11_add[0][0]']           \n",
            "                                                                                                  \n",
            " block_12_expand_BN (BatchNorma  (None, 14, 14, 576)  2304       ['block_12_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_12_expand_relu (ReLU)    (None, 14, 14, 576)  0           ['block_12_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_12_depthwise (DepthwiseC  (None, 14, 14, 576)  5184       ['block_12_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_12_depthwise_BN (BatchNo  (None, 14, 14, 576)  2304       ['block_12_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_12_depthwise_relu (ReLU)  (None, 14, 14, 576)  0          ['block_12_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_12_project (Conv2D)      (None, 14, 14, 96)   55296       ['block_12_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_12_project_BN (BatchNorm  (None, 14, 14, 96)  384         ['block_12_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_12_add (Add)             (None, 14, 14, 96)   0           ['block_11_add[0][0]',           \n",
            "                                                                  'block_12_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_13_expand (Conv2D)       (None, 14, 14, 576)  55296       ['block_12_add[0][0]']           \n",
            "                                                                                                  \n",
            " block_13_expand_BN (BatchNorma  (None, 14, 14, 576)  2304       ['block_13_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_13_expand_relu (ReLU)    (None, 14, 14, 576)  0           ['block_13_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_13_pad (ZeroPadding2D)   (None, 15, 15, 576)  0           ['block_13_expand_relu[0][0]']   \n",
            "                                                                                                  \n",
            " block_13_depthwise (DepthwiseC  (None, 7, 7, 576)   5184        ['block_13_pad[0][0]']           \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_13_depthwise_BN (BatchNo  (None, 7, 7, 576)   2304        ['block_13_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_13_depthwise_relu (ReLU)  (None, 7, 7, 576)   0           ['block_13_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_13_project (Conv2D)      (None, 7, 7, 160)    92160       ['block_13_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_13_project_BN (BatchNorm  (None, 7, 7, 160)   640         ['block_13_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_14_expand (Conv2D)       (None, 7, 7, 960)    153600      ['block_13_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_14_expand_BN (BatchNorma  (None, 7, 7, 960)   3840        ['block_14_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_14_expand_relu (ReLU)    (None, 7, 7, 960)    0           ['block_14_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_14_depthwise (DepthwiseC  (None, 7, 7, 960)   8640        ['block_14_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_14_depthwise_BN (BatchNo  (None, 7, 7, 960)   3840        ['block_14_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_14_depthwise_relu (ReLU)  (None, 7, 7, 960)   0           ['block_14_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_14_project (Conv2D)      (None, 7, 7, 160)    153600      ['block_14_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_14_project_BN (BatchNorm  (None, 7, 7, 160)   640         ['block_14_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_14_add (Add)             (None, 7, 7, 160)    0           ['block_13_project_BN[0][0]',    \n",
            "                                                                  'block_14_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_15_expand (Conv2D)       (None, 7, 7, 960)    153600      ['block_14_add[0][0]']           \n",
            "                                                                                                  \n",
            " block_15_expand_BN (BatchNorma  (None, 7, 7, 960)   3840        ['block_15_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_15_expand_relu (ReLU)    (None, 7, 7, 960)    0           ['block_15_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_15_depthwise (DepthwiseC  (None, 7, 7, 960)   8640        ['block_15_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_15_depthwise_BN (BatchNo  (None, 7, 7, 960)   3840        ['block_15_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_15_depthwise_relu (ReLU)  (None, 7, 7, 960)   0           ['block_15_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_15_project (Conv2D)      (None, 7, 7, 160)    153600      ['block_15_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_15_project_BN (BatchNorm  (None, 7, 7, 160)   640         ['block_15_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " block_15_add (Add)             (None, 7, 7, 160)    0           ['block_14_add[0][0]',           \n",
            "                                                                  'block_15_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " block_16_expand (Conv2D)       (None, 7, 7, 960)    153600      ['block_15_add[0][0]']           \n",
            "                                                                                                  \n",
            " block_16_expand_BN (BatchNorma  (None, 7, 7, 960)   3840        ['block_16_expand[0][0]']        \n",
            " lization)                                                                                        \n",
            "                                                                                                  \n",
            " block_16_expand_relu (ReLU)    (None, 7, 7, 960)    0           ['block_16_expand_BN[0][0]']     \n",
            "                                                                                                  \n",
            " block_16_depthwise (DepthwiseC  (None, 7, 7, 960)   8640        ['block_16_expand_relu[0][0]']   \n",
            " onv2D)                                                                                           \n",
            "                                                                                                  \n",
            " block_16_depthwise_BN (BatchNo  (None, 7, 7, 960)   3840        ['block_16_depthwise[0][0]']     \n",
            " rmalization)                                                                                     \n",
            "                                                                                                  \n",
            " block_16_depthwise_relu (ReLU)  (None, 7, 7, 960)   0           ['block_16_depthwise_BN[0][0]']  \n",
            "                                                                                                  \n",
            " block_16_project (Conv2D)      (None, 7, 7, 320)    307200      ['block_16_depthwise_relu[0][0]']\n",
            "                                                                                                  \n",
            " block_16_project_BN (BatchNorm  (None, 7, 7, 320)   1280        ['block_16_project[0][0]']       \n",
            " alization)                                                                                       \n",
            "                                                                                                  \n",
            " Conv_1 (Conv2D)                (None, 7, 7, 1280)   409600      ['block_16_project_BN[0][0]']    \n",
            "                                                                                                  \n",
            " Conv_1_bn (BatchNormalization)  (None, 7, 7, 1280)  5120        ['Conv_1[0][0]']                 \n",
            "                                                                                                  \n",
            " out_relu (ReLU)                (None, 7, 7, 1280)   0           ['Conv_1_bn[0][0]']              \n",
            "                                                                                                  \n",
            " global_average_pooling2d (Glob  (None, 1280)        0           ['out_relu[0][0]']               \n",
            " alAveragePooling2D)                                                                              \n",
            "                                                                                                  \n",
            "==================================================================================================\n",
            "Total params: 2,257,984\n",
            "Trainable params: 2,223,872\n",
            "Non-trainable params: 34,112\n",
            "__________________________________________________________________________________________________\n"
          ]
        }
      ],
      "source": [
        "new_model.summary()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "2a0adcf5",
      "metadata": {
        "id": "2a0adcf5"
      },
      "source": [
        "Tous les workeurs doivent pouvoir accéder au modèle ainsi qu'à ses poids. <br />\n",
        "Une bonne pratique consiste à charger le modèle sur le driver puis à diffuser <br />\n",
        "ensuite les poids aux différents workeurs."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1cc53ff0",
      "metadata": {
        "id": "1cc53ff0"
      },
      "outputs": [],
      "source": [
        "brodcast_weights = sc.broadcast(new_model.get_weights())"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "brodcast_weights"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zYV8p7pSPywL",
        "outputId": "7c4e3ae8-9719-498d-a7e0-0cc378b3a715"
      },
      "id": "zYV8p7pSPywL",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<pyspark.broadcast.Broadcast at 0x7f18e844bfd0>"
            ]
          },
          "metadata": {},
          "execution_count": 23
        }
      ]
    },
    {
      "cell_type": "markdown",
      "id": "8bc0e34e",
      "metadata": {
        "id": "8bc0e34e"
      },
      "source": [
        "<u>Mettons cela sous forme de fonction</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "3fd51ba9",
      "metadata": {
        "id": "3fd51ba9"
      },
      "outputs": [],
      "source": [
        "def model_fn():\n",
        "    \"\"\"\n",
        "    Returns a MobileNetV2 model with top layer removed \n",
