Modelos_Base.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.5-final"
    },
    "orig_nbformat": 2,
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3",
      "language": "python"
    },
    "colab": {
      "name": "Modelos_Base.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "machine_shape": "hm",
      "include_colab_link": true
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/jdiazram/DEEP_GEOM/blob/master/Modelos_Base.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-LE42J3GWkf4"
      },
      "source": [
        "# Importar librerías"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XKmmaLVDKHGL",
        "outputId": "7f0c3682-45b3-4c0b-cb0a-36fcf915d53d"
      },
      "source": [
        "import numpy as np #\n",
        "import glob\n",
        "import os\n",
        "import matplotlib.pyplot as plt\n",
        "import tensorflow as tf # Import TensorFlow \n",
        "print(tf.version.VERSION) # Check TensorFlow version\n",
        "import keras\n",
        "from keras.preprocessing.image import ImageDataGenerator, load_img, img_to_array, array_to_img\n",
        "\n",
        "%matplotlib inline \n",
        "#hará que las salidas de gráficos aparezcan y se almacenen en el cuaderno\n",
        "\n",
        "from IPython.display import clear_output #limpiar"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2.7.0\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "lHyJ2PA_aiZR"
      },
      "source": [
        "# Cargar base de datos"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ixfMdmwfUx1N",
        "outputId": "61f7bd70-0e08-46f3-db0f-a2f903993a27"
      },
      "source": [
        "# Carga de base de datos\n",
        "!wget https://www.dropbox.com/s/duigf1xgl38tb8o/DG_V2.zip\n",
        "!unzip DG_V2.zip\n",
        "!rm DG_V2.zip\n",
        "\n",
        "clear_output()\n",
        "print('Base de datos cargada')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Base de datos cargada\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rt1lGqfQasTT"
      },
      "source": [
        "# Preprocesamiento"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "15uoE-BCKHGN",
        "outputId": "9d48c5bf-1259-4452-d80a-ff8e30bbb5b6"
      },
      "source": [
        "#carga de datos en variables\n",
        "IMG_DIM = (150, 150)\n",
        "\n",
        "train_files = glob.glob('training_data/*')\n",
        "train_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in train_files]\n",
        "train_imgs = np.array(train_imgs)\n",
        "train_labels = [fn.split('/')[1].split('.')[0].strip() for fn in train_files]\n",
        "#print(train_labels)\n",
        "\n",
        "validation_files = glob.glob('validation_data/*')\n",
        "validation_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in validation_files]\n",
        "validation_imgs = np.array(validation_imgs)\n",
        "validation_labels = [fn.split('/')[1].split('.')[0].strip() for fn in validation_files]\n",
        "#print(validation_labels)\n",
        "print('Train dataset shape:', train_imgs.shape) \n",
        "print('Validation dataset shape:', validation_imgs.shape)\n",
        "mf_validation = validation_imgs.shape[0]"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Train dataset shape: (7000, 150, 150, 3)\n",
            "Validation dataset shape: (875, 150, 150, 3)\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 185
        },
        "id": "jz53HO4UKHGO",
        "outputId": "82653a32-218c-4be3-d622-ec1dedd68e96"
      },
      "source": [
        "#normalizacion\n",
        "train_imgs_scaled = train_imgs.astype('float32')\n",
        "validation_imgs_scaled  = validation_imgs.astype('float32')\n",
        "train_imgs_scaled /= 255\n",
        "validation_imgs_scaled /= 255\n",
        "\n",
        "print(train_imgs[2556].shape)\n",
        "array_to_img(train_imgs[2556])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "(150, 150, 3)\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<PIL.Image.Image image mode=RGB size=150x150 at 0x7FCE741BECD0>"
            ]
          },
          "metadata": {},
          "execution_count": 4
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "o1t39OUQKHGO",
        "outputId": "3344d798-b83d-494e-ed66-5ee53fc2dbde"
      },
      "source": [
        "#Codificación los LABEL mediante One Hot Encoding\n",
        "batch_size = 30\n",
        "num_classes = 7\n",
        "epochs = 30\n",
        "input_shape = (150, 150, 3)\n",
        "\n",
        "# encode text category labels\n",
        "from sklearn.preprocessing import LabelEncoder\n",
        "from tensorflow.keras.utils import to_categorical\n",
        "\n",
        "le = LabelEncoder()\n",
        "le.fit(train_labels)\n",
        "train_labels_enc = le.transform(train_labels)\n",
        "train_labels_enc = to_categorical(train_labels_enc, num_classes)\n",
        "validation_labels_enc = le.transform(validation_labels)\n",
        "validation_labels_enc = to_categorical(validation_labels_enc, num_classes)\n",
        "\n",
        "print(train_labels[455:465]) \n",
        "print(train_labels_enc[455:465]) #ejemplo de rango"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['elipse', 'cuadrado', 'trapecio', 'rombo', 'elipse', 'triangulo', 'rectangulo', 'rombo', 'cuadrado', 'rombo']\n",
            "[[0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 1. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 0. 1. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 0. 0. 1.]\n",
            " [0. 0. 0. 1. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0.]\n",
            " [0. 1. 0. 0. 0. 0. 0.]\n",
