Natural Images Recogniser.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cXD8sjjQLj-k"
      },
      "source": [
        "# 🤖 Natural Images Recogniser\n",
        "\n",
        "A neural network will be developed that classifies several types of images. To do so, we are going to use a dataset contained in our Google Drive account obtained through [Kaggle](https://www.kaggle.com/datasets/briansurezsantiago/natural-images). This dataset is made up of 8 different classes listed below, where each of these classes contains 560 images that will be used during the training process. Similarly, these same classes are available to be used in the validation phase, each of these classes containing 140 different images.\n",
        "\n",
        "*   Airplane\n",
        "*   Car\n",
        "*   Cat\n",
        "*   Dog\n",
        "*   Flower\n",
        "*   Fruit\n",
        "*   Motorbike\n",
        "*   Person"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "yZcKFL_TMkzR",
        "outputId": "62624276-822d-483f-c709-5ed042020738",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ]
        }
      ],
      "source": [
        "# Mounts the Google Drive directory in which the dataset is hosted\n",
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9GJsa2DPRE5C"
      },
      "source": [
        "Am I really using a GPU? Check in **Edit / Notebook Settings** or **Runtime / Change runtime type**."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 37
        },
        "id": "eKjrirLcRIkU",
        "outputId": "63de815f-5d42-4c6a-f7e1-d0c477fc67a0",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'/device:GPU:0'"
            ]
          },
          "execution_count": 2,
          "metadata": {},
          "output_type": "execute_result"
        }
      ],
      "source": [
        "# Check that we are making use of a GPU for training the neural network\n",
        "import tensorflow as tf\n",
        "tf.test.gpu_device_name()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Da8ekOhbrgGZ"
      },
      "source": [
        "Let's see an example image."
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 303
        },
        "id": "x3JBgGOsKu44",
        "outputId": "75970c95-0a6d-40b5-a64d-a4465059c4b3",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Training  Validation\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "<matplotlib.image.AxesImage at 0x7fd640a56e10>"
            ]
          },
          "execution_count": 3,
          "metadata": {},
          "output_type": "execute_result"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# List the directories available in the dataset to know the organisation of the data\n",
        "!ls \"/content/drive/My Drive/Colab Notebooks/datasets/NaturalDataset/\"\n",
        "\n",
        "# An example image is displayed, which will be used during the training phase of the network to determine the size of the image\n",
        "from matplotlib.pyplot import imshow\n",
        "from PIL import Image\n",
        "import numpy as np\n",
        "\n",
        "%matplotlib inline\n",
        "pil_im = Image.open('/content/drive/My Drive/Colab Notebooks/datasets/NaturalDataset/Training/cat/cat_0250.jpg', 'r')\n",
        "imshow(np.asarray(pil_im))"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "UCaBgMS8OdiV"
      },
      "source": [
        "## 1️⃣ First Model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "25NfZptPxjgD",
        "outputId": "403c2d08-18c7-4265-99ff-5231a879acb0",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 4480 files belonging to 8 classes.\n",
            "Using 3584 files for training.\n",
            "Found 1120 files belonging to 8 classes.\n",
            "Using 224 files for validation.\n"
          ]
        }
      ],
      "source": [
        "# Data source\n",
        "\n",
        "image_size = (100, 100)  # Images of 100x100 dimension\n",
        "batch_size = 32          # Number of images that are passed to the network per epoch\n",
        "\n",
        "train_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
        "    \"/content/drive/My Drive/Colab Notebooks/datasets/NaturalDataset/Training\",\n",
        "    validation_split=0.2,     # 20% of images for validation\n",
        "    subset=\"training\",\n",
