Covid_19_Detection_using_X_Ray_Images_Model_Building.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Covid-19 Detection using X-Ray Images - Model Building.ipynb",
      "provenance": [],
      "collapsed_sections": [],
      "toc_visible": true,
      "authorship_tag": "ABX9TyN7XdXKp7D5JfsZhguaLkGH",
      "include_colab_link": true
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/yakupkaplan/Covid-19-Detection-Using-X-Ray-Images/blob/main/Covid_19_Detection_using_X_Ray_Images_Model_Building.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sGK5bla2_EeX"
      },
      "source": [
        "# Covid-19 Detection using X-Ray Images"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EKdeeQoI_U2o"
      },
      "source": [
        "In this notebook we try to detect COVID-19 by using x-ray images.\r\n",
        "\r\n",
        "__Datasets to be used:__ \r\n",
        "1. [Covid Chest X Ray Dataset](https://github.com/ieee8023/covid-chestxray-dataset) -> used for covid-19 positive images\r\n",
        "2. [Pneumonia dataset by Praveen](https://www.kaggle.com/praveengovi/coronahack-chest-xraydataset) -> pneumonia negative images used for covid-19 negative images\r\n",
        "\r\n",
        "-> By using these datasets 500 Covid-19 positive images and 1600 Covid-19 negative images are created.\r\n",
        "\r\n",
        "__References:__\r\n",
        "1. https://www.youtube.com/watch?v=ol0OYJoBC4A&list=PLeR1DH-2kjIr_IZZIC64Mmf5wKEdekRZU&index=11\r\n",
        "2. https://www.kaggle.com/praveengovi/coronahack-chest-xraydataset"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "wZMtx-tEDRfS",
        "outputId": "4d91d430-4577-498a-e9b6-79d710657cda"
      },
      "source": [
        "# Google Colab Identification\r\n",
        "from google.colab import drive\r\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mounted at /content/drive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 36
        },
        "id": "zx9fbivzBuaL",
        "outputId": "57026408-4321-4082-f02c-82b651284922"
      },
      "source": [
        "# Changing working directory\r\n",
        "import os \r\n",
        "root = '/content/drive/MyDrive/Colab Notebooks/Covid-19 Detection Using X-Ray Images'\r\n",
        "os.chdir(root)\r\n",
        "os.getcwd()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'/content/drive/MyDrive/Colab Notebooks/Covid-19 Detection Using X-Ray Images'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 19
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "l0mEPWRgTSEb"
      },
      "source": [
        "# Import dependencies\r\n",
        "import numpy as np\r\n",
        "import pandas as pd\r\n",
        "import matplotlib.pyplot as plt\r\n",
        "import seaborn as sns\r\n",
        "\r\n",
        "from sklearn.model_selection import train_test_split\r\n",
        "\r\n",
        "from keras.models import Sequential,Model\r\n",
        "from keras.layers import Dense, Dropout, Flatten\r\n",
        "from keras.layers import Conv2D,Activation,MaxPooling2D\r\n",
        "from keras.utils import normalize\r\n",
        "from keras.layers import Concatenate\r\n",
        "from keras import Input\r\n",
        "from keras.callbacks import ModelCheckpoint"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ynjRRek0TSHL"
      },
      "source": [
        "# Import data and target that we saved earlier\r\n",
        "data = np.load('data.npy')\r\n",
        "target = np.load('target.npy')"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vtM0TeFaTSKF"
      },
      "source": [
        "# Define input features\r\n",
        "input_shape = data.shape[1:] #50,50,1\r\n",
        "inputs = Input(shape = input_shape)\r\n",
        "\r\n",
        "# Create/Design the model architecture\r\n",
        "convs = []\r\n",
        "\r\n",
        "parrallel_kernels = [3,5,7] # We will use parallel kernels\r\n",
        "\r\n",
        "for k in range(len(parrallel_kernels)):\r\n",
        "\r\n",
        "    conv = Conv2D(128, parrallel_kernels[k], padding = 'same', activation = 'relu', input_shape = input_shape, strides = 1)(inputs)\r\n",
        "    convs.append(conv)\r\n",
        "\r\n",
        "outputs = Concatenate()(convs)\r\n",
        "conv_model = Model(inputs=inputs, outputs=outputs)\r\n",
        "\r\n",
        "model = Sequential()\r\n",
        "model.add(conv_model)\r\n",
        "\r\n",
        "model.add(Conv2D(64,(3,3)))\r\n",
        "model.add(Activation('relu'))\r\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\r\n",
        "\r\n",
        "model.add(Conv2D(32,(3,3)))\r\n",
        "model.add(Activation('relu'))\r\n",
        "model.add(MaxPooling2D(pool_size=(2,2)))\r\n",
        "\r\n",
        "model.add(Flatten())\r\n",
        "model.add(Dropout(0.5))\r\n",
        "\r\n",