        "    and broadcasted pretrained weights.\n",
        "    \"\"\"\n",
        "    model = MobileNetV2(weights='imagenet',\n",
        "                        include_top=True,\n",
        "                        input_shape=(224, 224, 3))\n",
        "    for layer in model.layers:\n",
        "        layer.trainable = False\n",
        "    new_model = Model(inputs=model.input,\n",
        "                  outputs=model.layers[-2].output)\n",
        "    new_model.set_weights(brodcast_weights.value)\n",
        "    return new_model"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e5620876",
      "metadata": {
        "id": "e5620876"
      },
      "source": [
        "### 3.7.3 Définition du processus de chargement des images et application <br/>de leur featurisation à travers l'utilisation de pandas UDF\n",
        "\n",
        "Ce notebook définit la logique par étapes, jusqu'à Pandas UDF.\n",
        "\n",
        "<u>L'empilement des appels est la suivante</u> :\n",
        "\n",
        "- Pandas UDF\n",
        "  - featuriser une série d'images pd.Series\n",
        "   - prétraiter une image"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "dc4e5f69",
      "metadata": {
        "id": "dc4e5f69",
        "outputId": "6bb62856-7964-45de-b46c-60e776491a76",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stderr",
          "text": [
            "/usr/local/lib/python3.9/dist-packages/pyspark/sql/pandas/functions.py:389: UserWarning: In Python 3.6+ and Spark 3.0+, it is preferred to specify type hints for pandas UDF instead of specifying pandas UDF type which will be deprecated in the future releases. See SPARK-28264 for more details.\n",
            "  warnings.warn(\n"
          ]
        }
      ],
      "source": [
        "def preprocess(content):\n",
        "    \"\"\"\n",
        "    Preprocesses raw image bytes for prediction.\n",
        "    \"\"\"\n",
        "    img = Image.open(io.BytesIO(content)).resize([224, 224]) #Open image from binary objects content and resize it\n",
        "    arr = img_to_array(img) # Convert img to to array\n",
        "    return preprocess_input(arr) #Reshape to (224,224,3)\n",
        "\n",
        "def featurize_series(model, content_series):\n",
        "    \"\"\"\n",
        "    Featurize a pd.Series of raw images using the input model.\n",
        "    :return: a pd.Series of image features\n",
        "    \"\"\"\n",
        "    input = np.stack(content_series.map(preprocess))\n",
        "    preds = model.predict(input)\n",
        "    # For some layers, output features will be multi-dimensional tensors.\n",
        "    # We flatten the feature tensors to vectors for easier storage in Spark DataFrames.\n",
        "    output = [p.flatten() for p in preds]\n",
        "    return pd.Series(output)\n",
        "\n",
        "@pandas_udf('array<float>', PandasUDFType.SCALAR_ITER)\n",
        "def featurize_udf(content_series_iter):\n",
        "    '''\n",
        "    This method is a Scalar Iterator pandas UDF wrapping our featurization function.\n",
        "    The decorator specifies that this returns a Spark DataFrame column of type ArrayType(FloatType).\n",
        "\n",
        "    :param content_series_iter: This argument is an iterator over batches of data, where each batch\n",
        "                              is a pandas Series of image data.\n",
        "    '''\n",
        "    # With Scalar Iterator pandas UDFs, we can load the model once and then re-use it\n",
        "    # for multiple data batches.  This amortizes the overhead of loading big models.\n",
        "    model = model_fn() \n",
        "    #Weights broadcasting\n",
        "    model.set_weights(brodcast_weights.value)\n",
        "    for content_series in content_series_iter:\n",
        "        yield featurize_series(model, content_series)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "2bdf2ef9",
      "metadata": {
        "id": "2bdf2ef9"
      },
      "source": [
        "### 3.7.4 Exécution des actions d'extraction de features\n",
        "\n",
        "Les Pandas UDF, sur de grands enregistrements (par exemple, de très grandes images), <br />\n",
        "peuvent rencontrer des erreurs de type Out Of Memory (OOM).<br />\n",
        "Si vous rencontrez de telles erreurs dans la cellule ci-dessous, <br />\n",
        "essayez de réduire la taille du lot Arrow via 'maxRecordsPerBatch'\n",
        "\n",
        "Je n'utiliserai pas cette commande dans ce projet <br />\n",
        "et je laisse donc la commande en commentaire."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1f30d28c",
      "metadata": {
        "id": "1f30d28c"
      },
      "outputs": [],
      "source": [
        "# spark.conf.set(\"spark.sql.execution.arrow.maxRecordsPerBatch\", \"1024\")"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "70f8f95d",
      "metadata": {
        "id": "70f8f95d"
      },
      "source": [
        "Nous pouvons maintenant exécuter la featurisation sur l'ensemble de notre DataFrame Spark.<br />\n",
        "<u>REMARQUE</u> : Cela peut prendre beaucoup de temps, tout dépend du volume de données à traiter. <br />\n",
        "\n",
        "Notre jeu de données de **Test** contient **22819 images**. <br />\n",
        "Cependant, dans l'exécution en mode **local**, <br />\n",
        "nous <u>traiterons un ensemble réduit de **330 images**</u>."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "69c1767c",
      "metadata": {
        "id": "69c1767c",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 218
        },
        "outputId": "423721c9-e07b-4c83-9e11-47df4422b0f5"
      },
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-10-64fb9e1e1c76>\u001b[0m in \u001b[0;36m<cell line: 1>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m features_df = images.repartition(20).select(col(\"path\"),\n\u001b[1;32m      2\u001b[0m                                             \u001b[0mcol\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"label\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 3\u001b[0;31m                                             \u001b[0mfeaturize_udf\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"content\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0malias\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"features\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      4\u001b[0m                                            )\n",
            "\u001b[0;31mNameError\u001b[0m: name 'featurize_udf' is not defined"
          ]
        }
      ],
      "source": [
        "features_df = images.repartition(20).select(col(\"path\"),\n",
        "                                            col(\"label\"),\n",
        "                                            featurize_udf(\"content\").alias(\"features\")\n",
        "                                           )"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "eb5e83ec",
      "metadata": {
        "id": "eb5e83ec"
      },
      "source": [
        "<u>Rappel du PATH où seront inscrits les fichiers au format \"**parquet**\" <br />\n",
        "contenant nos résultats, à savoir, un DataFrame contenant 3 colonnes</u> :\n",
        " 1. Path des images\n",
        " 2. Label de l'image\n",
        " 3. Vecteur de caractéristiques de l'image"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "67fcdb0f",
      "metadata": {
        "id": "67fcdb0f",
        "outputId": "bffd48c1-08c2-4ac2-a1dc-ed2e162afbcd",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "/content/drive/MyDrive/Colab Notebooks/P8-Fruits/P8_Mode_opératoire/data/Results\n"
          ]
        }
      ],
      "source": [
        "print(PATH_Result)"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "type(images)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-ODt3F1IxdKM",
        "outputId": "5dd2c19d-00dc-40b0-e19f-90293dc9ca78"
      },
      "id": "-ODt3F1IxdKM",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "pyspark.sql.dataframe.DataFrame"
            ]
          },
          "metadata": {},
          "execution_count": 28
        }
      ]
    },
    {
      "cell_type": "markdown",
      "id": "f8901db3",
      "metadata": {
        "id": "f8901db3"
      },
      "source": [
        "<u>Enregistrement des données traitées au format \"**parquet**\"</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "95d07466",
      "metadata": {
        "id": "95d07466"
      },
      "outputs": [],
      "source": [
        "features_df.write.mode(\"overwrite\").parquet(PATH_Result)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "f9506f21",
      "metadata": {
        "id": "f9506f21"
      },
      "source": [
        "## 3.8 Chargement des données enregistrées et validation du résultat\n",
        "\n",
        "<u>On charge les données fraichement enregistrées dans un **DataFrame Pandas**</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "19243bf5",
      "metadata": {
        "id": "19243bf5"
      },
      "outputs": [],
      "source": [
        "df = pd.read_parquet(PATH_Result, engine='pyarrow')"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "27f15070",
      "metadata": {
        "id": "27f15070"
      },
      "source": [
        "<u>On affiche les 5 premières lignes du DataFrame</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "8a1bcdeb",
      "metadata": {
        "id": "8a1bcdeb",
        "outputId": "53e65787-4a77-45ae-dd52-c0e5d7d1ca41",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 206
        }
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "                                                path             label  \\\n",
              "0  file:/content/drive/MyDrive/Colab Notebooks/P8...            pear_1   \n",
              "1  file:/content/drive/MyDrive/Colab Notebooks/P8...  apple_braeburn_1   \n",
              "2  file:/content/drive/MyDrive/Colab Notebooks/P8...   zucchini_dark_1   \n",
              "3  file:/content/drive/MyDrive/Colab Notebooks/P8...  apple_braeburn_1   \n",
              "4  file:/content/drive/MyDrive/Colab Notebooks/P8...   zucchini_dark_1   \n",
              "\n",
              "                                            features  \n",
              "0  [1.0626231, 0.11354711, 0.0, 0.0, 0.0, 0.05484...  \n",
              "1  [0.04310427, 0.066478096, 0.0, 0.0, 0.0, 0.554...  \n",
              "2  [0.17094472, 0.008105326, 0.7820679, 0.0, 1.13...  \n",