        "    seed=1337,                # In an iteration, we do not feed the network with all the images but with small portions given by the batch_size\n",
        "    image_size=image_size,    # Size of our dataset images\n",
        "    batch_size=batch_size, \n",
        "    label_mode='categorical'  # The network classifies more than two classes (categorical), if the network would classify two classes (binary)\n",
        ")\n",
        "\n",
        "val_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
        "    \"/content/drive/My Drive/Colab Notebooks/datasets/NaturalDataset/Validation\",\n",
        "    validation_split=0.2,     # 80% of images for training\n",
        "    subset=\"validation\",\n",
        "    seed=1337,                # This is a specific way of randomising the data used during the training phase of the network\n",
        "    image_size=image_size,    # Recommended images sizes (100, 200, 512)\n",
        "    batch_size=batch_size,\n",
        "    label_mode='categorical' \n",
        ")\n",
        "\n",
        "train_ds = train_ds.prefetch(buffer_size=32)\n",
        "val_ds = val_ds.prefetch(buffer_size=32)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "e5tVrY9gxyfJ",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [],
      "source": [
        "# Convolutional Neural Network Model\n",
        "\n",
        "from tensorflow import keras        \n",
        "from tensorflow.keras.layers import Conv2D, MaxPooling2D, Dropout, Dense, Rescaling, Flatten\n",
        "from tensorflow.keras.callbacks import EarlyStopping\n",
        "import tensorflow as tf\n",
        "\n",
        "model = keras.Sequential()            # Model initialisation mode. Sequential mode adds one layer after another\n",
        "model.add(Rescaling(scale=(1./127.5), # Layer for the normalisation of the dataset image set\n",
        "                    offset=-1,\n",
        "                    input_shape=(100, 100, 3)))               # Size of the input images\n",
        "\n",
        "model.add(Conv2D(32, kernel_size=(3, 3), activation='relu'))  # First convolutional layer. Fully connected layer\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))                     # Halves the image (in our case, the image would be 49x49)\n",
        "\n",
        "model.add(Conv2D(64, kernel_size=(3, 3), activation='relu'))  # kernel_size = image filter\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "model.add(Dropout(0.25))  # Deactivates 25% of the neurons, forcing the network to rely on the remaining neurons available for training\n",
        "\n",
        "model.add(Conv2D(128, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "model.add(Dropout(0.25))  # Causes it to stop using the same neurons during the learning phase (switches them off in the previous layer)\n",
        "\n",
        "model.add(Conv2D(256, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "model.add(Dropout(0.5))\n",
        "\n",
        "model.add(Conv2D(512, kernel_size=(3, 3), activation='relu'))\n",
        "model.add(MaxPooling2D(pool_size=(2, 2)))\n",
        "model.add(Dropout(0.7))\n",
        "\n",
        "model.add(Flatten())                      # Converts the matrix into a vector\n",
        "model.add(Dense(128, activation='relu'))  # Simplification of the vector obtained above\n",
        "model.add(Dropout(0.5))\n",
        "\n",
        "model.add(Dense(8, activation='softmax')) # The first parameter specifies the number of classes the dataset has\n",
        "\n",
        "# Specify the model compilation mode\n",
        "model.compile(loss=tf.keras.losses.categorical_crossentropy,  # Loss function\n",
        "              optimizer=tf.keras.optimizers.Adam(1e-3),       # Adam optimiser (based on SGD = Stochastic Gradient Descent)\n",
        "              metrics=['accuracy'])                           # Metric that measures the learning accuracy of the network in each epoch"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "-AiE5BwazFFK",
        "outputId": "aa303034-9054-4789-a7e2-ad7462f9dce3",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/200\n",
            "112/112 [==============================] - 11s 65ms/step - loss: 1.9325 - accuracy: 0.2065 - val_loss: 1.7984 - val_accuracy: 0.4330\n",
            "Epoch 2/200\n",
