CNN_Classifier.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "CNN_Classifier.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    },
    "accelerator": "GPU"
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jpiy9Gvl-6Cx"
      },
      "source": [
        "## CNN Classifier (Mask, No Mask, Incorrectly wearing Mask)"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "8qbvfJ_S_MR7"
      },
      "source": [
        "### Import Packages"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qCfhYWYDT2z6"
      },
      "source": [
        "# Import Libraries \n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "import tensorflow\n",
        "import sklearn\n",
        "from sklearn.model_selection import train_test_split\n",
        "import cv2\n",
        "\n",
        "import seaborn as sns\n",
        "import pickle\n",
        "import os"
      ],
      "execution_count": 1,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rg1oz2jV93sK"
      },
      "source": [
        "### Load in the dataset that was preprocessed from local directory"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "GdZWFDrtZJy7"
      },
      "source": [
        "# Load in the files from Google Drive \n",
        "\n",
        "filename = '/content/drive/My Drive/Face Detection Dataset/data_pickle/X.pickle'\n",
        "pickle_in = open(filename, 'rb')\n",
        "X = pickle.load(pickle_in)\n",
        "\n",
        "filename = '/content/drive/My Drive/Face Detection Dataset/data_pickle/y.pickle'\n",
        "pickle_in = open(filename, 'rb')\n",
        "y = pickle.load(pickle_in)\n",
        "\n",
        "CATEGORIES = ['Without Mask', 'Incorrectly Wearing Mask', 'Wearing Mask']"
      ],
      "execution_count": 2,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "dIsabxBQaxwn",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "16a7aee5-3631-4734-8932-5f73d2e0997c"
      },
      "source": [
        "# double check to see the types of the loaded files\n",
        "print('Type of X:', type(X))\n",
        "print('Type of y:', type(y))"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Type of X: <class 'numpy.ndarray'>\n",
            "Type of y: <class 'numpy.ndarray'>\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e-_uuqKJjNXs",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cc3de2a8-337c-4e41-cfea-5ee068ad0fa5"
      },
      "source": [
        "print('Shape of X:', X.shape)\n",
        "print('Shape of y:', y.shape)"
      ],
      "execution_count": 4,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Shape of X: (17687, 224, 224)\n",
            "Shape of y: (17687,)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mOQ9lDgKa9ce",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c9c02854-e1ae-473d-ea2f-934d9fa5eb89"
      },
      "source": [
        "# print one example of dataset\n",
        "print('X:')\n",
        "print(X[13])\n",
        "print('y:')\n",
        "print(y[13])"
      ],
      "execution_count": 5,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "X:\n",
            "[[ 21  21  21 ... 243 243 243]\n",
            " [ 21  21  21 ... 243 243 243]\n",
            " [ 21  21  21 ... 243 243 243]\n",
            " ...\n",
            " [ 97  97  98 ...  51  52  52]\n",
            " [ 97  97  97 ...  51  52  52]\n",
            " [ 97  97  97 ...  51  52  52]]\n",
            "y:\n",
            "0\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 298
        },
        "id": "weVpXB2GFOgh",
        "outputId": "a52d952d-62c2-4351-d553-30178e222919"
      },
      "source": [
        "# Display the one example\n",
        "plt.imshow(X[10], cmap='gray')\n",
        "plt.title(CATEGORIES[y[10]], color='white')\n",
        "print('')"
      ],
      "execution_count": 6,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "\n"
          ],
          "name": "stdout"
        },
        {
          "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": "ktzFJvIRGQJz"
      },
      "source": [
        "#### EDA"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bvnLTGU2GPgP"
      },
      "source": [
        "resized_X = []\n",
        "for img in X:\n",
        "    resized_X.append(cv2.resize(img, (64, 64)))"
      ],
      "execution_count": 7,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Eqzux7kAGWIX",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "61eec3b5-35e6-4ea2-af46-22c43fc232ee"
      },
      "source": [
        "X = np.asarray(resized_X)\n",
        "X = X.reshape(-1, 64, 64, 1)\n",
        "print(X.shape)"
      ],
      "execution_count": 8,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(17687, 64, 64, 1)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yG3key7RGbwT"
      },
      "source": [
        "# normalize the pixel values\n",
        "X = X / 255.0"
      ],
      "execution_count": 9,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "SScCCjmIGmQj",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "db8b3dd3-7761-4a4f-9150-962f7b8dc734"
      },
      "source": [
        "IMG_DIM = X.shape[1]\n",
        "print('IMG_DIM:',IMG_DIM)"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "IMG_DIM: 64\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "s13FHAbgG1sj"