              "3  [0.120135434, 0.014146357, 0.0, 0.0, 0.0, 0.82...  \n",
              "4  [0.2212768, 0.0, 0.34022278, 0.030905694, 2.84...  "
            ],
            "text/html": [
              "\n",
              "  <div id=\"df-b8071254-7a71-4de7-8eb5-0121caadd2ef\">\n",
              "    <div class=\"colab-df-container\">\n",
              "      <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>path</th>\n",
              "      <th>label</th>\n",
              "      <th>features</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>file:/content/drive/MyDrive/Colab Notebooks/P8...</td>\n",
              "      <td>pear_1</td>\n",
              "      <td>[1.0626231, 0.11354711, 0.0, 0.0, 0.0, 0.05484...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>file:/content/drive/MyDrive/Colab Notebooks/P8...</td>\n",
              "      <td>apple_braeburn_1</td>\n",
              "      <td>[0.04310427, 0.066478096, 0.0, 0.0, 0.0, 0.554...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>file:/content/drive/MyDrive/Colab Notebooks/P8...</td>\n",
              "      <td>zucchini_dark_1</td>\n",
              "      <td>[0.17094472, 0.008105326, 0.7820679, 0.0, 1.13...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>file:/content/drive/MyDrive/Colab Notebooks/P8...</td>\n",
              "      <td>apple_braeburn_1</td>\n",
              "      <td>[0.120135434, 0.014146357, 0.0, 0.0, 0.0, 0.82...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>file:/content/drive/MyDrive/Colab Notebooks/P8...</td>\n",
              "      <td>zucchini_dark_1</td>\n",
              "      <td>[0.2212768, 0.0, 0.34022278, 0.030905694, 2.84...</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>\n",
              "      <button class=\"colab-df-convert\" onclick=\"convertToInteractive('df-b8071254-7a71-4de7-8eb5-0121caadd2ef')\"\n",
              "              title=\"Convert this dataframe to an interactive table.\"\n",
              "              style=\"display:none;\">\n",
              "        \n",
              "  <svg xmlns=\"http://www.w3.org/2000/svg\" height=\"24px\"viewBox=\"0 0 24 24\"\n",
              "       width=\"24px\">\n",
              "    <path d=\"M0 0h24v24H0V0z\" fill=\"none\"/>\n",
              "    <path d=\"M18.56 5.44l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94zm-11 1L8.5 8.5l.94-2.06 2.06-.94-2.06-.94L8.5 2.5l-.94 2.06-2.06.94zm10 10l.94 2.06.94-2.06 2.06-.94-2.06-.94-.94-2.06-.94 2.06-2.06.94z\"/><path d=\"M17.41 7.96l-1.37-1.37c-.4-.4-.92-.59-1.43-.59-.52 0-1.04.2-1.43.59L10.3 9.45l-7.72 7.72c-.78.78-.78 2.05 0 2.83L4 21.41c.39.39.9.59 1.41.59.51 0 1.02-.2 1.41-.59l7.78-7.78 2.81-2.81c.8-.78.8-2.07 0-2.86zM5.41 20L4 18.59l7.72-7.72 1.47 1.35L5.41 20z\"/>\n",
              "  </svg>\n",
              "      </button>\n",
              "      \n",
              "  <style>\n",
              "    .colab-df-container {\n",
              "      display:flex;\n",
              "      flex-wrap:wrap;\n",
              "      gap: 12px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert {\n",
              "      background-color: #E8F0FE;\n",
              "      border: none;\n",
              "      border-radius: 50%;\n",
              "      cursor: pointer;\n",
              "      display: none;\n",
              "      fill: #1967D2;\n",
              "      height: 32px;\n",
              "      padding: 0 0 0 0;\n",
              "      width: 32px;\n",
              "    }\n",
              "\n",
              "    .colab-df-convert:hover {\n",
              "      background-color: #E2EBFA;\n",
              "      box-shadow: 0px 1px 2px rgba(60, 64, 67, 0.3), 0px 1px 3px 1px rgba(60, 64, 67, 0.15);\n",
              "      fill: #174EA6;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert {\n",
              "      background-color: #3B4455;\n",
              "      fill: #D2E3FC;\n",
              "    }\n",
              "\n",
              "    [theme=dark] .colab-df-convert:hover {\n",
              "      background-color: #434B5C;\n",
              "      box-shadow: 0px 1px 3px 1px rgba(0, 0, 0, 0.15);\n",
              "      filter: drop-shadow(0px 1px 2px rgba(0, 0, 0, 0.3));\n",
              "      fill: #FFFFFF;\n",
              "    }\n",
              "  </style>\n",
              "\n",
              "      <script>\n",
              "        const buttonEl =\n",
              "          document.querySelector('#df-b8071254-7a71-4de7-8eb5-0121caadd2ef button.colab-df-convert');\n",
              "        buttonEl.style.display =\n",
              "          google.colab.kernel.accessAllowed ? 'block' : 'none';\n",
              "\n",
              "        async function convertToInteractive(key) {\n",
              "          const element = document.querySelector('#df-b8071254-7a71-4de7-8eb5-0121caadd2ef');\n",
              "          const dataTable =\n",
              "            await google.colab.kernel.invokeFunction('convertToInteractive',\n",
              "                                                     [key], {});\n",
              "          if (!dataTable) return;\n",
              "\n",
              "          const docLinkHtml = 'Like what you see? Visit the ' +\n",
              "            '<a target=\"_blank\" href=https://colab.research.google.com/notebooks/data_table.ipynb>data table notebook</a>'\n",
              "            + ' to learn more about interactive tables.';\n",
              "          element.innerHTML = '';\n",
              "          dataTable['output_type'] = 'display_data';\n",
              "          await google.colab.output.renderOutput(dataTable, element);\n",
              "          const docLink = document.createElement('div');\n",
              "          docLink.innerHTML = docLinkHtml;\n",
              "          element.appendChild(docLink);\n",
              "        }\n",
              "      </script>\n",
              "    </div>\n",
              "  </div>\n",
              "  "
            ]
          },
          "metadata": {},
          "execution_count": 12
        }
      ],
      "source": [
        "df.head()"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "e2794fca",
      "metadata": {
        "id": "e2794fca"
      },
      "source": [
        "<u>On valide que la dimension du vecteur de caractéristiques des images est bien de dimension 1280</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "0bb933b9",
      "metadata": {
        "id": "0bb933b9",
        "outputId": "eb3dc40f-b4bb-41ef-dfcd-601f4eeef894",
        "colab": {
          "base_uri": "https://localhost:8080/"
        }
      },
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "(1280,)"
            ]
          },
          "metadata": {},
          "execution_count": 13
        }
      ],
      "source": [
        "df.loc[0,'features'].shape"
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df1 = spark.read.parquet(PATH_Result)"
      ],
      "metadata": {
        "id": "6YQ-oi5WxD3x"
      },
      "id": "6YQ-oi5WxD3x",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df1.printSchema()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Yl8Xxl9Y1hQj",
        "outputId": "2d89ad82-5aad-4a9b-fdd6-f7f7435ccd88"
      },
      "id": "Yl8Xxl9Y1hQj",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "root\n",
            " |-- path: string (nullable = true)\n",
            " |-- label: string (nullable = true)\n",
            " |-- features: array (nullable = true)\n",
            " |    |-- element: float (containsNull = true)\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 3.9 Standard scalering image features"
      ],
      "metadata": {
        "id": "0oA1Ta7vhu6R"
      },
      "id": "0oA1Ta7vhu6R"
    },
    {
      "cell_type": "code",
      "source": [
        "from pyspark.ml.linalg import VectorUDT, Vectors\n",
        "from pyspark.sql.functions import udf\n",
        "from pyspark.sql.types import ArrayType, DoubleType\n",
        "\n",
        "# define a UDF for converting numpy arrays to PySpark VectorUDT objects\n",
        "numpy_to_vector = udf(lambda a: Vectors.dense(a), returnType=VectorUDT())\n",
        "\n",
        "# apply the UDF to the 'features' column to create a new column of VectorUDT objects\n",
        "df1 = df1.withColumn('features_vector', numpy_to_vector(df1['features']))\n",
        "df1 = df1.withColumn('label_red', element_at(split(df1['label'], '_'),1))\n",
        "# drop the original 'features' column\n",
        "#df = df.drop('features')"
      ],
      "metadata": {
        "id": "6lyAv4zq2Xjk"
      },
      "id": "6lyAv4zq2Xjk",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "df1.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ThLtKwUo8Ini",
        "outputId": "7b6d5aa6-3cc7-4b5b-a4a3-1f30240029d9"
      },
      "id": "ThLtKwUo8Ini",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "+--------------------+---------------+--------------------+--------------------+---------+\n",
            "|                path|          label|            features|     features_vector|label_red|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+\n",
            "|file:/content/dri...|     cucumber_3|[0.47135264, 0.64...|[0.47135263681411...| cucumber|\n",
            "|file:/content/dri...|         pear_3|[1.4832611, 0.0, ...|[1.48326110839843...|     pear|\n",
            "|file:/content/dri...|         pear_1|[1.0241469, 0.115...|[1.02414691448211...|     pear|\n",
            "|file:/content/dri...|     zucchini_1|[0.0, 0.70493853,...|[0.0,0.7049385309...| zucchini|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.34860915, 0.0,...|[0.34860914945602...| zucchini|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.17964712, 0.0,...|[0.17964711785316...| zucchini|\n",
            "|file:/content/dri...|        apple_6|[0.022195164, 0.3...|[0.02219516411423...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.10118419, 0.85...|[0.10118418931961...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.013050653, 0.4...|[0.01305065304040...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.39738423, 0.45...|[0.39738422632217...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.033550497, 0.4...|[0.03355049714446...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.021785308, 1.2...|[0.02178530767560...|    apple|\n",
            "|file:/content/dri...|        apple_6|[0.053463314, 1.2...|[0.05346331372857...|    apple|\n",