            " [0. 0. 0. 0. 1. 0. 0.]]\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "s5JxDaUKKHGP",
        "outputId": "0e924374-8313-4715-f836-131e4186593c"
      },
      "source": [
        "#data augmentation\n",
        "train_datagen = ImageDataGenerator(rescale=1./255, zoom_range=0.3, rotation_range=40,\n",
        "                                   width_shift_range=0.2, height_shift_range=0.2, shear_range=0.2, \n",
        "                                   horizontal_flip=True, fill_mode='nearest')\n",
        "\n",
        "val_datagen = ImageDataGenerator(rescale=1./255)\n",
        "\n",
        "train_generator = train_datagen.flow(train_imgs, train_labels_enc, batch_size=30, shuffle=True)\n",
        "val_generator = val_datagen.flow(validation_imgs, validation_labels_enc, batch_size=30, shuffle=True)\n",
        "#input_shape = (150, 150, 3)\n",
        "print(train_generator.n)\n",
        "print(val_generator.n)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "7000\n",
            "875\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Yrb14D5eeAe1"
      },
      "source": [
        "# Modelo"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wO9kAdxOoCfU"
      },
      "source": [
        "#modelos sin dropout, solo usar 1. Para usar Dropout elimine comentario en la capa correspodiente\n",
        "from keras.layers import Conv2D, MaxPooling2D, Flatten, Dense, Dropout\n",
        "from keras.models import Sequential\n",
        "from keras import optimizers\n",
        "\n",
        "modelo_1 = [    \n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_2 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_3 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_4 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_5 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_6 = [    \n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(128, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_7 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(128, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(256, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_8 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(128, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(256, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(512, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]\n",
        "\n",
        "modelo_9 = [\n",
        "    Conv2D(16, (3,3), activation='relu',input_shape=(150,150,3)),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(32, (3,3), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(64, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Conv2D(128, (5,5), activation='relu'),\n",
        "    MaxPooling2D(2,2),\n",
        "    Flatten(),\n",
        "    #Dropout(0.3),\n",
        "    Dense(128, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(256, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(512, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(512, activation='relu'),\n",
        "    #Dropout(0.3),\n",
        "    Dense(7, activation='softmax')\n",
        "]"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "RXf6IC_fn8eX",
        "outputId": "b44d2d44-87b4-47c6-f3e7-11f6afd0f444"
      },
      "source": [
        "with tf.device('/device:GPU:0'):\n",
        "  model = Sequential(modelo_6)\n",
        "\n",
        "  model.compile(loss='categorical_crossentropy',\n",
        "                optimizer='adam',\n",
        "                metrics=['accuracy'])\n",
        "\n",
        "  model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            " Layer (type)                Output Shape              Param #   \n",
            "=================================================================\n",
            " conv2d_15 (Conv2D)          (None, 148, 148, 16)      448       \n",
            "                                                                 \n",
            " max_pooling2d_15 (MaxPoolin  (None, 74, 74, 16)       0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " conv2d_16 (Conv2D)          (None, 72, 72, 32)        4640      \n",
            "                                                                 \n",
            " max_pooling2d_16 (MaxPoolin  (None, 36, 36, 32)       0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " conv2d_17 (Conv2D)          (None, 32, 32, 64)        51264     \n",
            "                                                                 \n",
            " max_pooling2d_17 (MaxPoolin  (None, 16, 16, 64)       0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " conv2d_18 (Conv2D)          (None, 12, 12, 128)       204928    \n",
            "                                                                 \n",
            " max_pooling2d_18 (MaxPoolin  (None, 6, 6, 128)        0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " conv2d_19 (Conv2D)          (None, 2, 2, 128)         409728    \n",
            "                                                                 \n",
            " max_pooling2d_19 (MaxPoolin  (None, 1, 1, 128)        0         \n",
            " g2D)                                                            \n",
            "                                                                 \n",
            " flatten_5 (Flatten)         (None, 128)               0         \n",
            "                                                                 \n",
            " dense_5 (Dense)             (None, 128)               16512     \n",
            "                                                                 \n",