      },
      "source": [
        "# Split dataset into Training and Testing\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)"
      ],
      "execution_count": 11,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ipm0aO5lJbCL",
        "outputId": "ba4e8ccf-7098-463a-d1df-45a43937a91b"
      },
      "source": [
        "print('Training Size:', len(X_train))\n",
        "print('Testing  Size:', len(X_test))"
      ],
      "execution_count": 12,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Training Size: 14149\n",
            "Testing  Size: 3538\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "6q0JzNt3GmqB"
      },
      "source": [
        "#### Building and fitting the CNN Model "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jmlMbhIgG3i4"
      },
      "source": [
        "cnn_model = tensorflow.keras.models.Sequential()"
      ],
      "execution_count": 13,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JCbnzA1dHAYa"
      },
      "source": [
        "# Start of Convolution Layers & Maxpooling\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu', input_shape=(IMG_DIM, IMG_DIM, 1)))\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.MaxPool2D())\n",
        "\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.MaxPool2D())\n",
        "\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu', input_shape=(IMG_DIM, IMG_DIM, 1)))\n",
        "cnn_model.add(tensorflow.keras.layers.Conv2D(filters=64, kernel_size=3, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.MaxPool2D())\n",
        "\n",
        "# Start of Neural Nets\n",
        "cnn_model.add(tensorflow.keras.layers.Flatten())\n",
        "\n",
        "cnn_model.add(tensorflow.keras.layers.Dense(512, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.Dropout(0.3))\n",
        "cnn_model.add(tensorflow.keras.layers.Dense(512, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.Dropout(0.3))\n",
        "cnn_model.add(tensorflow.keras.layers.Dense(256, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.Dense(128, activation='relu'))\n",
        "cnn_model.add(tensorflow.keras.layers.Dense(3, activation='softmax'))"
      ],
      "execution_count": 14,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "vQkdPxKdHD0F",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "a1ae4be3-07be-479c-bf08-f6c0a03715d6"
      },
      "source": [
        "cnn_model.summary()"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Model: \"sequential\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "conv2d (Conv2D)              (None, 62, 62, 64)        640       \n",
            "_________________________________________________________________\n",
            "conv2d_1 (Conv2D)            (None, 60, 60, 64)        36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d (MaxPooling2D) (None, 30, 30, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_2 (Conv2D)            (None, 28, 28, 64)        36928     \n",
            "_________________________________________________________________\n",
            "conv2d_3 (Conv2D)            (None, 26, 26, 64)        36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_1 (MaxPooling2 (None, 13, 13, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_4 (Conv2D)            (None, 11, 11, 64)        36928     \n",
            "_________________________________________________________________\n",
            "conv2d_5 (Conv2D)            (None, 9, 9, 64)          36928     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_2 (MaxPooling2 (None, 4, 4, 64)          0         \n",
            "_________________________________________________________________\n",
            "flatten (Flatten)            (None, 1024)              0         \n",
            "_________________________________________________________________\n",
            "dense (Dense)                (None, 512)               524800    \n",
            "_________________________________________________________________\n",
            "dropout (Dropout)            (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_1 (Dense)              (None, 512)               262656    \n",
            "_________________________________________________________________\n",
            "dropout_1 (Dropout)          (None, 512)               0         \n",
            "_________________________________________________________________\n",
            "dense_2 (Dense)              (None, 256)               131328    \n",
            "_________________________________________________________________\n",
            "dense_3 (Dense)              (None, 128)               32896     \n",
            "_________________________________________________________________\n",
            "dense_4 (Dense)              (None, 3)                 387       \n",
            "=================================================================\n",
            "Total params: 1,137,347\n",
            "Trainable params: 1,137,347\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OHpoSlZVHKc8"
      },
      "source": [
        "# Compile the Model\n",
        "cnn_model.compile(optimizer=tensorflow.keras.optimizers.Adam(), loss='sparse_categorical_crossentropy', metrics=['acc'])"