            "|file:/content/dri...|     cucumber_3|[0.84237564, 0.76...|[0.84237563610076...| cucumber|\n",
            "|file:/content/dri...|     cucumber_1|[1.6268797, 0.010...|[1.62687969207763...| cucumber|\n",
            "|file:/content/dri...|     zucchini_1|[0.0019634208, 0....|[0.00196342077106...| zucchini|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.13799514, 0.0,...|[0.13799513876438...| zucchini|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.07329515, 0.0,...|[0.07329515367746...| zucchini|\n",
            "|file:/content/dri...|       carrot_1|[0.2981051, 0.160...|[0.29810509085655...|   carrot|\n",
            "|file:/content/dri...|        apple_6|[0.0, 0.19224256,...|[0.0,0.1922425627...|    apple|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+\n",
            "only showing top 20 rows\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "scaler = StandardScaler(\n",
        "    inputCol = 'features_vector', \n",
        "    outputCol = 'scaledFeatures',\n",
        "    withMean = True,\n",
        "    withStd = True\n",
        ").fit(df1)\n",
        "\n",
        "# when we transform the dataframe, the old\n",
        "# feature will still remain in it\n",
        "df_scaled = scaler.transform(df1)\n",
        "df_scaled.show(6)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "7zYqiS7Yh-3m",
        "outputId": "a7e2e6e5-9639-4c39-e0c0-5087f8501fd9"
      },
      "id": "7zYqiS7Yh-3m",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+\n",
            "|                path|          label|            features|     features_vector|label_red|      scaledFeatures|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+\n",
            "|file:/content/dri...|     cucumber_3|[0.47135264, 0.64...|[0.47135263681411...| cucumber|[0.42540941346383...|\n",
            "|file:/content/dri...|         pear_3|[1.4832611, 0.0, ...|[1.48326110839843...|     pear|[2.87794599661208...|\n",
            "|file:/content/dri...|         pear_1|[1.0241469, 0.115...|[1.02414691448211...|     pear|[1.76520271197543...|\n",
            "|file:/content/dri...|     zucchini_1|[0.0, 0.70493853,...|[0.0,0.7049385309...| zucchini|[-0.7169958710181...|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.34860915, 0.0,...|[0.34860914945602...| zucchini|[0.12791917443303...|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.17964712, 0.0,...|[0.17964711785316...| zucchini|[-0.2815897636829...|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+\n",
            "only showing top 6 rows\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "df_scaled.printSchema()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "gZYIsykV6g_a",
        "outputId": "d602ea45-b693-462b-f00a-99224d9e6426"
      },
      "id": "gZYIsykV6g_a",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "root\n",
            " |-- path: string (nullable = true)\n",
            " |-- label: string (nullable = true)\n",
            " |-- features: array (nullable = true)\n",
            " |    |-- element: float (containsNull = true)\n",
            " |-- features_vector: vector (nullable = true)\n",
            " |-- label_red: string (nullable = true)\n",
            " |-- scaledFeatures: vector (nullable = true)\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "## 3.10 Performing PCA"
      ],
      "metadata": {
        "id": "f9fG7puLk_fJ"
      },
      "id": "f9fG7puLk_fJ"
    },
    {
      "cell_type": "code",
      "source": [
        "n_components = 20\n",
        "pca = PCA(\n",
        "    k = n_components, \n",
        "    inputCol = 'scaledFeatures', \n",
        "    outputCol = 'pcaFeatures'\n",
        ").fit(df_scaled)\n",
        "print('Cumulative variance', pca.explainedVariance.toArray().sum())"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 235
        },
        "id": "yA96VxQElGlQ",
        "outputId": "67d2f476-17a7-4807-feec-d9b8c6b80a11"
      },
      "id": "yA96VxQElGlQ",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "error",
          "ename": "NameError",
          "evalue": "ignored",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)",
            "\u001b[0;32m<ipython-input-1-1754d21ee66d>\u001b[0m in \u001b[0;36m<cell line: 2>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      1\u001b[0m \u001b[0mn_components\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m20\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 2\u001b[0;31m pca = PCA(\n\u001b[0m\u001b[1;32m      3\u001b[0m     \u001b[0mk\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mn_components\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m     \u001b[0minputCol\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'scaledFeatures'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      5\u001b[0m     \u001b[0moutputCol\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m'pcaFeatures'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'PCA' is not defined"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "variances = pca.explainedVariance.toArray()\n",
        "# Compute the cumulative explained variance\n",
        "cumulative_variances = np.cumsum(variances)\n",
        "\n",
        "# Plot the explained variance ratios\n",
        "plt.bar(range(1, n_components + 1), variances, label='Individual explained variance')\n",
        "plt.plot(range(1, n_components + 1), cumulative_variances,'-s', label='Cumulative explained variance', color='red')\n",
        "plt.xlabel('Number of principal components')\n",
        "plt.ylabel('Explained variance ratio')\n",
        "plt.legend(loc='best')\n",
        "plt.show()"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 449
        },
        "id": "3fLfP7yNmQtQ",
        "outputId": "8e04eef1-f3cd-43b8-bf7d-4d543ce64828"
      },
      "id": "3fLfP7yNmQtQ",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 640x480 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "n_components = 2\n",
        "Pca = PCA(\n",
        "    k = n_components, \n",
        "    inputCol = 'scaledFeatures', \n",
        "    outputCol = 'pcaFeatures'\n",
        ").fit(df_scaled)\n",
        "\n",
        "df_pca = Pca.transform(df_scaled)\n",
        "print('Explained Variance Ratio', Pca.explainedVariance.toArray())\n",
        "df_pca.show(6)\n",
        "X_pca = df_pca.rdd.map(lambda row: row.pcaFeatures).collect()\n",
        "X_pca = np.array(X_pca)"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "9-eCglmzmMXq",
        "outputId": "2ed3a98e-a22b-4367-c231-08ca1eef2ce1"
      },
      "id": "9-eCglmzmMXq",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Explained Variance Ratio [0.15652207 0.07429275]\n",
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+--------------------+\n",
            "|                path|          label|            features|     features_vector|label_red|      scaledFeatures|         pcaFeatures|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+--------------------+\n",
            "|file:/content/dri...|     cucumber_3|[0.47135264, 0.64...|[0.47135263681411...| cucumber|[0.42540941346383...|[11.3596894459071...|\n",
            "|file:/content/dri...|         pear_3|[1.4832611, 0.0, ...|[1.48326110839843...|     pear|[2.87794599661208...|[-4.1495691193120...|\n",
            "|file:/content/dri...|         pear_1|[1.0241469, 0.115...|[1.02414691448211...|     pear|[1.76520271197543...|[-7.5617186423112...|\n",
            "|file:/content/dri...|     zucchini_1|[0.0, 0.70493853,...|[0.0,0.7049385309...| zucchini|[-0.7169958710181...|[19.1052957723427...|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.34860915, 0.0,...|[0.34860914945602...| zucchini|[0.12791917443303...|[27.9604274490659...|\n",
            "|file:/content/dri...|zucchini_dark_1|[0.17964712, 0.0,...|[0.17964711785316...| zucchini|[-0.2815897636829...|[30.0047288403945...|\n",
            "+--------------------+---------------+--------------------+--------------------+---------+--------------------+--------------------+\n",
            "only showing top 6 rows\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "source": [
        "# change default style figure and font size\n",
        "def plot_pca(X, y):\n",
        "    \"\"\"Plot of the 2-dimensional image features\"\"\"\n",
        "    labels = np.array(df2.select('label').rdd.flatMap(lambda x: x).collect())\n",
        "    u_labels = np.unique(labels)\n",
        "\n",
        "    for i in u_labels:\n",
        "        plt.scatter(X[labels == i , 0] , X[labels == i , 1] , label = f'{i}')\n",
        "    plt.title(f'PCA Scatter plot', fontsize=14 )\n",
        "    plt.legend(loc='upper right')# bbox_to_anchor=(0.5, -0.2), ncol=len(u_labels)) #,\n",
        "    plt.show()"
      ],
      "metadata": {
        "id": "e0BAe4bV-djr"
      },
      "id": "e0BAe4bV-djr",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "source": [
        "plot_pca(X_pca, \"label_red\")"
      ],
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 569
        },
        "id": "sMoFTtrF_PeV",
        "outputId": "b7f7069a-5f7b-4822-d226-3005cd362005"
      },
      "id": "sMoFTtrF_PeV",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 800x600 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {}
        }
      ]
    },
    {
      "cell_type": "markdown",
      "source": [
        "Save PCA dataframe"
      ],
      "metadata": {
        "id": "2SD1mEXzvk7Y"
      },
      "id": "2SD1mEXzvk7Y"
    },
    {
      "cell_type": "code",
      "source": [
        "PATH_pca = PATH+'/data/Results/PcaFeatures'\n",
        "df_pca.write.mode(\"overwrite\").parquet(PATH_pca)"
      ],
      "metadata": {
        "id": "QbHC6suDvCoo"
      },
      "id": "QbHC6suDvCoo",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "id": "efe5348d",
      "metadata": {
        "id": "efe5348d"
      },
      "source": [
        "Nous venons de valider le processus sur un jeu de données allégé en local <br />\n",
        "où nous avons simulé un cluster de machines en répartissant la charge de travail <br />\n",
        "sur différents cœurs de processeur au sein d'une même machine.\n",
        "\n",
        "Nous allons maintenant généraliser le processus en déployant notre solution <br />\n",
        "sur un réel cluster de machines et nous travaillerons désormais sur la totalité <br />\n",
        "des 22819 images de notre dossier \"Test\"."