            " dense_6 (Dense)             (None, 7)                 903       \n",
            "                                                                 \n",
            "=================================================================\n",
            "Total params: 688,423\n",
            "Trainable params: 688,423\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Q8BpQBQxKHGK"
      },
      "source": [
        "#para calcular el tiempo de entrenamiento\n",
        "import time\n",
        "start = time.time() #partiendo"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Gi9HBHggf6bD"
      },
      "source": [
        "# Entrenamiento"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sga06mSsf_lX"
      },
      "source": [
        "Los steps_per_epoch y epochs para los experimentos variaron de la siguiente forma:\n",
        "*   steps_per_epoch = 30 | epochs = 30\n",
        "*   steps_per_epoch = 30 | epochs = 100\n",
        "*   steps_per_epoch = 86 | epochs = 100\n",
        "*   steps_per_epoch = 233 | epochs = 100\n",
        "\n",
        "Y según la configuracion anterior, se usó Dropout o no\n",
        "\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "CbCL5nZ5KHGQ",
        "outputId": "5d1f244d-a892-443f-bce5-e25e827b6f19"
      },
      "source": [
        "with tf.device('/device:GPU:0'):\n",
        "  steps_per_epoch = train_generator.n//train_generator.batch_size\n",
        "  epochs = 100\n",
        "\n",
        "  STEP_SIZE_VAL=val_generator.n//val_generator.batch_size\n",
        "\n",
        "  history = model.fit(train_generator, steps_per_epoch=steps_per_epoch, epochs=epochs, validation_data=val_generator, validation_steps=STEP_SIZE_VAL)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/100\n",
            "233/233 [==============================] - 39s 133ms/step - loss: 1.8760 - accuracy: 0.1914 - val_loss: 1.7166 - val_accuracy: 0.2425\n",
            "Epoch 2/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 1.7333 - accuracy: 0.2600 - val_loss: 1.6111 - val_accuracy: 0.3115\n",
            "Epoch 3/100\n",
            "233/233 [==============================] - 32s 137ms/step - loss: 1.5976 - accuracy: 0.3334 - val_loss: 1.4082 - val_accuracy: 0.3931\n",
            "Epoch 4/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.3843 - accuracy: 0.4349 - val_loss: 1.2221 - val_accuracy: 0.5356\n",
            "Epoch 5/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.2628 - accuracy: 0.4839 - val_loss: 1.1631 - val_accuracy: 0.5218\n",
            "Epoch 6/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 1.2348 - accuracy: 0.4927 - val_loss: 1.1070 - val_accuracy: 0.5851\n",
            "Epoch 7/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.1763 - accuracy: 0.5195 - val_loss: 1.1124 - val_accuracy: 0.5517\n",
            "Epoch 8/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.1508 - accuracy: 0.5261 - val_loss: 1.1434 - val_accuracy: 0.5276\n",
            "Epoch 9/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.1186 - accuracy: 0.5350 - val_loss: 1.0211 - val_accuracy: 0.6184\n",
            "Epoch 10/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.1039 - accuracy: 0.5429 - val_loss: 1.0392 - val_accuracy: 0.5989\n",
            "Epoch 11/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.0881 - accuracy: 0.5613 - val_loss: 1.1275 - val_accuracy: 0.5563\n",
            "Epoch 12/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.0678 - accuracy: 0.5571 - val_loss: 1.0194 - val_accuracy: 0.5977\n",
            "Epoch 13/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.0550 - accuracy: 0.5661 - val_loss: 1.1563 - val_accuracy: 0.5529\n",
            "Epoch 14/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 1.0349 - accuracy: 0.5808 - val_loss: 0.9583 - val_accuracy: 0.6322\n",
            "Epoch 15/100\n",
            "233/233 [==============================] - 31s 131ms/step - loss: 1.0028 - accuracy: 0.5931 - val_loss: 0.9847 - val_accuracy: 0.5943\n",
            "Epoch 16/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9937 - accuracy: 0.5956 - val_loss: 1.2895 - val_accuracy: 0.5494\n",
            "Epoch 17/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9966 - accuracy: 0.6049 - val_loss: 0.9565 - val_accuracy: 0.6218\n",
            "Epoch 18/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9780 - accuracy: 0.6004 - val_loss: 0.9380 - val_accuracy: 0.6161\n",
            "Epoch 19/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9671 - accuracy: 0.6046 - val_loss: 0.9346 - val_accuracy: 0.6299\n",
            "Epoch 20/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9487 - accuracy: 0.6139 - val_loss: 0.9968 - val_accuracy: 0.6230\n",
            "Epoch 21/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.9472 - accuracy: 0.6185 - val_loss: 0.9601 - val_accuracy: 0.6253\n",
            "Epoch 22/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.9348 - accuracy: 0.6264 - val_loss: 0.9275 - val_accuracy: 0.6425\n",
            "Epoch 23/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.9557 - accuracy: 0.6088 - val_loss: 0.8913 - val_accuracy: 0.6609\n",
            "Epoch 24/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9161 - accuracy: 0.6307 - val_loss: 0.8989 - val_accuracy: 0.6483\n",
            "Epoch 25/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9173 - accuracy: 0.6321 - val_loss: 0.8594 - val_accuracy: 0.6701\n",
            "Epoch 26/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9004 - accuracy: 0.6331 - val_loss: 0.8749 - val_accuracy: 0.6586\n",
            "Epoch 27/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.9045 - accuracy: 0.6356 - val_loss: 0.8273 - val_accuracy: 0.6713\n",