            "112/112 [==============================] - 10s 81ms/step - loss: 1.4857 - accuracy: 0.4428 - val_loss: 1.3820 - val_accuracy: 0.5045\n",
            "Epoch 3/200\n",
            "112/112 [==============================] - 9s 80ms/step - loss: 1.1363 - accuracy: 0.5896 - val_loss: 1.0133 - val_accuracy: 0.6339\n",
            "Epoch 4/200\n",
            "112/112 [==============================] - 9s 79ms/step - loss: 0.9718 - accuracy: 0.6526 - val_loss: 0.9870 - val_accuracy: 0.6518\n",
            "Epoch 5/200\n",
            "112/112 [==============================] - 9s 80ms/step - loss: 0.8339 - accuracy: 0.6953 - val_loss: 0.8054 - val_accuracy: 0.6696\n",
            "Epoch 6/200\n",
            "112/112 [==============================] - 9s 80ms/step - loss: 0.7076 - accuracy: 0.7344 - val_loss: 0.6261 - val_accuracy: 0.7634\n",
            "Epoch 7/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.7012 - accuracy: 0.7517 - val_loss: 0.6872 - val_accuracy: 0.7545\n",
            "Epoch 8/200\n",
            "112/112 [==============================] - 10s 86ms/step - loss: 0.5466 - accuracy: 0.7944 - val_loss: 0.5420 - val_accuracy: 0.8036\n",
            "Epoch 9/200\n",
            "112/112 [==============================] - 10s 82ms/step - loss: 0.5463 - accuracy: 0.8050 - val_loss: 0.5053 - val_accuracy: 0.7857\n",
            "Epoch 10/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.4871 - accuracy: 0.8175 - val_loss: 0.5416 - val_accuracy: 0.7946\n",
            "Epoch 11/200\n",
            "112/112 [==============================] - 10s 82ms/step - loss: 0.4282 - accuracy: 0.8396 - val_loss: 0.5067 - val_accuracy: 0.8214\n",
            "Epoch 12/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.4438 - accuracy: 0.8362 - val_loss: 0.4661 - val_accuracy: 0.8259\n",
            "Epoch 13/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.4201 - accuracy: 0.8465 - val_loss: 0.4019 - val_accuracy: 0.8527\n",
            "Epoch 14/200\n",
            "112/112 [==============================] - 10s 81ms/step - loss: 0.4044 - accuracy: 0.8418 - val_loss: 0.3908 - val_accuracy: 0.8795\n",
            "Epoch 15/200\n",
            "112/112 [==============================] - 10s 81ms/step - loss: 0.3637 - accuracy: 0.8669 - val_loss: 0.4073 - val_accuracy: 0.8616\n",
            "Epoch 16/200\n",
            "112/112 [==============================] - 10s 86ms/step - loss: 0.3587 - accuracy: 0.8650 - val_loss: 0.4339 - val_accuracy: 0.8750\n",
            "Epoch 17/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.3127 - accuracy: 0.8806 - val_loss: 0.4777 - val_accuracy: 0.8527\n",
            "Epoch 18/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.3429 - accuracy: 0.8758 - val_loss: 0.3724 - val_accuracy: 0.8839\n",
            "Epoch 19/200\n",
            "112/112 [==============================] - 10s 81ms/step - loss: 0.2990 - accuracy: 0.8887 - val_loss: 0.4469 - val_accuracy: 0.8616\n",
            "Epoch 20/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2906 - accuracy: 0.8923 - val_loss: 0.2766 - val_accuracy: 0.9062\n",
            "Epoch 21/200\n",
            "112/112 [==============================] - 10s 83ms/step - loss: 0.3092 - accuracy: 0.8895 - val_loss: 0.4210 - val_accuracy: 0.8571\n",
            "Epoch 22/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.3178 - accuracy: 0.8811 - val_loss: 0.5338 - val_accuracy: 0.8438\n",
            "Epoch 23/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.2808 - accuracy: 0.8954 - val_loss: 0.2956 - val_accuracy: 0.9062\n",
            "Epoch 24/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2853 - accuracy: 0.8943 - val_loss: 0.3308 - val_accuracy: 0.9107\n",
            "Epoch 25/200\n",
            "112/112 [==============================] - 10s 83ms/step - loss: 0.2349 - accuracy: 0.9138 - val_loss: 0.3600 - val_accuracy: 0.8884\n",
            "Epoch 26/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2592 - accuracy: 0.9004 - val_loss: 0.3702 - val_accuracy: 0.8750\n",
            "Epoch 27/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2456 - accuracy: 0.9152 - val_loss: 0.3046 - val_accuracy: 0.9107\n",
            "Epoch 28/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2153 - accuracy: 0.9235 - val_loss: 0.4136 - val_accuracy: 0.8705\n",