      ]
    },
    {
      "cell_type": "markdown",
      "id": "4e759c1e",
      "metadata": {
        "id": "4e759c1e"
      },
      "source": [
        "### 4.10.1 Démarrage de la session Spark"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "e5f0fbe1",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "e5f0fbe1",
        "outputId": "cc838294-1257-4d72-8adb-162085a69b4e"
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Starting Spark application\n"
          ]
        },
        {
          "data": {
            "text/html": [
              "<table>\n",
              "<tr><th>ID</th><th>YARN Application ID</th><th>Kind</th><th>State</th><th>Spark UI</th><th>Driver log</th><th>Current session?</th></tr><tr><td>0</td><td>application_1626050279029_0001</td><td>pyspark</td><td>idle</td><td><a target=\"_blank\" href=\"http://ip-172-31-10-195.eu-west-3.compute.internal:20888/proxy/application_1626050279029_0001/\">Link</a></td><td><a target=\"_blank\" href=\"http://ip-172-31-2-58.eu-west-3.compute.internal:8042/node/containerlogs/container_1626050279029_0001_01_000001/livy\">Link</a></td><td>✔</td></tr></table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "SparkSession available as 'spark'.\n"
          ]
        },
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# L'exécution de cette cellule démarre l'application Spark"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "3aba202f",
      "metadata": {
        "id": "3aba202f"
      },
      "source": [
        "<u>Affichage des informations sur la session en cours et liens vers Spark UI</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "fb788991",
      "metadata": {
        "id": "fb788991",
        "outputId": "fdac2f1d-fd5c-4a9b-e14e-9c6d8947f88c"
      },
      "outputs": [
        {
          "data": {
            "text/html": [
              "Current session configs: <tt>{'driverMemory': '1000M', 'executorCores': 2, 'proxyUser': 'jovyan', 'kind': 'pyspark'}</tt><br>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "text/html": [
              "<table>\n",
              "<tr><th>ID</th><th>YARN Application ID</th><th>Kind</th><th>State</th><th>Spark UI</th><th>Driver log</th><th>Current session?</th></tr><tr><td>0</td><td>application_1626050279029_0001</td><td>pyspark</td><td>idle</td><td><a target=\"_blank\" href=\"http://ip-172-31-10-195.eu-west-3.compute.internal:20888/proxy/application_1626050279029_0001/\">Link</a></td><td><a target=\"_blank\" href=\"http://ip-172-31-2-58.eu-west-3.compute.internal:8042/node/containerlogs/container_1626050279029_0001_01_000001/livy\">Link</a></td><td>✔</td></tr></table>"
            ],
            "text/plain": [
              "<IPython.core.display.HTML object>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "%%info"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "27ac9832",
      "metadata": {
        "id": "27ac9832"
      },
      "source": [
        "### 4.10.2 Installation des packages\n",
        "\n",
        "Les packages nécessaires ont été installé via l'étape de **bootstrap** à l'instanciation du serveur.\n",
        "\n",
        "### 4.10.3 Import des librairies"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "ad562eab",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "ad562eab",
        "outputId": "b2e0eaac-a13b-4c92-a053-62b0e4c3a2dd"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import pandas as pd\n",
        "import numpy as np\n",
        "import io\n",
        "import os\n",
        "import tensorflow as tf\n",
        "from PIL import Image\n",
        "from tensorflow.keras.applications.mobilenet_v2 import MobileNetV2, preprocess_input\n",
        "from tensorflow.keras.preprocessing.image import img_to_array\n",
        "from tensorflow.keras import Model\n",
        "from pyspark.sql.functions import col, pandas_udf, PandasUDFType, element_at, split"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "83663cbd",
      "metadata": {
        "id": "83663cbd"
      },
      "source": [
        "### 4.10.4 Définition des PATH pour charger les images et enregistrer les résultats\n",
        "\n",
        "Nous accédons directement à nos **données sur S3** comme si elles étaient **stockées localement**."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "46be859d",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "46be859d",
        "outputId": "3ffc80fe-e628-40ea-854e-06cebfcb5a75"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "PATH:        s3://p8-data\n",
            "PATH_Data:   s3://p8-data/Test\n",
            "PATH_Result: s3://p8-data/Results"
          ]
        }
      ],
      "source": [
        "PATH = 's3://p8-data'\n",
        "PATH_Data = PATH+'/Test'\n",
        "PATH_Result = PATH+'/Results'\n",
        "print('PATH:        '+\\\n",
        "      PATH+'\\nPATH_Data:   '+\\\n",
        "      PATH_Data+'\\nPATH_Result: '+PATH_Result)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "cf883c20",
      "metadata": {
        "id": "cf883c20"
      },
      "source": [
        "### 4.10.5 Traitement des données"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "2ffe93f5",
      "metadata": {
        "id": "2ffe93f5"
      },
      "source": [
        "#### 4.10.5.1 Chargement des données"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "7e4b319a",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "7e4b319a",
        "outputId": "5a256601-d4bc-4802-e118-f3f36bae54d2"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "images = spark.read.format(\"binaryFile\") \\\n",
        "  .option(\"pathGlobFilter\", \"*.jpg\") \\\n",
        "  .option(\"recursiveFileLookup\", \"true\") \\\n",
        "  .load(PATH_Data)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "16bfeb4d",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "16bfeb4d",
        "outputId": "3c2ca103-25e7-4091-b1fa-380ba0704f6e"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "+--------------------+-------------------+------+--------------------+\n",
            "|                path|   modificationTime|length|             content|\n",
            "+--------------------+-------------------+------+--------------------+\n",
            "|s3://p8-data/Test...|2021-07-03 09:00:08|  7353|[FF D8 FF E0 00 1...|\n",
            "|s3://p8-data/Test...|2021-07-03 09:00:08|  7350|[FF D8 FF E0 00 1...|\n",
            "|s3://p8-data/Test...|2021-07-03 09:00:08|  7349|[FF D8 FF E0 00 1...|\n",
            "|s3://p8-data/Test...|2021-07-03 09:00:08|  7348|[FF D8 FF E0 00 1...|\n",
            "|s3://p8-data/Test...|2021-07-03 09:00:09|  7328|[FF D8 FF E0 00 1...|\n",
            "+--------------------+-------------------+------+--------------------+\n",
            "only showing top 5 rows"
          ]
        }
      ],
      "source": [
        "images.show(5)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "8b32ac34",
      "metadata": {
        "id": "8b32ac34"
      },
      "source": [
        "<u>Je ne conserve que le **path** de l'image et j'ajoute <br />\n",
        "    une colonne contenant les **labels** de chaque image</u> :"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a52ab808",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "a52ab808",
        "outputId": "03149032-68b4-476b-9127-248afc5356e8"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "root\n",
            " |-- path: string (nullable = true)\n",
            " |-- modificationTime: timestamp (nullable = true)\n",
            " |-- length: long (nullable = true)\n",
            " |-- content: binary (nullable = true)\n",
            " |-- label: string (nullable = true)\n",
            "\n",
            "None\n",
            "+------------------------------------------+----------+\n",
            "|path                                      |label     |\n",
            "+------------------------------------------+----------+\n",
            "|s3://p8-data/Test/Watermelon/r_106_100.jpg|Watermelon|\n",
            "|s3://p8-data/Test/Watermelon/r_109_100.jpg|Watermelon|\n",
            "|s3://p8-data/Test/Watermelon/r_108_100.jpg|Watermelon|\n",
            "|s3://p8-data/Test/Watermelon/r_107_100.jpg|Watermelon|\n",
            "|s3://p8-data/Test/Watermelon/r_95_100.jpg |Watermelon|\n",
            "+------------------------------------------+----------+\n",
            "only showing top 5 rows\n",
            "\n",
            "None"
          ]
        }
      ],
      "source": [
        "images = images.withColumn('label', element_at(split(images['path'], '/'),-2))\n",
        "print(images.printSchema())\n",
        "print(images.select('path','label').show(5,False))"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "8f15b199",
      "metadata": {
        "id": "8f15b199"
      },
      "source": [
        "#### 4.10.5.2 Préparation du modèle"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "ec7c7165",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "ec7c7165",
        "outputId": "2adadb0d-599a-4371-f3da-bea11a180217"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Downloading data from https://storage.googleapis.com/tensorflow/keras-applications/mobilenet_v2/mobilenet_v2_weights_tf_dim_ordering_tf_kernels_1.0_224.h5\n",
            "\r",
            "    8192/14536120 [..............................] - ETA: 0s\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
            " 4202496/14536120 [=======>......................] - ETA: 0s\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\b\r",
            "14540800/14536120 [==============================] - 0s 0us/step"
          ]
        }
      ],
      "source": [
        "model = MobileNetV2(weights='imagenet',\n",
        "                    include_top=True,\n",
        "                    input_shape=(224, 224, 3))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1b9bc650",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "1b9bc650",
        "outputId": "7996a048-907c-473a-be69-92e547b037c7"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "new_model = Model(inputs=model.input,\n",
        "                  outputs=model.layers[-2].output)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "a0d497f2",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "a0d497f2",
        "outputId": "a81c63bd-d231-4753-c793-a7c65ba434bb"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "brodcast_weights = sc.broadcast(new_model.get_weights())"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "1bc0bf14",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "collapsed": true,
        "id": "1bc0bf14",
        "outputId": "e927eec0-be14-4d07-feae-3c637391b69e"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"model\"\n",
            "__________________________________________________________________________________________________\n",
            "Layer (type)                    Output Shape         Param #     Connected to                     \n",
            "==================================================================================================\n",
            "input_1 (InputLayer)            [(None, 224, 224, 3) 0                                            \n",
            "__________________________________________________________________________________________________\n",
            "Conv1 (Conv2D)                  (None, 112, 112, 32) 864         input_1[0][0]                    \n",
            "__________________________________________________________________________________________________\n",