            "Epoch 28/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.8726 - accuracy: 0.6429 - val_loss: 0.8201 - val_accuracy: 0.6678\n",
            "Epoch 29/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.8647 - accuracy: 0.6462 - val_loss: 0.8238 - val_accuracy: 0.6897\n",
            "Epoch 30/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.8629 - accuracy: 0.6512 - val_loss: 0.9234 - val_accuracy: 0.6517\n",
            "Epoch 31/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.8743 - accuracy: 0.6478 - val_loss: 0.8054 - val_accuracy: 0.6851\n",
            "Epoch 32/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.8548 - accuracy: 0.6555 - val_loss: 0.8116 - val_accuracy: 0.6943\n",
            "Epoch 33/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8521 - accuracy: 0.6525 - val_loss: 1.1949 - val_accuracy: 0.6621\n",
            "Epoch 34/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8451 - accuracy: 0.6532 - val_loss: 0.8620 - val_accuracy: 0.6690\n",
            "Epoch 35/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8371 - accuracy: 0.6562 - val_loss: 0.8048 - val_accuracy: 0.6701\n",
            "Epoch 36/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8464 - accuracy: 0.6595 - val_loss: 0.7838 - val_accuracy: 0.6920\n",
            "Epoch 37/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8527 - accuracy: 0.6575 - val_loss: 0.8374 - val_accuracy: 0.6920\n",
            "Epoch 38/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8173 - accuracy: 0.6756 - val_loss: 0.8304 - val_accuracy: 0.6931\n",
            "Epoch 39/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8127 - accuracy: 0.6805 - val_loss: 0.7646 - val_accuracy: 0.7080\n",
            "Epoch 40/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.8188 - accuracy: 0.6722 - val_loss: 0.9906 - val_accuracy: 0.6885\n",
            "Epoch 41/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.8104 - accuracy: 0.6780 - val_loss: 0.7822 - val_accuracy: 0.6989\n",
            "Epoch 42/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7887 - accuracy: 0.6846 - val_loss: 0.7607 - val_accuracy: 0.7138\n",
            "Epoch 43/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7782 - accuracy: 0.6895 - val_loss: 0.7594 - val_accuracy: 0.7299\n",
            "Epoch 44/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.7753 - accuracy: 0.6904 - val_loss: 0.7563 - val_accuracy: 0.7034\n",
            "Epoch 45/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.7754 - accuracy: 0.6884 - val_loss: 0.7472 - val_accuracy: 0.7115\n",
            "Epoch 46/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.7717 - accuracy: 0.6931 - val_loss: 0.7550 - val_accuracy: 0.7276\n",
            "Epoch 47/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.7639 - accuracy: 0.6960 - val_loss: 0.7198 - val_accuracy: 0.7414\n",
            "Epoch 48/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7520 - accuracy: 0.7052 - val_loss: 0.7369 - val_accuracy: 0.7287\n",
            "Epoch 49/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.7468 - accuracy: 0.7042 - val_loss: 0.7533 - val_accuracy: 0.7276\n",
            "Epoch 50/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.7435 - accuracy: 0.7077 - val_loss: 0.7616 - val_accuracy: 0.7253\n",
            "Epoch 51/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7362 - accuracy: 0.7099 - val_loss: 0.6733 - val_accuracy: 0.7540\n",
            "Epoch 52/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.7379 - accuracy: 0.7142 - val_loss: 0.6934 - val_accuracy: 0.7506\n",
            "Epoch 53/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7126 - accuracy: 0.7222 - val_loss: 0.7211 - val_accuracy: 0.7402\n",
            "Epoch 54/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6980 - accuracy: 0.7270 - val_loss: 0.6791 - val_accuracy: 0.7598\n",
            "Epoch 55/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7155 - accuracy: 0.7224 - val_loss: 0.6928 - val_accuracy: 0.7586\n",
            "Epoch 56/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.7060 - accuracy: 0.7319 - val_loss: 0.6634 - val_accuracy: 0.7540\n",
            "Epoch 57/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.6944 - accuracy: 0.7303 - val_loss: 0.6528 - val_accuracy: 0.7736\n",
            "Epoch 58/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6951 - accuracy: 0.7336 - val_loss: 0.6363 - val_accuracy: 0.7586\n",
            "Epoch 59/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6915 - accuracy: 0.7265 - val_loss: 0.6349 - val_accuracy: 0.7759\n",
            "Epoch 60/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6787 - accuracy: 0.7353 - val_loss: 0.6318 - val_accuracy: 0.7575\n",
            "Epoch 61/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6642 - accuracy: 0.7435 - val_loss: 0.7365 - val_accuracy: 0.7333\n",
            "Epoch 62/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6634 - accuracy: 0.7449 - val_loss: 0.6109 - val_accuracy: 0.7793\n",
            "Epoch 63/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.6598 - accuracy: 0.7443 - val_loss: 0.6586 - val_accuracy: 0.7747\n",
            "Epoch 64/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6534 - accuracy: 0.7479 - val_loss: 0.6669 - val_accuracy: 0.7586\n",
            "Epoch 65/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6460 - accuracy: 0.7473 - val_loss: 0.6530 - val_accuracy: 0.7793\n",
            "Epoch 66/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.6431 - accuracy: 0.7561 - val_loss: 0.5905 - val_accuracy: 0.7874\n",
