DR_Train.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "DR_Train.ipynb",
      "version": "0.3.2",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "metadata": {
        "id": "S2_f8vvu72Df",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 2133
        },
        "outputId": "d91026e8-c26e-453a-c84c-d817c2a072ec"
      },
      "cell_type": "code",
      "source": [
        "#Training classifier network\n",
        "\n",
        "#import keras,numpy,os, matplotlib libraries\n",
        "import datetime as dt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import keras\n",
        "import os\n",
        "from matplotlib import pyplot as plt\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Model\n",
        "from keras.models import Sequential, load_model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.optimizers import SGD\n",
        "from keras.applications.vgg16 import VGG16\n",
        "\n",
        "#reloading CNN outputs compressed image data  from file\n",
        "features_input = np.load(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/imagedata_train_0_4000.npy\")\n",
        "\n",
        "#Converting labels to categorial one hot encoding\n",
        "labels = pd.read_csv(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/train002Labels.csv\",header = None)\n",
        "print(labels.head())\n",
        "labels = labels.values\n",
        "labels = labels[0:4000,1]\n",
        "print(labels.shape)\n",
        "print(labels[0])\n",
        "labels = keras.utils.to_categorical(labels, num_classes=5)\n",
        "\n",
        "#Adding classifier neural network (Dense + Softmax)\n",
        "new_model = Sequential()\n",
        "#add a fc1 layer with 256 output neurons and a dropout layer1\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#add a fc2 layer with 256 output neurons and a dropout layer2\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#adding a softmax layer with 5 classes\n",
        "new_model.add(Dense(units = 5))\n",
        "new_model.add(keras.layers.Activation('softmax'))\n",
        "#configures the model for training\n",
        "sgd = SGD(lr=0.001, decay=0.0, momentum=0.9,nesterov=True)\n",
        "new_model.compile(loss = 'categorical_crossentropy', optimizer = sgd, metrics=['accuracy'] )\n",
        "\n",
        "#train the model for given number of epochs\n",
        "training_history = {}\n",
        "print(dt.datetime.now())\n",
        "try:\n",
        "    training_history = new_model.fit(x = features_input, y = labels, batch_size = 32, epochs = 30, verbose = 2, validation_split = 0.2)\n",
        "except Exception as error:\n",
        "    print(error)\n",
        "finally:\n",
        "    print(training_history.history)  #validation accuracy and loss, training accuracy and loss\n",
        "    new_model.save('my_model_train002_0_4000.h5')  # creates a HDF5 file 'my_model.h5'\n",
        "    new_model.save_weights('my_model_weigths_train002_0_4000.h5') #creates a HDF5 'my_model_weights.h5' weight file\n",
        "#End of training and va;idation   \n",
        "print(dt.datetime.now())\n",
        "#plotting accuracy of training and validation\n",
        "plt.plot(training_history.history['acc'])\n",
        "plt.plot(training_history.history['val_acc'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "#plotting loss of training and validation\n",
        "plt.plot(training_history.history['loss'])\n",
        "plt.plot(training_history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n"
      ],
      "execution_count": 3,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Using TensorFlow backend.\n"
          ],
          "name": "stderr"
        },
        {
          "output_type": "stream",
          "text": [
            "             0  1\n",
            "0   19500_left  0\n",
            "1  19500_right  0\n",
            "2   19503_left  0\n",
            "3  19503_right  0\n",
            "4   19504_left  1\n",
            "(4000,)\n",
            "0\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Colocations handled automatically by placer.\n",
            "2019-02-21 14:46:26.862966\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/keras/backend/tensorflow_backend.py:3445: calling dropout (from tensorflow.python.ops.nn_ops) with keep_prob is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Please use `rate` instead of `keep_prob`. Rate should be set to `rate = 1 - keep_prob`.\n",
            "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
            "Instructions for updating:\n",
            "Use tf.cast instead.\n",
            "Train on 3200 samples, validate on 800 samples\n",
            "Epoch 1/30\n",