      ],
      "execution_count": 16,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-vypBIABHMNl",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "beca7aad-cef9-4786-c2c1-8ada6d916a9b"
      },
      "source": [
        "# Train Model\n",
        "epochs = 10\n",
        "cnn_model.fit(X_train, y_train, epochs=epochs, validation_split=0.1)"
      ],
      "execution_count": 17,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "398/398 [==============================] - 6s 16ms/step - loss: 0.3631 - acc: 0.8272 - val_loss: 0.1033 - val_acc: 0.9675\n",
            "Epoch 2/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0984 - acc: 0.9685 - val_loss: 0.0944 - val_acc: 0.9717\n",
            "Epoch 3/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0729 - acc: 0.9753 - val_loss: 0.0628 - val_acc: 0.9788\n",
            "Epoch 4/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0522 - acc: 0.9825 - val_loss: 0.0563 - val_acc: 0.9845\n",
            "Epoch 5/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0549 - acc: 0.9823 - val_loss: 0.0883 - val_acc: 0.9703\n",
            "Epoch 6/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0391 - acc: 0.9870 - val_loss: 0.0544 - val_acc: 0.9767\n",
            "Epoch 7/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0365 - acc: 0.9873 - val_loss: 0.0683 - val_acc: 0.9717\n",
            "Epoch 8/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0346 - acc: 0.9904 - val_loss: 0.0704 - val_acc: 0.9753\n",
            "Epoch 9/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0294 - acc: 0.9910 - val_loss: 0.0688 - val_acc: 0.9781\n",
            "Epoch 10/10\n",
            "398/398 [==============================] - 6s 15ms/step - loss: 0.0268 - acc: 0.9912 - val_loss: 0.0365 - val_acc: 0.9866\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<tensorflow.python.keras.callbacks.History at 0x7fe6a01cf5f8>"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 17
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Bu67aEqUHrjq"
      },
      "source": [
        "#### Evalute Performance"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JVTqv1ENHNvL",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "c1121eed-fc94-43c2-c9ff-b2abe9520e2c"
      },
      "source": [
        "# Evaluate performance\n",
        "cnn_model.evaluate(X_test, y_test)"
      ],
      "execution_count": 18,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "111/111 [==============================] - 1s 6ms/step - loss: 0.0380 - acc: 0.9876\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.03797728940844536, 0.9875636100769043]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 18
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_meFAampHuay"
      },
      "source": [
        "y_pred = np.argmax(cnn_model.predict(X_test), axis=-1)"
      ],
      "execution_count": 19,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 690
        },
        "id": "LyfAXslHHz9X",
        "outputId": "531490e9-831f-492c-8a26-23d5ad999e6e"
      },
      "source": [
        "cm = sklearn.metrics.confusion_matrix(y_test, y_pred)\n",
        "\n",
        "plt.figure(figsize=(9, 9))\n",
        "sns.heatmap(cm, annot=True, fmt='.0f', square=True, linewidths=.5, cmap='Blues_r')\n",
        "plt.ylabel('Actual Label')\n",
        "plt.xlabel('Predicted Label')\n",
        "\n",
        "print(sklearn.metrics.classification_report(y_test, y_pred))"
      ],
      "execution_count": 20,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.99      0.98      0.99      1154\n",
            "           1       1.00      0.99      0.99      1179\n",
            "           2       0.98      0.99      0.98      1205\n",
            "\n",
            "    accuracy                           0.99      3538\n",
            "   macro avg       0.99      0.99      0.99      3538\n",
            "weighted avg       0.99      0.99      0.99      3538\n",
            "\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 648x648 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "47jJyhTtHkud"
      },
      "source": [
        "#### Save model for future use"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "1_CCuP0mIHvZ",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e2bcd397-70b9-4b1a-c4e6-8331380744af"
      },
      "source": [
        "path = './cnn'\n",
        "cnn_model.save(path)\n",
        "\n",
        "# To load\n",
        "# model = tensorflow.keras.models.load_model(path)"
      ],
      "execution_count": 21,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/tracking/tracking.py:111: Model.state_updates (from tensorflow.python.keras.engine.training) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/training/tracking/tracking.py:111: Layer.updates (from tensorflow.python.keras.engine.base_layer) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "This property should not be used in TensorFlow 2.0, as updates are applied automatically.\n",
            "INFO:tensorflow:Assets written to: ./cnn/assets\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "e0QgVB6SILAV"
      },
      "source": [
        ""
      ],
      "execution_count": 21,
      "outputs": []
    }
  ]
}