            "Epoch 67/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.6777 - accuracy: 0.7403 - val_loss: 0.6329 - val_accuracy: 0.7724\n",
            "Epoch 68/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6530 - accuracy: 0.7495 - val_loss: 0.5953 - val_accuracy: 0.7908\n",
            "Epoch 69/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.6268 - accuracy: 0.7575 - val_loss: 0.5487 - val_accuracy: 0.8057\n",
            "Epoch 70/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.5969 - accuracy: 0.7687 - val_loss: 0.5629 - val_accuracy: 0.8057\n",
            "Epoch 71/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.6130 - accuracy: 0.7654 - val_loss: 0.5561 - val_accuracy: 0.7839\n",
            "Epoch 72/100\n",
            "233/233 [==============================] - 31s 132ms/step - loss: 0.6047 - accuracy: 0.7624 - val_loss: 0.6231 - val_accuracy: 0.7816\n",
            "Epoch 73/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5934 - accuracy: 0.7745 - val_loss: 0.5647 - val_accuracy: 0.8069\n",
            "Epoch 74/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.6247 - accuracy: 0.7660 - val_loss: 0.5313 - val_accuracy: 0.8023\n",
            "Epoch 75/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.5915 - accuracy: 0.7773 - val_loss: 0.5410 - val_accuracy: 0.8046\n",
            "Epoch 76/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.5850 - accuracy: 0.7714 - val_loss: 0.5231 - val_accuracy: 0.8126\n",
            "Epoch 77/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.5842 - accuracy: 0.7816 - val_loss: 0.5350 - val_accuracy: 0.8069\n",
            "Epoch 78/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.6058 - accuracy: 0.7692 - val_loss: 0.5393 - val_accuracy: 0.7897\n",
            "Epoch 79/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5997 - accuracy: 0.7730 - val_loss: 0.5551 - val_accuracy: 0.8023\n",
            "Epoch 80/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.6003 - accuracy: 0.7737 - val_loss: 0.5437 - val_accuracy: 0.7966\n",
            "Epoch 81/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.5637 - accuracy: 0.7861 - val_loss: 0.5847 - val_accuracy: 0.7874\n",
            "Epoch 82/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.5784 - accuracy: 0.7766 - val_loss: 0.5427 - val_accuracy: 0.8046\n",
            "Epoch 83/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5856 - accuracy: 0.7747 - val_loss: 0.5591 - val_accuracy: 0.8126\n",
            "Epoch 84/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.5646 - accuracy: 0.7841 - val_loss: 0.5428 - val_accuracy: 0.8057\n",
            "Epoch 85/100\n",
            "233/233 [==============================] - 32s 136ms/step - loss: 0.5721 - accuracy: 0.7867 - val_loss: 0.5433 - val_accuracy: 0.8000\n",
            "Epoch 86/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5631 - accuracy: 0.7904 - val_loss: 0.5202 - val_accuracy: 0.8023\n",
            "Epoch 87/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5784 - accuracy: 0.7841 - val_loss: 0.5240 - val_accuracy: 0.8069\n",
            "Epoch 88/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.6031 - accuracy: 0.7755 - val_loss: 0.5171 - val_accuracy: 0.8172\n",
            "Epoch 89/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5690 - accuracy: 0.7839 - val_loss: 0.5107 - val_accuracy: 0.8149\n",
            "Epoch 90/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.5591 - accuracy: 0.7927 - val_loss: 0.4996 - val_accuracy: 0.8161\n",
            "Epoch 91/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5635 - accuracy: 0.7859 - val_loss: 0.5035 - val_accuracy: 0.8023\n",
            "Epoch 92/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.5518 - accuracy: 0.7875 - val_loss: 0.6094 - val_accuracy: 0.7747\n",
            "Epoch 93/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5520 - accuracy: 0.7914 - val_loss: 0.4930 - val_accuracy: 0.8057\n",
            "Epoch 94/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.5443 - accuracy: 0.7966 - val_loss: 0.4770 - val_accuracy: 0.8149\n",
            "Epoch 95/100\n",
            "233/233 [==============================] - 32s 135ms/step - loss: 0.5558 - accuracy: 0.7914 - val_loss: 0.5122 - val_accuracy: 0.8126\n",
            "Epoch 96/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5461 - accuracy: 0.7960 - val_loss: 0.5123 - val_accuracy: 0.8161\n",
            "Epoch 97/100\n",
            "233/233 [==============================] - 31s 135ms/step - loss: 0.5378 - accuracy: 0.7943 - val_loss: 0.4933 - val_accuracy: 0.8138\n",
            "Epoch 98/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5435 - accuracy: 0.7957 - val_loss: 0.4539 - val_accuracy: 0.8322\n",
            "Epoch 99/100\n",
            "233/233 [==============================] - 31s 134ms/step - loss: 0.5212 - accuracy: 0.7971 - val_loss: 0.5046 - val_accuracy: 0.8103\n",
            "Epoch 100/100\n",
            "233/233 [==============================] - 31s 133ms/step - loss: 0.5468 - accuracy: 0.7984 - val_loss: 0.4832 - val_accuracy: 0.8276\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "1UlHBOvzKHGT",
        "outputId": "6d1c33a2-c79d-474a-834e-2a57297af79c"
      },
      "source": [
        "done = time.time()\n",
        "elapsed = done - start #terminando\n",
        "print(\"Tiempo total de ejecución training/validation\", elapsed)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Tiempo total de ejecución training/validation 3135.1372678279877\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "riBexwbbha7B"