            " - 8s - loss: 4.3477 - acc: 0.7278 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 2/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 3/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 4/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 5/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 6/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 7/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 8/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 9/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 10/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 11/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 12/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 13/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 14/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 15/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 16/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 17/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 18/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 19/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 20/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 21/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 22/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 23/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 24/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 25/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 26/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 27/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 28/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 29/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "Epoch 30/30\n",
            " - 7s - loss: 4.2461 - acc: 0.7366 - val_loss: 4.6138 - val_acc: 0.7137\n",
            "{'val_loss': [4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964, 4.613804807662964], 'val_acc': [0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375, 0.71375], 'loss': [4.34768835067749, 4.246110765933991, 4.246110780239105, 4.246110775470734, 4.246110796928406, 4.2461107730865475, 4.2461107730865475, 4.246110751628875, 4.24611077785492, 4.246110804080963, 4.246110770702362, 4.246110780239105, 4.246110771298408, 4.246110780239105, 4.246110775470734, 4.246110808849335, 4.246110792160034, 4.246110787391663, 4.246110780239105, 4.246110765933991, 4.246110775470734, 4.246110796928406, 4.246110785007477, 4.246110770702362, 4.246110792160034, 4.246110796928406, 4.246110770702362, 4.246110775470734, 4.246110782623291, 4.246110780239105], 'acc': [0.7278125, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625, 0.7365625]}\n",
            "2019-02-21 14:49:59.171336\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfoAAAFnCAYAAABO7YvUAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nRK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "XDGzXN_lgfih",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 121
        },
        "outputId": "2d228155-5dc5-45b7-ba54-1e4ce0dddde1"
      },
      "cell_type": "code",
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/gdrive')"
      ],
      "execution_count": 1,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Go to this URL in a browser: https://accounts.google.com/o/oauth2/auth?client_id=947318989803-6bn6qk8qdgf4n4g3pfee6491hc0brc4i.apps.googleusercontent.com&redirect_uri=urn%3Aietf%3Awg%3Aoauth%3A2.0%3Aoob&scope=email%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdocs.test%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fdrive.photos.readonly%20https%3A%2F%2Fwww.googleapis.com%2Fauth%2Fpeopleapi.readonly&response_type=code\n",
            "\n",
            "Enter your authorization code:\n",
            "··········\n",
            "Mounted at /gdrive\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "wJUpsb6OgvxD",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "b4881fc0-8799-4ce9-b2e4-e9bd5d58cb26"
      },
      "cell_type": "code",
      "source": [
        "!ls \"/gdrive/My Drive/DiabeticRetinopathy_Detection\""
      ],
      "execution_count": 2,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "DR_Train.ipynb\timagedata_train_0_4000.npy  train002Labels.csv\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "metadata": {
        "id": "UVD7r2lCxGBh",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1629
        },
        "outputId": "229a8a27-7f12-4c0e-b1e1-e1d46585a567"
      },
      "cell_type": "code",
      "source": [
        "#Training classifier network\n",
        "\n",
        "#import keras,numpy,os, matplotlib libraries\n",
        "import datetime as dt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import keras\n",
        "import os\n",
        "from matplotlib import pyplot as plt\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Model\n",
        "from keras.models import Sequential, load_model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.optimizers import SGD\n",
        "from keras.applications.vgg16 import VGG16\n",