        "model.add(Dense(128,activation='relu'))\r\n",
        "model.add(Dropout(0.5))\r\n",
        "\r\n",
        "model.add(Dense(64,activation='relu'))\r\n",
        "model.add(Dropout(0.5))\r\n",
        "\r\n",
        "model.add(Dense(2,input_dim=128,activation='softmax'))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "toVpckjbVltE"
      },
      "source": [
        "# Compile the model\r\n",
        "model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "65kJdVwdVc_C",
        "outputId": "e8bf3523-0248-40e9-b40c-be8716ce0930"
      },
      "source": [
        "# Let's take a look at the model architecture\r\n",
        "model.summary()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential_1\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "model_1 (Functional)         (None, 100, 100, 384)     11008     \n",
            "_________________________________________________________________\n",
            "conv2d_8 (Conv2D)            (None, 98, 98, 64)        221248    \n",
            "_________________________________________________________________\n",
            "activation_2 (Activation)    (None, 98, 98, 64)        0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 49, 49, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_9 (Conv2D)            (None, 47, 47, 32)        18464     \n",
            "_________________________________________________________________\n",
            "activation_3 (Activation)    (None, 47, 47, 32)        0         \n",
            "_________________________________________________________________\n",
            "max_pooling2d_3 (MaxPooling2 (None, 23, 23, 32)        0         \n",
            "_________________________________________________________________\n",
            "flatten_1 (Flatten)          (None, 16928)             0         \n",
            "_________________________________________________________________\n",
            "dropout_3 (Dropout)          (None, 16928)             0         \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 128)               2166912   \n",
            "_________________________________________________________________\n",
            "dropout_4 (Dropout)          (None, 128)               0         \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 64)                8256      \n",
            "_________________________________________________________________\n",
            "dropout_5 (Dropout)          (None, 64)                0         \n",
            "_________________________________________________________________\n",
            "dense_5 (Dense)              (None, 2)                 130       \n",
            "=================================================================\n",
            "Total params: 2,426,018\n",
            "Trainable params: 2,426,018\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "YjCi66fxVc7d"
      },
      "source": [
        "# Split the dataset into train and test sets -> 10% data will be used for testing\r\n",
        "train_data, test_data, train_target, test_target = train_test_split(data, target, test_size = 0.1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "xtcKyTrfYgEV",
        "outputId": "6b3fb6f1-8d18-43d5-8e8b-b3087cd802b3"
      },
      "source": [
        "# Create a model checkpoint - The models will be saved that surpase the model before\r\n",
        "checkpoint = ModelCheckpoint('model-{epoch:03d}.model', monitor='val_loss', verbose=0, save_best_only=True, mode='auto')\r\n",
        "\r\n",
        "# Train the model - 10% as the validation data\r\n",
        "history = model.fit(train_data, train_target, epochs=20, callbacks=[checkpoint], validation_split=0.1)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/20\n",
            "50/50 [==============================] - 7s 131ms/step - loss: 0.5588 - accuracy: 0.7065 - val_loss: 0.1377 - val_accuracy: 0.9548\n",
            "INFO:tensorflow:Assets written to: model-001.model/assets\n",
            "Epoch 2/20\n",
            "50/50 [==============================] - 6s 128ms/step - loss: 0.1696 - accuracy: 0.9494 - val_loss: 0.0442 - val_accuracy: 0.9944\n",
            "INFO:tensorflow:Assets written to: model-002.model/assets\n",
            "Epoch 3/20\n",
            "50/50 [==============================] - 6s 128ms/step - loss: 0.1089 - accuracy: 0.9687 - val_loss: 0.0535 - val_accuracy: 0.9887\n",
            "Epoch 4/20\n",
            "50/50 [==============================] - 6s 129ms/step - loss: 0.0793 - accuracy: 0.9734 - val_loss: 0.0238 - val_accuracy: 0.9887\n",
            "INFO:tensorflow:Assets written to: model-004.model/assets\n",
            "Epoch 5/20\n",
            "50/50 [==============================] - 6s 130ms/step - loss: 0.0661 - accuracy: 0.9741 - val_loss: 0.0434 - val_accuracy: 0.9831\n",