      },
      "source": [
        "# Gráfico de entrenamiento"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 308
        },
        "id": "NUYIwi3QKHGT",
        "outputId": "97eb0b87-a552-4257-f147-bac69fb8fd31"
      },
      "source": [
        "f, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 4))\n",
        "t = f.suptitle('Modelo 6 Sin Dropout Data Augmentation 233-100', fontsize=12)\n",
        "f.subplots_adjust(top=0.85, wspace=0.3)\n",
        "\n",
        "epoch_list = list(range(1,101))\n",
        "ax1.plot(epoch_list, history.history['accuracy'], label='Train Accuracy')\n",
        "ax1.plot(epoch_list, history.history['val_accuracy'], label='Validation Accuracy')\n",
        "ax1.set_xticks(np.arange(1, 101, 5))\n",
        "ax1.set_ylabel('Accuracy Value')\n",
        "ax1.set_xlabel('Epoch')\n",
        "ax1.set_title('Accuracy')\n",
        "l1 = ax1.legend(loc=\"best\")\n",
        "\n",
        "ax2.plot(epoch_list, history.history['loss'], label='Train Loss')\n",
        "ax2.plot(epoch_list, history.history['val_loss'], label='Validation Loss')\n",
        "ax2.set_xticks(np.arange(0, 101, 5))\n",
        "ax2.set_ylabel('Loss Value')\n",
        "ax2.set_xlabel('Epoch')\n",
        "ax2.set_title('Loss')\n",
        "l2 = ax2.legend(loc=\"best\")"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Wo6N90qHKHGT"
      },
      "source": [
        "#se guarda modelo luego del entrenamiento\n",
        "model.save('modelo6_sDP-DA-233-100.h5') #"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nhFDDlUahtXL"
      },
      "source": [
        "# Carga del modelo entrenado\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xis9T5hjKHGT",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "43030f6f-f819-44f9-c484-1d86b6571335"
      },
      "source": [
        "from keras.models import load_model\n",
        "\n",
        "# cargar el modelo\n",
        "modelo_6_sDpDA = load_model('modelo6_sDP-DA-233-100.h5')\n",
        "\n",
        "clear_output()\n",
        "print('Modelo cargado')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Modelo cargado\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CT8qPPKvkNjt"
      },
      "source": [
        "# Preprocesamiento de datos de Test"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "lZGQ__ASKHGU",
        "outputId": "69633d14-8ce5-41ce-85da-5c60d193dfb7"
      },
      "source": [
        "IMG_DIM = (150, 150)\n",
        "\n",
        "test_files = glob.glob('test_data/*')\n",
        "test_imgs = [img_to_array(load_img(img, target_size=IMG_DIM)) for img in test_files]\n",
        "test_imgs = np.array(test_imgs)\n",
        "test_labels = [fn.split('/')[1].split('.')[0].strip() for fn in test_files]\n",
        "\n",
        "test_imgs_scaled = test_imgs.astype('float32')\n",
        "test_imgs_scaled /= 255\n",
        "\n",
        "print('Test dataset shape:', test_imgs.shape)\n",
        "print(test_labels[100:105])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Test dataset shape: (875, 150, 150, 3)\n",
            "['cuadrado', 'rombo', 'cuadrado', 'rectangulo', 'circulo']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "3439ZrEeKHGU",
        "outputId": "a5be48b7-9425-475b-e136-47bbe727c524"
      },
      "source": [
        "test_labels_enc = le.transform(test_labels)\n",
        "test_labels_enc = to_categorical(test_labels_enc, num_classes)\n",
        "\n",
        "test_eval = modelo_6_sDpDA.evaluate(test_imgs_scaled, test_labels_enc, verbose=1)\n",
        " \n",
        "print('Test loss:', test_eval[0])\n",
        "print('Test accuracy:', test_eval[1])"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "28/28 [==============================] - 0s 8ms/step - loss: 0.5111 - accuracy: 0.8046\n",
            "Test loss: 0.5110574960708618\n",
            "Test accuracy: 0.8045714497566223\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ozykGKxFKHGU",
        "outputId": "4bc354ba-bcf4-461a-bd17-225de90e65e1"
      },
      "source": [
        "from sklearn import metrics\n",
        "import numpy as np\n",
        "\n",
        "# de numero a etiquetas, según prediccion del modelo\n",
        "predictions = np.argmax(modelo_6_sDpDA.predict(test_imgs_scaled), axis=-1)\n",
        "\n",
        "test_labels_pred = []\n",
        "\n",
        "for i in range(875):\n",
        "    if predictions[i] == 0:\n",
        "        test_labels_pred.append('circulo')\n",
        "    elif predictions[i] == 1:\n",
        "        test_labels_pred.append('cuadrado')\n",
        "    elif predictions[i] == 2:\n",
        "        test_labels_pred.append('elipse')\n",
        "    elif predictions[i] == 3:\n",
        "        test_labels_pred.append('rectangulo')\n",
        "    elif predictions[i] == 4:\n",
        "        test_labels_pred.append('rombo')\n",
        "    elif predictions[i] == 5:\n",
        "        test_labels_pred.append('trapecio')\n",
        "    else:\n",
        "        test_labels_pred.append('triangulo')\n",
        "\n",