            "bn_Conv1 (BatchNormalization)   (None, 112, 112, 32) 128         Conv1[0][0]                      \n",
            "__________________________________________________________________________________________________\n",
            "Conv1_relu (ReLU)               (None, 112, 112, 32) 0           bn_Conv1[0][0]                   \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise (Depthw (None, 112, 112, 32) 288         Conv1_relu[0][0]                 \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise_BN (Bat (None, 112, 112, 32) 128         expanded_conv_depthwise[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_depthwise_relu (R (None, 112, 112, 32) 0           expanded_conv_depthwise_BN[0][0] \n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_project (Conv2D)  (None, 112, 112, 16) 512         expanded_conv_depthwise_relu[0][0\n",
            "__________________________________________________________________________________________________\n",
            "expanded_conv_project_BN (Batch (None, 112, 112, 16) 64          expanded_conv_project[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand (Conv2D)         (None, 112, 112, 96) 1536        expanded_conv_project_BN[0][0]   \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand_BN (BatchNormali (None, 112, 112, 96) 384         block_1_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_1_expand_relu (ReLU)      (None, 112, 112, 96) 0           block_1_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_1_pad (ZeroPadding2D)     (None, 113, 113, 96) 0           block_1_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise (DepthwiseCon (None, 56, 56, 96)   864         block_1_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise_BN (BatchNorm (None, 56, 56, 96)   384         block_1_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_1_depthwise_relu (ReLU)   (None, 56, 56, 96)   0           block_1_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_1_project (Conv2D)        (None, 56, 56, 24)   2304        block_1_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_1_project_BN (BatchNormal (None, 56, 56, 24)   96          block_1_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand (Conv2D)         (None, 56, 56, 144)  3456        block_1_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand_BN (BatchNormali (None, 56, 56, 144)  576         block_2_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_2_expand_relu (ReLU)      (None, 56, 56, 144)  0           block_2_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise (DepthwiseCon (None, 56, 56, 144)  1296        block_2_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise_BN (BatchNorm (None, 56, 56, 144)  576         block_2_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_2_depthwise_relu (ReLU)   (None, 56, 56, 144)  0           block_2_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_2_project (Conv2D)        (None, 56, 56, 24)   3456        block_2_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_2_project_BN (BatchNormal (None, 56, 56, 24)   96          block_2_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_2_add (Add)               (None, 56, 56, 24)   0           block_1_project_BN[0][0]         \n",
            "                                                                 block_2_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand (Conv2D)         (None, 56, 56, 144)  3456        block_2_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand_BN (BatchNormali (None, 56, 56, 144)  576         block_3_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_3_expand_relu (ReLU)      (None, 56, 56, 144)  0           block_3_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_3_pad (ZeroPadding2D)     (None, 57, 57, 144)  0           block_3_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise (DepthwiseCon (None, 28, 28, 144)  1296        block_3_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise_BN (BatchNorm (None, 28, 28, 144)  576         block_3_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_3_depthwise_relu (ReLU)   (None, 28, 28, 144)  0           block_3_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_3_project (Conv2D)        (None, 28, 28, 32)   4608        block_3_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_3_project_BN (BatchNormal (None, 28, 28, 32)   128         block_3_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand (Conv2D)         (None, 28, 28, 192)  6144        block_3_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand_BN (BatchNormali (None, 28, 28, 192)  768         block_4_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_4_expand_relu (ReLU)      (None, 28, 28, 192)  0           block_4_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise (DepthwiseCon (None, 28, 28, 192)  1728        block_4_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise_BN (BatchNorm (None, 28, 28, 192)  768         block_4_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_4_depthwise_relu (ReLU)   (None, 28, 28, 192)  0           block_4_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_4_project (Conv2D)        (None, 28, 28, 32)   6144        block_4_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_4_project_BN (BatchNormal (None, 28, 28, 32)   128         block_4_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_4_add (Add)               (None, 28, 28, 32)   0           block_3_project_BN[0][0]         \n",
            "                                                                 block_4_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand (Conv2D)         (None, 28, 28, 192)  6144        block_4_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand_BN (BatchNormali (None, 28, 28, 192)  768         block_5_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_5_expand_relu (ReLU)      (None, 28, 28, 192)  0           block_5_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise (DepthwiseCon (None, 28, 28, 192)  1728        block_5_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise_BN (BatchNorm (None, 28, 28, 192)  768         block_5_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_5_depthwise_relu (ReLU)   (None, 28, 28, 192)  0           block_5_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_5_project (Conv2D)        (None, 28, 28, 32)   6144        block_5_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_5_project_BN (BatchNormal (None, 28, 28, 32)   128         block_5_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_5_add (Add)               (None, 28, 28, 32)   0           block_4_add[0][0]                \n",
            "                                                                 block_5_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand (Conv2D)         (None, 28, 28, 192)  6144        block_5_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand_BN (BatchNormali (None, 28, 28, 192)  768         block_6_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_6_expand_relu (ReLU)      (None, 28, 28, 192)  0           block_6_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_6_pad (ZeroPadding2D)     (None, 29, 29, 192)  0           block_6_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise (DepthwiseCon (None, 14, 14, 192)  1728        block_6_pad[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise_BN (BatchNorm (None, 14, 14, 192)  768         block_6_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_6_depthwise_relu (ReLU)   (None, 14, 14, 192)  0           block_6_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_6_project (Conv2D)        (None, 14, 14, 64)   12288       block_6_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_6_project_BN (BatchNormal (None, 14, 14, 64)   256         block_6_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand (Conv2D)         (None, 14, 14, 384)  24576       block_6_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand_BN (BatchNormali (None, 14, 14, 384)  1536        block_7_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_7_expand_relu (ReLU)      (None, 14, 14, 384)  0           block_7_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise (DepthwiseCon (None, 14, 14, 384)  3456        block_7_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise_BN (BatchNorm (None, 14, 14, 384)  1536        block_7_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_7_depthwise_relu (ReLU)   (None, 14, 14, 384)  0           block_7_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_7_project (Conv2D)        (None, 14, 14, 64)   24576       block_7_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_7_project_BN (BatchNormal (None, 14, 14, 64)   256         block_7_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_7_add (Add)               (None, 14, 14, 64)   0           block_6_project_BN[0][0]         \n",
            "                                                                 block_7_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand (Conv2D)         (None, 14, 14, 384)  24576       block_7_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand_BN (BatchNormali (None, 14, 14, 384)  1536        block_8_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_8_expand_relu (ReLU)      (None, 14, 14, 384)  0           block_8_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise (DepthwiseCon (None, 14, 14, 384)  3456        block_8_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise_BN (BatchNorm (None, 14, 14, 384)  1536        block_8_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_8_depthwise_relu (ReLU)   (None, 14, 14, 384)  0           block_8_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_8_project (Conv2D)        (None, 14, 14, 64)   24576       block_8_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_8_project_BN (BatchNormal (None, 14, 14, 64)   256         block_8_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_8_add (Add)               (None, 14, 14, 64)   0           block_7_add[0][0]                \n",
            "                                                                 block_8_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand (Conv2D)         (None, 14, 14, 384)  24576       block_8_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand_BN (BatchNormali (None, 14, 14, 384)  1536        block_9_expand[0][0]             \n",
            "__________________________________________________________________________________________________\n",
            "block_9_expand_relu (ReLU)      (None, 14, 14, 384)  0           block_9_expand_BN[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise (DepthwiseCon (None, 14, 14, 384)  3456        block_9_expand_relu[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise_BN (BatchNorm (None, 14, 14, 384)  1536        block_9_depthwise[0][0]          \n",
            "__________________________________________________________________________________________________\n",
            "block_9_depthwise_relu (ReLU)   (None, 14, 14, 384)  0           block_9_depthwise_BN[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_9_project (Conv2D)        (None, 14, 14, 64)   24576       block_9_depthwise_relu[0][0]     \n",
            "__________________________________________________________________________________________________\n",
            "block_9_project_BN (BatchNormal (None, 14, 14, 64)   256         block_9_project[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_9_add (Add)               (None, 14, 14, 64)   0           block_8_add[0][0]                \n",