            "Epoch 6/20\n",
            "50/50 [==============================] - 6s 128ms/step - loss: 0.1130 - accuracy: 0.9585 - val_loss: 0.0188 - val_accuracy: 0.9944\n",
            "INFO:tensorflow:Assets written to: model-006.model/assets\n",
            "Epoch 7/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0481 - accuracy: 0.9851 - val_loss: 0.0570 - val_accuracy: 0.9887\n",
            "Epoch 8/20\n",
            "50/50 [==============================] - 6s 128ms/step - loss: 0.0382 - accuracy: 0.9889 - val_loss: 0.0239 - val_accuracy: 0.9831\n",
            "Epoch 9/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0428 - accuracy: 0.9892 - val_loss: 0.0973 - val_accuracy: 0.9774\n",
            "Epoch 10/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0838 - accuracy: 0.9733 - val_loss: 0.0437 - val_accuracy: 0.9887\n",
            "Epoch 11/20\n",
            "50/50 [==============================] - 6s 126ms/step - loss: 0.0377 - accuracy: 0.9875 - val_loss: 0.0646 - val_accuracy: 0.9944\n",
            "Epoch 12/20\n",
            "50/50 [==============================] - 6s 126ms/step - loss: 0.0549 - accuracy: 0.9849 - val_loss: 0.0747 - val_accuracy: 0.9887\n",
            "Epoch 13/20\n",
            "50/50 [==============================] - 6s 126ms/step - loss: 0.0511 - accuracy: 0.9863 - val_loss: 0.0482 - val_accuracy: 0.9944\n",
            "Epoch 14/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0187 - accuracy: 0.9941 - val_loss: 0.0506 - val_accuracy: 0.9944\n",
            "Epoch 15/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0222 - accuracy: 0.9941 - val_loss: 0.0511 - val_accuracy: 0.9944\n",
            "Epoch 16/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0288 - accuracy: 0.9929 - val_loss: 0.1309 - val_accuracy: 0.9831\n",
            "Epoch 17/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0200 - accuracy: 0.9924 - val_loss: 0.0363 - val_accuracy: 0.9944\n",
            "Epoch 18/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0125 - accuracy: 0.9949 - val_loss: 0.0822 - val_accuracy: 0.9944\n",
            "Epoch 19/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0131 - accuracy: 0.9944 - val_loss: 0.0945 - val_accuracy: 0.9887\n",
            "Epoch 20/20\n",
            "50/50 [==============================] - 6s 127ms/step - loss: 0.0159 - accuracy: 0.9971 - val_loss: 0.1613 - val_accuracy: 0.9831\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "hRFleO9LaOJr",
        "outputId": "1d755f6e-737f-4f5e-f9e9-155b8936b536"
      },
      "source": [
        "# Let's interpret the model performance by comparing the train and test loss\r\n",
        "plt.plot(history.history['loss'],'r',label='training loss')\r\n",
        "plt.plot(history.history['val_loss'],label='validation loss')\r\n",
        "plt.xlabel('# epochs')\r\n",
        "plt.ylabel('loss')\r\n",
        "plt.legend()\r\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 279
        },
        "id": "c5SgNwCGbIIa",
        "outputId": "6d68ac5d-6a57-410d-8b41-4bba0bad1aaa"
      },
      "source": [
        "# Let's interpret the model performance by comparing the train and test accuracy\r\n",
        "plt.plot(history.history['accuracy'],'r',label='training accuracy')\r\n",
        "plt.plot(history.history['val_accuracy'],label='validation accuracy')\r\n",
        "plt.xlabel('# epochs')\r\n",
        "plt.ylabel('loss')\r\n",
        "plt.legend()\r\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "09uz7c52bAlM"
      },
      "source": [
        "- For both train and test data losses are decreasing and accuracies are increasing.\r\n",
        "- So it is a good sign for the model.\r\n",
        "- There is no sign for Overfitting.\r\n",
        "- The model can be trained for more epochs for better accuracy.\r\n",
        "- By plot for losss and accuracy, we have similar indications. It is a good sign."
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "XiUY9bW1cKIJ",
        "outputId": "164d8e5e-b0d7-4d1d-c2c6-639e4deeffe3"
      },
      "source": [
        "# Evaluate the model\r\n",
        "print(model.evaluate(test_data,test_target))"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "7/7 [==============================] - 0s 45ms/step - loss: 0.0932 - accuracy: 0.9694\n",
            "[0.09315331280231476, 0.9693877696990967]\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "28V2_ODMD3tY"
      },
      "source": [
        "# Create requirements.txt file to save dependencies versions\r\n",
        "# !pip install -r requirements.txt\r\n",
        "# !pip3 freeze > requirements.txt"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "iJ2I9nvye-qp"
      },
      "source": [
        ""
      ],
      "execution_count": null,
      "outputs": []
    }
  ]
}