            "Epoch 29/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2393 - accuracy: 0.9210 - val_loss: 0.3819 - val_accuracy: 0.8884\n",
            "Epoch 30/200\n",
            "112/112 [==============================] - 10s 86ms/step - loss: 0.2283 - accuracy: 0.9205 - val_loss: 0.3103 - val_accuracy: 0.9152\n",
            "Epoch 31/200\n",
            "112/112 [==============================] - 10s 82ms/step - loss: 0.2176 - accuracy: 0.9199 - val_loss: 0.3977 - val_accuracy: 0.8839\n",
            "Epoch 32/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2193 - accuracy: 0.9227 - val_loss: 0.3593 - val_accuracy: 0.8839\n",
            "Epoch 33/200\n",
            "112/112 [==============================] - 10s 83ms/step - loss: 0.1620 - accuracy: 0.9397 - val_loss: 0.3422 - val_accuracy: 0.8973\n",
            "Epoch 34/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.1963 - accuracy: 0.9305 - val_loss: 0.3724 - val_accuracy: 0.8973\n",
            "Epoch 35/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.1985 - accuracy: 0.9330 - val_loss: 0.3224 - val_accuracy: 0.9018\n",
            "Epoch 36/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.2349 - accuracy: 0.9252 - val_loss: 0.3912 - val_accuracy: 0.8973\n",
            "Epoch 37/200\n",
            "112/112 [==============================] - 10s 83ms/step - loss: 0.1822 - accuracy: 0.9333 - val_loss: 0.3061 - val_accuracy: 0.8929\n",
            "Epoch 38/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.1713 - accuracy: 0.9403 - val_loss: 0.2625 - val_accuracy: 0.9286\n",
            "Epoch 39/200\n",
            "112/112 [==============================] - 10s 83ms/step - loss: 0.1698 - accuracy: 0.9464 - val_loss: 0.4673 - val_accuracy: 0.8661\n",
            "Epoch 40/200\n",
            "112/112 [==============================] - 10s 87ms/step - loss: 0.1695 - accuracy: 0.9442 - val_loss: 0.2918 - val_accuracy: 0.9018\n",
            "Epoch 41/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.1246 - accuracy: 0.9548 - val_loss: 0.2486 - val_accuracy: 0.9196\n",
            "Epoch 42/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.1284 - accuracy: 0.9534 - val_loss: 0.4760 - val_accuracy: 0.8973\n",
            "Epoch 43/200\n",
            "112/112 [==============================] - 10s 86ms/step - loss: 0.1652 - accuracy: 0.9442 - val_loss: 0.2735 - val_accuracy: 0.9241\n",
            "Epoch 44/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.1872 - accuracy: 0.9414 - val_loss: 0.3191 - val_accuracy: 0.9152\n",
            "Epoch 45/200\n",
            "112/112 [==============================] - 10s 84ms/step - loss: 0.1752 - accuracy: 0.9420 - val_loss: 0.3670 - val_accuracy: 0.9018\n",
            "Epoch 46/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.1448 - accuracy: 0.9503 - val_loss: 0.3007 - val_accuracy: 0.9107\n",
            "Epoch 47/200\n",
            "112/112 [==============================] - 10s 85ms/step - loss: 0.1008 - accuracy: 0.9671 - val_loss: 0.3497 - val_accuracy: 0.8929\n",
            "Epoch 48/200\n",
            "112/112 [==============================] - ETA: 0s - loss: 0.1468 - accuracy: 0.9526Restoring model weights from the end of the best epoch: 38.\n",
            "112/112 [==============================] - 10s 86ms/step - loss: 0.1468 - accuracy: 0.9526 - val_loss: 0.3878 - val_accuracy: 0.9152\n",
            "Epoch 00048: early stopping\n"
          ]
        }
      ],
      "source": [
        "# Training Phase\n",
        "\n",
        "epochs = 200\n",
        "\n",
        "# Once the highest value is reached in the network and it stabilises or starts\n",
        "# to have losses, the training process is stopped. The parameter patience=10 makes\n",
        "# it stop after 10 waiting epochs when the highest value has been recorded and the\n",
        "# model has not improved, so that finally we will have the best value and therefore\n",
        "# the best weights.\n",
        "es = EarlyStopping(monitor='val_accuracy', mode='max', verbose=1, patience=10, restore_best_weights=True)\n",
        "\n",
        "# The model fits the given training and validation data, the number of epochs and each of the previously implemented layers\n",
        "h = model.fit(\n",
        "        train_ds,\n",