        "\n",
        "#reloading CNN outputs compressed image data  from file\n",
        "features_input = np.load(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/imagedata_train_0_4000.npy\")\n",
        "\n",
        "#Converting labels to categorial one hot encoding\n",
        "labels = pd.read_csv(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/train002Labels.csv\",header = None)\n",
        "print(labels.head())\n",
        "labels = labels.values\n",
        "labels = labels[0:4000,1]\n",
        "print(labels.shape)\n",
        "print(labels[0])\n",
        "labels = keras.utils.to_categorical(labels, num_classes=5)\n",
        "\n",
        "#Adding classifier neural network (Dense + Softmax)\n",
        "new_model = Sequential()\n",
        "#add a fc1 layer with 128 output neurons and a dropout layer1\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#add a fc2 layer with 128 output neurons and a dropout layer2\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#adding a softmax layer with 5 classes\n",
        "new_model.add(Dense(units = 5))\n",
        "new_model.add(keras.layers.Activation('softmax'))\n",
        "#configures the model for training\n",
        "sgd = SGD(lr=0.00001, decay=0.0, momentum=0.9,nesterov=True)\n",
        "new_model.compile(loss = 'categorical_crossentropy', optimizer = sgd, metrics=['accuracy'] )\n",
        "\n",
        "#train the model for given number of epochs\n",
        "training_history = {}\n",
        "print(dt.datetime.now())\n",
        "try:\n",
        "    training_history = new_model.fit(x = features_input, y = labels, batch_size = 25, epochs = 20, verbose = 2, validation_split = 0.05)\n",
        "except Exception as error:\n",
        "    print(error)\n",
        "finally:\n",
        "    print(training_history.history)  #validation accuracy and loss, training accuracy and loss\n",
        "    new_model.save('my_model_train002_0_4000_128.h5')  # creates a HDF5 file 'my_model.h5'\n",
        "    #new_model.save_weights('my_model_weigths_train002_0_4000_128.h5') #creates a HDF5 'my_model_weights.h5' weight file\n",
        "#End of training and va;idation   \n",
        "print(dt.datetime.now())\n",
        "#plotting accuracy of training and validation\n",
        "plt.plot(training_history.history['acc'])\n",
        "plt.plot(training_history.history['val_acc'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "#plotting loss of training and validation\n",
        "plt.plot(training_history.history['loss'])\n",
        "plt.plot(training_history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n"
      ],
      "execution_count": 7,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "             0  1\n",
            "0   19500_left  0\n",
            "1  19500_right  0\n",
            "2   19503_left  0\n",
            "3  19503_right  0\n",
            "4   19504_left  1\n",
            "(4000,)\n",
            "0\n",
            "2019-02-21 16:02:41.587027\n",
            "Train on 3800 samples, validate on 200 samples\n",
            "Epoch 1/20\n",
            " - 10s - loss: 5.2355 - acc: 0.6176 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 2/20\n",
            " - 10s - loss: 4.3614 - acc: 0.7166 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 3/20\n",
            " - 10s - loss: 4.3598 - acc: 0.7174 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 4/20\n",
            " - 10s - loss: 4.3886 - acc: 0.7166 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 5/20\n",
            " - 10s - loss: 4.2920 - acc: 0.7200 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 6/20\n",
            " - 10s - loss: 4.3025 - acc: 0.7100 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 7/20\n",
            " - 10s - loss: 4.2498 - acc: 0.7084 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 8/20\n",
            " - 10s - loss: 4.2892 - acc: 0.6968 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 9/20\n",
            " - 10s - loss: 4.2145 - acc: 0.7000 - val_loss: 3.5973 - val_acc: 0.7750\n",
            "Epoch 10/20\n",
            " - 10s - loss: 4.0600 - acc: 0.6847 - val_loss: 3.5147 - val_acc: 0.7750\n",
            "Epoch 11/20\n",
            " - 10s - loss: 4.0658 - acc: 0.6945 - val_loss: 3.4705 - val_acc: 0.7750\n",
            "Epoch 12/20\n",
            " - 10s - loss: 4.0276 - acc: 0.6805 - val_loss: 3.2337 - val_acc: 0.7750\n",
            "Epoch 13/20\n",
            " - 10s - loss: 3.9265 - acc: 0.6747 - val_loss: 2.9801 - val_acc: 0.7750\n",
            "Epoch 14/20\n",
            " - 10s - loss: 3.8161 - acc: 0.6668 - val_loss: 2.9651 - val_acc: 0.7750\n",
            "Epoch 15/20\n",
            " - 10s - loss: 3.6833 - acc: 0.6661 - val_loss: 2.4692 - val_acc: 0.7750\n",
            "Epoch 16/20\n",