            "                                                                 block_9_project_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand (Conv2D)        (None, 14, 14, 384)  24576       block_9_add[0][0]                \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand_BN (BatchNormal (None, 14, 14, 384)  1536        block_10_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_10_expand_relu (ReLU)     (None, 14, 14, 384)  0           block_10_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise (DepthwiseCo (None, 14, 14, 384)  3456        block_10_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise_BN (BatchNor (None, 14, 14, 384)  1536        block_10_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_10_depthwise_relu (ReLU)  (None, 14, 14, 384)  0           block_10_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_10_project (Conv2D)       (None, 14, 14, 96)   36864       block_10_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_10_project_BN (BatchNorma (None, 14, 14, 96)   384         block_10_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand (Conv2D)        (None, 14, 14, 576)  55296       block_10_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand_BN (BatchNormal (None, 14, 14, 576)  2304        block_11_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_11_expand_relu (ReLU)     (None, 14, 14, 576)  0           block_11_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise (DepthwiseCo (None, 14, 14, 576)  5184        block_11_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise_BN (BatchNor (None, 14, 14, 576)  2304        block_11_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_11_depthwise_relu (ReLU)  (None, 14, 14, 576)  0           block_11_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_11_project (Conv2D)       (None, 14, 14, 96)   55296       block_11_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_11_project_BN (BatchNorma (None, 14, 14, 96)   384         block_11_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_11_add (Add)              (None, 14, 14, 96)   0           block_10_project_BN[0][0]        \n",
            "                                                                 block_11_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand (Conv2D)        (None, 14, 14, 576)  55296       block_11_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand_BN (BatchNormal (None, 14, 14, 576)  2304        block_12_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_12_expand_relu (ReLU)     (None, 14, 14, 576)  0           block_12_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise (DepthwiseCo (None, 14, 14, 576)  5184        block_12_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise_BN (BatchNor (None, 14, 14, 576)  2304        block_12_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_12_depthwise_relu (ReLU)  (None, 14, 14, 576)  0           block_12_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_12_project (Conv2D)       (None, 14, 14, 96)   55296       block_12_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_12_project_BN (BatchNorma (None, 14, 14, 96)   384         block_12_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_12_add (Add)              (None, 14, 14, 96)   0           block_11_add[0][0]               \n",
            "                                                                 block_12_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand (Conv2D)        (None, 14, 14, 576)  55296       block_12_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand_BN (BatchNormal (None, 14, 14, 576)  2304        block_13_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_13_expand_relu (ReLU)     (None, 14, 14, 576)  0           block_13_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_13_pad (ZeroPadding2D)    (None, 15, 15, 576)  0           block_13_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise (DepthwiseCo (None, 7, 7, 576)    5184        block_13_pad[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise_BN (BatchNor (None, 7, 7, 576)    2304        block_13_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_13_depthwise_relu (ReLU)  (None, 7, 7, 576)    0           block_13_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_13_project (Conv2D)       (None, 7, 7, 160)    92160       block_13_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_13_project_BN (BatchNorma (None, 7, 7, 160)    640         block_13_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand (Conv2D)        (None, 7, 7, 960)    153600      block_13_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand_BN (BatchNormal (None, 7, 7, 960)    3840        block_14_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_14_expand_relu (ReLU)     (None, 7, 7, 960)    0           block_14_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise (DepthwiseCo (None, 7, 7, 960)    8640        block_14_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise_BN (BatchNor (None, 7, 7, 960)    3840        block_14_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_14_depthwise_relu (ReLU)  (None, 7, 7, 960)    0           block_14_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_14_project (Conv2D)       (None, 7, 7, 160)    153600      block_14_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_14_project_BN (BatchNorma (None, 7, 7, 160)    640         block_14_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_14_add (Add)              (None, 7, 7, 160)    0           block_13_project_BN[0][0]        \n",
            "                                                                 block_14_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand (Conv2D)        (None, 7, 7, 960)    153600      block_14_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand_BN (BatchNormal (None, 7, 7, 960)    3840        block_15_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_15_expand_relu (ReLU)     (None, 7, 7, 960)    0           block_15_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise (DepthwiseCo (None, 7, 7, 960)    8640        block_15_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise_BN (BatchNor (None, 7, 7, 960)    3840        block_15_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_15_depthwise_relu (ReLU)  (None, 7, 7, 960)    0           block_15_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_15_project (Conv2D)       (None, 7, 7, 160)    153600      block_15_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_15_project_BN (BatchNorma (None, 7, 7, 160)    640         block_15_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "block_15_add (Add)              (None, 7, 7, 160)    0           block_14_add[0][0]               \n",
            "                                                                 block_15_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand (Conv2D)        (None, 7, 7, 960)    153600      block_15_add[0][0]               \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand_BN (BatchNormal (None, 7, 7, 960)    3840        block_16_expand[0][0]            \n",
            "__________________________________________________________________________________________________\n",
            "block_16_expand_relu (ReLU)     (None, 7, 7, 960)    0           block_16_expand_BN[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise (DepthwiseCo (None, 7, 7, 960)    8640        block_16_expand_relu[0][0]       \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise_BN (BatchNor (None, 7, 7, 960)    3840        block_16_depthwise[0][0]         \n",
            "__________________________________________________________________________________________________\n",
            "block_16_depthwise_relu (ReLU)  (None, 7, 7, 960)    0           block_16_depthwise_BN[0][0]      \n",
            "__________________________________________________________________________________________________\n",
            "block_16_project (Conv2D)       (None, 7, 7, 320)    307200      block_16_depthwise_relu[0][0]    \n",
            "__________________________________________________________________________________________________\n",
            "block_16_project_BN (BatchNorma (None, 7, 7, 320)    1280        block_16_project[0][0]           \n",
            "__________________________________________________________________________________________________\n",
            "Conv_1 (Conv2D)                 (None, 7, 7, 1280)   409600      block_16_project_BN[0][0]        \n",
            "__________________________________________________________________________________________________\n",
            "Conv_1_bn (BatchNormalization)  (None, 7, 7, 1280)   5120        Conv_1[0][0]                     \n",
            "__________________________________________________________________________________________________\n",
            "out_relu (ReLU)                 (None, 7, 7, 1280)   0           Conv_1_bn[0][0]                  \n",
            "__________________________________________________________________________________________________\n",
            "global_average_pooling2d (Globa (None, 1280)         0           out_relu[0][0]                   \n",
            "==================================================================================================\n",
            "Total params: 2,257,984\n",
            "Trainable params: 2,223,872\n",
            "Non-trainable params: 34,112\n",
            "__________________________________________________________________________________________________"
          ]
        }
      ],
      "source": [
        "new_model.summary()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "be8fe2b9",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "be8fe2b9",
        "outputId": "0b09a155-fabc-490d-efb1-ed2ff7207984"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "def model_fn():\n",
        "    \"\"\"\n",
        "    Returns a MobileNetV2 model with top layer removed \n",
        "    and broadcasted pretrained weights.\n",
        "    \"\"\"\n",
        "    model = MobileNetV2(weights='imagenet',\n",
        "                        include_top=True,\n",
        "                        input_shape=(224, 224, 3))\n",
        "    for layer in model.layers:\n",
        "        layer.trainable = False\n",
        "    new_model = Model(inputs=model.input,\n",
        "                  outputs=model.layers[-2].output)\n",
        "    new_model.set_weights(brodcast_weights.value)\n",
        "    return new_model"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "c032f135",
      "metadata": {
        "id": "c032f135"
      },
      "source": [
        "#### 4.10.5.3 Définition du processus de chargement des images <br/> et application de leur featurisation à travers l'utilisation de pandas UDF"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "933100cf",
      "metadata": {
        "scrolled": true,
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "933100cf",
        "outputId": "adcc95ba-151d-44c3-cf09-b1f71972813a"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "/usr/lib/spark/python/lib/pyspark.zip/pyspark/sql/pandas/functions.py:392: UserWarning: In Python 3.6+ and Spark 3.0+, it is preferred to specify type hints for pandas UDF instead of specifying pandas UDF type which will be deprecated in the future releases. See SPARK-28264 for more details."