        "print(test_labels_pred)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "['rombo', 'triangulo', 'triangulo', 'trapecio', 'trapecio', 'rombo', 'rombo', 'trapecio', 'rectangulo', 'circulo', 'triangulo', 'circulo', 'cuadrado', 'rectangulo', 'cuadrado', 'triangulo', 'cuadrado', 'rombo', 'rectangulo', 'trapecio', 'circulo', 'triangulo', 'triangulo', 'triangulo', 'trapecio', 'triangulo', 'rectangulo', 'rectangulo', 'elipse', 'trapecio', 'trapecio', 'rectangulo', 'cuadrado', 'cuadrado', 'rectangulo', 'cuadrado', 'circulo', 'cuadrado', 'trapecio', 'triangulo', 'triangulo', 'triangulo', 'rombo', 'rombo', 'triangulo', 'elipse', 'cuadrado', 'rectangulo', 'triangulo', 'circulo', 'rombo', 'triangulo', 'rombo', 'circulo', 'rombo', 'rombo', 'rombo', 'cuadrado', 'rectangulo', 'elipse', 'rectangulo', 'triangulo', 'cuadrado', 'circulo', 'trapecio', 'triangulo', 'circulo', 'elipse', 'cuadrado', 'circulo', 'trapecio', 'circulo', 'cuadrado', 'circulo', 'triangulo', 'rombo', 'triangulo', 'cuadrado', 'circulo', 'circulo', 'cuadrado', 'circulo', 'cuadrado', 'elipse', 'rombo', 'elipse', 'rectangulo', 'rombo', 'cuadrado', 'trapecio', 'rombo', 'cuadrado', 'rectangulo', 'trapecio', 'trapecio', 'triangulo', 'cuadrado', 'cuadrado', 'elipse', 'elipse', 'triangulo', 'circulo', 'triangulo', 'rectangulo', 'circulo', 'rectangulo', 'rombo', 'triangulo', 'circulo', 'cuadrado', 'elipse', 'rombo', 'rectangulo', 'trapecio', 'rectangulo', 'rombo', 'triangulo', 'rectangulo', 'rombo', 'trapecio', 'rectangulo', 'rectangulo', 'circulo', 'circulo', 'triangulo', 'elipse', 'rectangulo', 'elipse', 'circulo', 'trapecio', 'circulo', 'rombo', 'rectangulo', 'rectangulo', 'cuadrado', 'triangulo', 'triangulo', 'rombo', 'rectangulo', 'elipse', 'circulo', 'rectangulo', 'rectangulo', 'triangulo', 'triangulo', 'elipse', 'cuadrado', 'triangulo', 'trapecio', 'rombo', 'cuadrado', 'circulo', 'trapecio', 'triangulo', 'triangulo', 'circulo', 'circulo', 'triangulo', 'triangulo', 'circulo', 'elipse', 'rombo', 'elipse', 'rombo', 'trapecio', 'rectangulo', 'circulo', 'rombo', 'circulo', 'cuadrado', 'cuadrado', 'triangulo', 'elipse', 'elipse', 'circulo', 'circulo', 'trapecio', 'rombo', 'circulo', 'triangulo', 'triangulo', 'cuadrado', 'rombo', 'circulo', 'circulo', 'rectangulo', 'rombo', 'triangulo', 'trapecio', 'circulo', 'circulo', 'circulo', 'circulo', 'circulo', 'trapecio', 'triangulo', 'trapecio', 'rectangulo', 'trapecio', 'cuadrado', 'trapecio', 'circulo', 'elipse', 'cuadrado', 'cuadrado', 'rombo', 'elipse', 'triangulo', 'triangulo', 'trapecio', 'rectangulo', 'trapecio', 'rombo', 'rombo', 'elipse', 'trapecio', 'rombo', 'rectangulo', 'rombo', 'triangulo', 'rombo', 'rectangulo', 'elipse', 'cuadrado', 'trapecio', 'elipse', 'elipse', 'rombo', 'trapecio', 'rombo', 'trapecio', 'cuadrado', 'circulo', 'rombo', 'rombo', 'triangulo', 'cuadrado', 'triangulo', 'trapecio', 'circulo', 'triangulo', 'triangulo', 'cuadrado', 'rombo', 'triangulo', 'circulo', 'rombo', 'cuadrado', 'triangulo', 'triangulo', 'rombo', 'triangulo', 'triangulo', 'rombo', 'trapecio', 'cuadrado', 'triangulo', 'triangulo', 'elipse', 'rombo', 'cuadrado', 'rectangulo', 'circulo', 'trapecio', 'rombo', 'triangulo', 'rombo', 'elipse', 'triangulo', 'triangulo', 'rombo', 'elipse', 'triangulo', 'rectangulo', 'rombo', 'elipse', 'triangulo', 'triangulo', 'cuadrado', 'triangulo', 'triangulo', 'circulo', 'triangulo', 'circulo', 'triangulo', 'circulo', 'trapecio', 'circulo', 'triangulo', 'rombo', 'circulo', 'cuadrado', 'circulo', 'triangulo', 'circulo', 'cuadrado', 'cuadrado', 'elipse', 'cuadrado', 'cuadrado', 'trapecio', 'trapecio', 'rectangulo', 'rectangulo', 'cuadrado', 'cuadrado', 'circulo', 'elipse', 'elipse', 'rombo', 'circulo', 'cuadrado', 'circulo', 'trapecio', 'triangulo', 'trapecio', 'cuadrado', 'circulo', 'rectangulo', 'trapecio', 'elipse', 'rombo', 'triangulo', 'triangulo', 'triangulo', 'circulo', 'triangulo', 'rectangulo', 'rectangulo', 'rombo', 'rectangulo', 'circulo', 'rectangulo', 'circulo', 'triangulo', 'cuadrado', 'circulo', 'cuadrado', 'circulo', 'cuadrado', 'cuadrado', 'elipse', 'elipse', 'triangulo', 'trapecio', 'cuadrado', 'triangulo', 'rectangulo', 'cuadrado', 'elipse', 'rectangulo', 'trapecio', 'circulo', 'rombo', 'circulo', 'rombo', 'circulo', 'elipse', 'cuadrado', 'circulo', 'rectangulo', 'elipse', 'circulo', 'triangulo', 'triangulo', 'triangulo', 'circulo', 'rectangulo', 'circulo', 'trapecio', 'circulo', 'cuadrado', 'rectangulo', 'rombo', 'cuadrado', 'circulo', 'circulo', 'circulo', 'trapecio', 'triangulo', 'circulo', 'triangulo', 'circulo', 'trapecio', 'rombo', 'cuadrado', 'triangulo', 'rombo', 'trapecio', 'rombo', 'rectangulo', 'trapecio', 'rombo', 'trapecio', 'triangulo', 'cuadrado', 'triangulo', 'rombo', 'circulo', 'trapecio', 'rectangulo', 'trapecio', 'trapecio', 'elipse', 'trapecio', 'rectangulo', 'elipse', 'rombo', 'circulo', 'triangulo', 'circulo', 'circulo', 'trapecio', 'elipse', 'circulo', 'triangulo', 'circulo', 'cuadrado', 'trapecio', 'trapecio', 'trapecio', 'rombo', 'circulo', 'rombo', 'circulo', 'triangulo', 'trapecio', 'rombo', 'cuadrado', 'cuadrado', 'elipse', 'trapecio', 'trapecio', 'rombo', 'trapecio', 'trapecio', 'circulo', 'rombo', 'rombo', 'rombo', 'elipse', 'triangulo', 'rectangulo', 'triangulo', 'circulo', 'circulo', 'circulo', 'trapecio', 'circulo', 'elipse', 'trapecio', 'rectangulo', 'trapecio', 'circulo', 'rectangulo', 'trapecio', 'cuadrado', 'elipse', 'circulo', 'cuadrado', 'circulo', 'rombo', 'triangulo', 'triangulo', 'rombo', 'triangulo', 'cuadrado', 'triangulo', 'triangulo', 'trapecio', 'circulo', 'rectangulo', 'cuadrado', 'rectangulo', 'rectangulo', 'circulo', 'rombo', 'triangulo', 'triangulo', 'elipse', 'rombo', 'circulo', 'rectangulo', 'rectangulo', 'cuadrado', 'circulo', 'trapecio', 'rombo', 'cuadrado', 'elipse', 'cuadrado', 'cuadrado', 'triangulo', 'rectangulo', 'triangulo', 'triangulo', 'elipse', 'rombo', 'cuadrado', 'circulo', 'circulo', 'rombo', 'rectangulo', 'rombo', 'triangulo', 'elipse', 'triangulo', 'cuadrado', 'circulo', 'triangulo', 'trapecio', 'cuadrado', 'rombo', 'rectangulo', 'circulo', 'rombo', 'circulo', 'rombo', 'cuadrado', 'rombo', 'triangulo', 'cuadrado', 'cuadrado', 'elipse', 'trapecio', 'elipse', 'trapecio', 'circulo', 'cuadrado', 'elipse', 