        "        epochs=epochs, \n",
        "        validation_data=val_ds,\n",
        "        callbacks = [es]\n",
        ")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 295
        },
        "id": "2PPKm5euoRAs",
        "outputId": "75322679-5efa-4867-84c0-029df58ea4a0",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Graphical representation of the results obtained during the training phase\n",
        "\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "plt.plot(h.history['accuracy'])\n",
        "plt.plot(h.history['val_accuracy'])\n",
        "plt.plot(h.history['loss'])\n",
        "plt.title('Model accuracy')\n",
        "plt.ylabel('accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['training', 'validation','loss'], loc='upper right')\n",
        "plt.show()"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "TVOKAhbsm_pw"
      },
      "source": [
        "## 📉 Evaluation of the results"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 520
        },
        "id": "IkyaTtnAKfqA",
        "outputId": "bf82d8a9-7714-446f-ac4b-920bee3a5f1c",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0     0.8333    0.9524    0.8889        21\n",
            "           1     0.9189    1.0000    0.9577        34\n",
            "           2     0.9259    0.7576    0.8333        33\n",
            "           3     0.7812    0.8333    0.8065        30\n",
            "           4     1.0000    1.0000    1.0000        31\n",
            "           5     1.0000    1.0000    1.0000        25\n",
            "           6     1.0000    0.9091    0.9524        22\n",
            "           7     1.0000    1.0000    1.0000        28\n",
            "\n",
            "    accuracy                         0.9286       224\n",
            "   macro avg     0.9324    0.9315    0.9299       224\n",
            "weighted avg     0.9319    0.9286    0.9280       224\n",
            "\n"
          ]
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 2 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          },
          "output_type": "display_data"
        }
      ],
      "source": [
        "# Confusion matrix. Allow visualisation of neural network performance\n",
        "\n",
        "from sklearn.metrics import classification_report, confusion_matrix\n",
        "import seaborn as sns\n",
        "import numpy as np\n",
        "\n",
        "results = np.concatenate([(y, model.predict(x=x)) for x, y in val_ds], axis=1)\n",
        "\n",
        "predictions = np.argmax(results[0], axis=1)\n",
        "labels = np.argmax(results[1], axis=1)\n",
        "\n",
        "cf_matrix = confusion_matrix(labels, predictions)\n",
        "\n",
        "sns.heatmap(cf_matrix, annot=True, fmt=\"d\", cmap=\"Blues\")\n",
        "\n",
        "print(classification_report(labels, predictions, digits = 4))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "djAlHGLz-Cjj",
        "outputId": "f5cda494-9d6f-4b95-843c-862d68696bdb",
        "vscode": {
          "languageId": "python"
        }
      },
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "3\n"
          ]
        }
      ],
      "source": [
        "# Validation Phase\n",
        "\n",
        "img = keras.preprocessing.image.load_img(\n",
        "    '/content/drive/My Drive/Colab Notebooks/datasets/NaturalDataset/Validation/dog/dog_0002.jpg', target_size=image_size\n",
        ")\n",
        "\n",
        "img_array = keras.preprocessing.image.img_to_array(img)\n",
        "img_array = tf.expand_dims(img_array, 0)  # Create batch axis\n",
        "\n",
        "predictions = model.predict(img_array)    # Once trained, the network tries to predict to which class the given validation image belongs\n",
        "print(np.argmax(predictions[0])) \n",
        "\n",
        "# The result obtained below represents the classification performed by\n",
        "# the neural network in this validation process after the training phase.\n",
        "# From an image belonging to the set of validation images and not used in\n",
        "# the training process, the network classifies and returns the number of\n",
        "# the class to which this image belongs."
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "name": "CNN. Natural Images Recogniser.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}