          ]
        }
      ],
      "source": [
        "def preprocess(content):\n",
        "    \"\"\"\n",
        "    Preprocesses raw image bytes for prediction.\n",
        "    \"\"\"\n",
        "    img = Image.open(io.BytesIO(content)).resize([224, 224])\n",
        "    arr = img_to_array(img)\n",
        "    return preprocess_input(arr)\n",
        "\n",
        "def featurize_series(model, content_series):\n",
        "    \"\"\"\n",
        "    Featurize a pd.Series of raw images using the input model.\n",
        "    :return: a pd.Series of image features\n",
        "    \"\"\"\n",
        "    input = np.stack(content_series.map(preprocess))\n",
        "    preds = model.predict(input)\n",
        "    # For some layers, output features will be multi-dimensional tensors.\n",
        "    # We flatten the feature tensors to vectors for easier storage in Spark DataFrames.\n",
        "    output = [p.flatten() for p in preds]\n",
        "    return pd.Series(output)\n",
        "\n",
        "@pandas_udf('array<float>', PandasUDFType.SCALAR_ITER)\n",
        "def featurize_udf(content_series_iter):\n",
        "    '''\n",
        "    This method is a Scalar Iterator pandas UDF wrapping our featurization function.\n",
        "    The decorator specifies that this returns a Spark DataFrame column of type ArrayType(FloatType).\n",
        "\n",
        "    :param content_series_iter: This argument is an iterator over batches of data, where each batch\n",
        "                              is a pandas Series of image data.\n",
        "    '''\n",
        "    # With Scalar Iterator pandas UDFs, we can load the model once and then re-use it\n",
        "    # for multiple data batches.  This amortizes the overhead of loading big models.\n",
        "    model = model_fn()\n",
        "    for content_series in content_series_iter:\n",
        "        yield featurize_series(model, content_series)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "f23206e8",
      "metadata": {
        "id": "f23206e8"
      },
      "source": [
        "#### 4.10.5.4 Exécutions des actions d'extractions de features"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "22d760c2",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "22d760c2",
        "outputId": "726980e7-363d-4467-cbde-0642dae1a413"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "# spark.conf.set(\"spark.sql.execution.arrow.maxRecordsPerBatch\", \"1024\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "5e07fd68",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "5e07fd68",
        "outputId": "da9ca07b-345b-4d91-dcbd-43a5d82aaf6d"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "features_df = images.repartition(24).select(col(\"path\"),\n",
        "                                            col(\"label\"),\n",
        "                                            featurize_udf(\"content\").alias(\"features\")\n",
        "                                           )"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "06a930b3",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "06a930b3",
        "outputId": "1dd6b923-2e8b-4de1-ea33-fa3371b1729b"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "s3://p8-data/Results"
          ]
        }
      ],
      "source": [
        "print(PATH_Result)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "7c53ddd5",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "7c53ddd5",
        "outputId": "cb0de832-84a4-48c9-bc32-bdf9d940e8af"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "features_df.write.mode(\"overwrite\").parquet(PATH_Result)"
      ]
    },
    {
      "cell_type": "markdown",
      "id": "1fe01b72",
      "metadata": {
        "id": "1fe01b72"
      },
      "source": [
        "### 4.10.6 Chargement des données enregistrées et validation du résultat"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "db18a784",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "db18a784",
        "outputId": "5e5c2b02-946f-4e88-a416-6f8c43a8fdb3"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "df = pd.read_parquet(PATH_Result, engine='pyarrow')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "d750d2a8",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "d750d2a8",
        "outputId": "e0829b27-5e66-471b-cb78-b35ece1b3fec"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "                                           path  ...                                           features\n",
            "0    s3://p8-data/Test/Watermelon/r_174_100.jpg  ...  [0.0059991637, 0.44703647, 0.0, 0.0, 3.3713572...\n",
            "1  s3://p8-data/Test/Pineapple Mini/128_100.jpg  ...  [0.0146466885, 4.080593, 0.055877004, 0.0, 0.0...\n",
            "2  s3://p8-data/Test/Pineapple Mini/137_100.jpg  ...  [0.0, 4.9659867, 0.0, 0.0, 0.0, 0.0, 0.5144821...\n",
            "3      s3://p8-data/Test/Watermelon/275_100.jpg  ...  [0.22511952, 0.07235509, 0.0, 0.0, 1.690149, 0...\n",
            "4      s3://p8-data/Test/Watermelon/271_100.jpg  ...  [0.3286234, 0.18830013, 0.0, 0.0, 1.9123534, 0...\n",
            "\n",
            "[5 rows x 3 columns]"
          ]
        }
      ],
      "source": [
        "df.head()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "b29205ff",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "b29205ff",
        "outputId": "6ddf9063-e3d8-4f92-826a-6a8915bd538b"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(1280,)"
          ]
        }
      ],
      "source": [
        "df.loc[0,'features'].shape"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "id": "4fba6455",
      "metadata": {
        "colab": {
          "referenced_widgets": [
            ""
          ]
        },
        "id": "4fba6455",
        "outputId": "5a06d1bd-f114-48dd-c5b9-f14caff4986d"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.jupyter.widget-view+json": {
              "model_id": "",
              "version_major": 2,
              "version_minor": 0
            },
            "text/plain": [
              "FloatProgress(value=0.0, bar_style='info', description='Progress:', layout=Layout(height='25px', width='50%'),…"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "(22688, 3)"
          ]
        }
      ],
      "source": [
        "df.shape"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3 (ipykernel)",
      "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": "3.9.12"
    },
    "toc": {
      "base_numbering": 1,
      "nav_menu": {},
      "number_sections": true,
      "sideBar": true,
      "skip_h1_title": false,
      "title_cell": "Table of Contents",
      "title_sidebar": "Contents",
      "toc_cell": false,
      "toc_position": {
        "height": "calc(100% - 180px)",
        "left": "10px",
        "top": "150px",
        "width": "432.4px"
      },
      "toc_section_display": true,
      "toc_window_display": true
    },
    "colab": {
      "provenance": [],
      "toc_visible": true,
      "collapsed_sections": [
        "0oA1Ta7vhu6R",
        "27ac9832",
        "83663cbd",
        "2ffe93f5",
        "8f15b199",
        "c032f135",
        "f23206e8",
        "1fe01b72"
      ],
      "gpuClass": "premium",
      "include_colab_link": true
    },
    "accelerator": "GPU",
    "gpuClass": "premium"
  },
  "nbformat": 4,
  "nbformat_minor": 5
}