            " - 10s - loss: 3.4551 - acc: 0.6632 - val_loss: 2.3870 - val_acc: 0.7750\n",
            "Epoch 17/20\n",
            " - 10s - loss: 3.2617 - acc: 0.6421 - val_loss: 2.1381 - val_acc: 0.7750\n",
            "Epoch 18/20\n",
            " - 10s - loss: 3.1479 - acc: 0.6353 - val_loss: 1.9761 - val_acc: 0.7750\n",
            "Epoch 19/20\n",
            " - 10s - loss: 2.8782 - acc: 0.6337 - val_loss: 1.8710 - val_acc: 0.7750\n",
            "Epoch 20/20\n",
            " - 10s - loss: 2.7471 - acc: 0.6350 - val_loss: 1.6619 - val_acc: 0.7750\n",
            "{'val_loss': [3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.5973427444696426, 3.5146952718496323, 3.4705040901899338, 3.233675628900528, 2.9800971895456314, 2.96514992415905, 2.4692172706127167, 2.387048602104187, 2.1380801796913147, 1.9760978892445564, 1.8709987327456474, 1.661928467452526], 'val_acc': [0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000059604645, 0.7750000059604645, 0.7750000059604645, 0.7750000059604645], 'loss': [5.235508841903586, 4.361369189463164, 4.3597689384692595, 4.388561800514397, 4.292030370941288, 4.302456311097271, 4.249843980136671, 4.289206642853586, 4.2144668196377, 4.059972186229731, 4.065849254005833, 4.027577425304212, 3.9264772259875347, 3.816132583900502, 3.683254032542831, 3.455065273140606, 3.261722390588961, 3.147918876456587, 2.8781946917113506, 2.747133882422196], 'acc': [0.6176315820040671, 0.7165789512035093, 0.7173684233505475, 0.7165789486546266, 0.720000001160722, 0.7100000057957674, 0.7084210550314501, 0.6968421098825178, 0.7000000041566397, 0.6847368465050271, 0.6944736863829588, 0.680526318714807, 0.6747368448658994, 0.6668421112393078, 0.6660526343866399, 0.663157899913035, 0.642105265471496, 0.6352631616357126, 0.6336842157730931, 0.6350000020312635]}\n",
            "2019-02-21 16:06:03.017061\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\niIiYjIq3iIiIyfw/WHjkBn2iJI4AAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "O5PBKHCez22P",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1293
        },
        "outputId": "a159c8e0-0b4a-46df-9971-247c8bc54314"
      },
      "cell_type": "code",
      "source": [
        "#Training classifier network\n",
        "\n",
        "#import keras,numpy,os, matplotlib libraries\n",
        "import datetime as dt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import keras\n",
        "import os\n",
        "from matplotlib import pyplot as plt\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Model\n",
        "from keras.models import Sequential, load_model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.optimizers import SGD\n",
        "from keras.applications.vgg16 import VGG16\n",
        "\n",
        "#reloading CNN outputs compressed image data  from file\n",
        "features_input = np.load(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/imagedata_train_0_4000.npy\")\n",
        "\n",
        "#Converting labels to categorial one hot encoding\n",
        "labels = pd.read_csv(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/train002Labels.csv\",header = None)\n",
        "print(labels.head())\n",
        "labels = labels.values\n",
        "labels = labels[0:4000,1]\n",
        "print(labels.shape)\n",
        "print(labels[0])\n",
        "labels = keras.utils.to_categorical(labels, num_classes=5)\n",
        "\n",
        "#Adding classifier neural network (Dense + Softmax)\n",
        "new_model = Sequential()\n",
        "#add a fc1 layer with 128 output neurons and a dropout layer1\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#add a fc2 layer with 128 output neurons and a dropout layer2\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#adding a softmax layer with 5 classes\n",
        "new_model.add(Dense(units = 5))\n",
        "new_model.add(keras.layers.Activation('softmax'))\n",
        "#configures the model for training\n",
        "sgd = SGD(lr=0.0001, decay=1e-6, momentum=0.9,nesterov=True)\n",
        "new_model.compile(loss = 'categorical_crossentropy', optimizer = sgd, metrics=['accuracy'] )\n",
        "\n",
        "#train the model for given number of epochs\n",
        "training_history = {}\n",
        "print(dt.datetime.now())\n",
        "try:\n",
        "    training_history = new_model.fit(x = features_input, y = labels, batch_size = 25, epochs = 10, verbose = 2, validation_split = 0.05)\n",
        "except Exception as error:\n",
        "    print(error)\n",
        "finally:\n",
        "    print(training_history.history)  #validation accuracy and loss, training accuracy and loss\n",
        "    #new_model.save('my_model_train002_0_4000_128.h5')  # creates a HDF5 file 'my_model.h5'\n",
        "    #new_model.save_weights('my_model_weigths_train002_0_4000_128.h5') #creates a HDF5 'my_model_weights.h5' weight file\n",
        "#End of training and va;idation   \n",