'circulo', 'circulo', 'trapecio', 'circulo', 'circulo', 'elipse', 'rectangulo', 'circulo', 'rombo', 'triangulo', 'trapecio', 'rectangulo', 'rombo', 'rombo', 'triangulo', 'elipse', 'cuadrado', 'circulo', 'trapecio', 'rectangulo', 'elipse', 'elipse', 'rombo', 'circulo', 'cuadrado', 'cuadrado', 'cuadrado', 'rombo', 'trapecio', 'cuadrado', 'elipse', 'triangulo', 'elipse', 'circulo', 'trapecio', 'rectangulo', 'triangulo', 'rectangulo', 'cuadrado', 'rectangulo', 'trapecio', 'circulo', 'rombo', 'trapecio', 'cuadrado', 'circulo', 'elipse', 'trapecio', 'triangulo', 'trapecio', 'elipse', 'rectangulo', 'trapecio', 'rombo', 'elipse', 'elipse', 'circulo', 'cuadrado', 'elipse', 'rectangulo', 'elipse', 'elipse', 'triangulo', 'triangulo', 'triangulo', 'trapecio', 'trapecio', 'rectangulo', 'rombo', 'elipse', 'cuadrado', 'triangulo', 'triangulo', 'circulo', 'circulo', 'circulo', 'circulo', 'rectangulo', 'rombo', 'circulo', 'cuadrado', 'rectangulo', 'rombo', 'elipse', 'triangulo', 'cuadrado', 'circulo', 'cuadrado', 'cuadrado', 'trapecio', 'triangulo', 'rombo', 'circulo', 'rombo', 'cuadrado', 'trapecio', 'rectangulo', 'cuadrado', 'cuadrado', 'triangulo', 'rombo', 'trapecio', 'cuadrado', 'elipse', 'trapecio', 'rombo', 'triangulo', 'elipse', 'triangulo', 'triangulo', 'cuadrado', 'trapecio', 'triangulo', 'elipse', 'triangulo', 'rombo', 'rectangulo', 'triangulo', 'cuadrado', 'cuadrado', 'circulo', 'triangulo', 'cuadrado', 'circulo', 'elipse', 'circulo', 'cuadrado', 'rombo', 'cuadrado', 'trapecio', 'trapecio', 'trapecio', 'cuadrado', 'rectangulo', 'elipse', 'elipse', 'cuadrado', 'rombo', 'rectangulo', 'cuadrado', 'rectangulo', 'rectangulo', 'rectangulo', 'rectangulo', 'cuadrado', 'cuadrado', 'rectangulo', 'triangulo', 'elipse', 'circulo', 'trapecio', 'circulo', 'triangulo', 'rombo', 'rombo', 'rectangulo', 'trapecio', 'cuadrado', 'elipse', 'rectangulo', 'triangulo', 'rombo', 'triangulo', 'circulo', 'triangulo', 'rectangulo', 'rombo', 'cuadrado', 'circulo', 'elipse', 'triangulo', 'rectangulo', 'cuadrado', 'cuadrado', 'rombo', 'rectangulo', 'triangulo', 'circulo', 'circulo', 'trapecio', 'circulo', 'triangulo', 'rectangulo', 'rombo', 'circulo', 'triangulo', 'circulo', 'rectangulo', 'triangulo', 'cuadrado', 'triangulo', 'triangulo', 'triangulo', 'circulo', 'elipse', 'trapecio', 'rombo', 'rombo', 'triangulo', 'circulo', 'trapecio', 'cuadrado', 'elipse', 'trapecio', 'trapecio', 'triangulo', 'triangulo', 'rombo', 'rombo', 'cuadrado', 'cuadrado', 'rombo', 'trapecio', 'elipse', 'rombo', 'circulo', 'triangulo', 'circulo', 'cuadrado', 'circulo', 'circulo', 'rectangulo', 'elipse', 'circulo', 'triangulo', 'circulo', 'triangulo', 'circulo', 'triangulo', 'cuadrado', 'cuadrado', 'triangulo', 'cuadrado', 'cuadrado', 'rombo', 'triangulo', 'circulo', 'rectangulo', 'cuadrado', 'elipse', 'triangulo', 'elipse', 'triangulo', 'cuadrado', 'rectangulo', 'trapecio', 'cuadrado', 'circulo', 'circulo', 'rectangulo', 'cuadrado', 'rombo', 'elipse', 'triangulo', 'elipse', 'trapecio', 'cuadrado', 'rombo', 'triangulo', 'circulo', 'rombo', 'circulo', 'rectangulo', 'trapecio', 'elipse', 'elipse', 'elipse', 'elipse', 'rectangulo', 'cuadrado', 'rombo', 'circulo', 'cuadrado', 'rectangulo', 'rectangulo', 'rombo', 'triangulo', 'cuadrado', 'cuadrado', 'elipse', 'cuadrado', 'triangulo', 'elipse', 'rectangulo', 'triangulo', 'circulo', 'elipse', 'cuadrado', 'elipse', 'circulo', 'rectangulo', 'rombo', 'elipse', 'trapecio', 'trapecio', 'rombo', 'circulo', 'triangulo', 'elipse', 'triangulo', 'trapecio', 'triangulo', 'triangulo', 'circulo', 'triangulo', 'cuadrado', 'triangulo', 'elipse', 'trapecio', 'elipse', 'cuadrado', 'rectangulo', 'rombo', 'circulo', 'rectangulo', 'rombo', 'circulo', 'rombo', 'cuadrado', 'triangulo', 'cuadrado', 'cuadrado', 'triangulo', 'cuadrado', 'cuadrado', 'rectangulo', 'circulo', 'trapecio', 'trapecio', 'circulo', 'cuadrado', 'cuadrado', 'rombo', 'cuadrado', 'rombo', 'elipse', 'circulo', 'rectangulo', 'triangulo', 'rombo', 'trapecio', 'rombo', 'circulo', 'trapecio', 'triangulo']\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Tzi5BjaCKHGV",
        "outputId": "f451c237-d46f-415a-cfaa-4b067a8b4a99"
      },
      "source": [
        "from sklearn import metrics\n",
        "\n",
        "# Matriz de confusión\n",
        "print(metrics.confusion_matrix(test_labels, test_labels_pred))\n",
        "\n",
        "# Precision and recall, F1-score y support\n",
        "print(metrics.classification_report(test_labels, test_labels_pred, digits=3))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "[[116   1   1   0   4   0   3]\n",
            " [  4  98   0   4   6   3  10]\n",
            " [ 18   3  92   0   2   3   7]\n",
            " [  1  20   1  95   2   4   2]\n",
            " [  4   6   0   0 100   1  14]\n",
            " [  6   8   0   1   3  94  13]\n",
            " [  4   1   1   0   6   4 109]]\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "     circulo      0.758     0.928     0.835       125\n",
            "    cuadrado      0.715     0.784     0.748       125\n",
            "      elipse      0.968     0.736     0.836       125\n",
            "  rectangulo      0.950     0.760     0.844       125\n",
            "       rombo      0.813     0.800     0.806       125\n",
            "    trapecio      0.862     0.752     0.803       125\n",
            "   triangulo      0.690     0.872     0.770       125\n",
            "\n",
            "    accuracy                          0.805       875\n",
            "   macro avg      0.822     0.805     0.806       875\n",
            "weighted avg      0.822     0.805     0.806       875\n",
            "\n"
          ]
        }
      ]
    }
  ]
}