        "print(dt.datetime.now())\n",
        "#plotting accuracy of training and validation\n",
        "plt.plot(training_history.history['acc'])\n",
        "plt.plot(training_history.history['val_acc'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "#plotting loss of training and validation\n",
        "plt.plot(training_history.history['loss'])\n",
        "plt.plot(training_history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n"
      ],
      "execution_count": 10,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "             0  1\n",
            "0   19500_left  0\n",
            "1  19500_right  0\n",
            "2   19503_left  0\n",
            "3  19503_right  0\n",
            "4   19504_left  1\n",
            "(4000,)\n",
            "0\n",
            "2019-02-21 16:13:33.619026\n",
            "Train on 3800 samples, validate on 200 samples\n",
            "Epoch 1/10\n",
            " - 11s - loss: 4.5251 - acc: 0.7084 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 2/10\n",
            " - 10s - loss: 4.3564 - acc: 0.7292 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 3/10\n",
            " - 10s - loss: 4.3562 - acc: 0.7295 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 4/10\n",
            " - 10s - loss: 4.3544 - acc: 0.7297 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 5/10\n",
            " - 10s - loss: 4.3557 - acc: 0.7289 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 6/10\n",
            " - 10s - loss: 4.3557 - acc: 0.7297 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 7/10\n",
            " - 10s - loss: 4.3562 - acc: 0.7292 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 8/10\n",
            " - 10s - loss: 4.3552 - acc: 0.7297 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 9/10\n",
            " - 10s - loss: 4.3551 - acc: 0.7292 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 10/10\n",
            " - 10s - loss: 4.3393 - acc: 0.7271 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "{'val_loss': [3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196], 'val_acc': [0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451], 'loss': [4.52508638563909, 4.356377490256962, 4.3562414340282745, 4.35435045079181, 4.35567150304192, 4.355727372200866, 4.356160834431648, 4.355170369932526, 4.355092774096288, 4.339329261528818], 'acc': [0.7084210566980275, 0.7292105288882005, 0.7294736884926495, 0.7297368465285552, 0.7289473730090418, 0.7297368443718082, 0.72921052986854, 0.7297368445678761, 0.7292105304567438, 0.7271052683262449]}\n",
            "2019-02-21 16:15:14.369944\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAfUAAAFnCAYAAAC/5tBZAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\noQAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhiG8AQAwzP8Hjoqe4GrMrpEAAAAASUVORK5CYII=\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    },
    {
      "metadata": {
        "id": "oZ6VLyQm10lg",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1293
        },
        "outputId": "2b301922-5f80-4545-f811-42d2bd57a37f"
      },
      "cell_type": "code",
      "source": [
        "#Training classifier network\n",
        "\n",
        "#import keras,numpy,os, matplotlib libraries\n",
        "import datetime as dt\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "import keras\n",
        "import os\n",
        "from matplotlib import pyplot as plt\n",
        "from keras.preprocessing.image import ImageDataGenerator\n",
        "from keras.models import Model\n",
        "from keras.models import Sequential, load_model\n",
        "from keras.layers import Dense, Dropout, Flatten\n",
        "from keras.layers import Conv2D, MaxPooling2D\n",
        "from keras.optimizers import SGD\n",
        "from keras.applications.vgg16 import VGG16\n",
        "\n",
        "#reloading CNN outputs compressed image data  from file\n",
        "features_input = np.load(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/imagedata_train_0_4000.npy\")\n",
        "\n",
        "#Converting labels to categorial one hot encoding\n",
        "labels = pd.read_csv(\"/gdrive/My Drive/DiabeticRetinopathy_Detection/train002Labels.csv\",header = None)\n",
        "print(labels.head())\n",
        "labels = labels.values\n",
        "labels = labels[0:4000,1]\n",
        "print(labels.shape)\n",
        "print(labels[0])\n",
        "labels = keras.utils.to_categorical(labels, num_classes=5)\n",
        "\n",
        "#Adding classifier neural network (Dense + Softmax)\n",
        "new_model = Sequential()\n",
        "#add a fc1 layer with 128 output neurons and a dropout layer1\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#add a fc2 layer with 128 output neurons and a dropout layer2\n",
        "new_model.add(Dense(units = 256, activation = 'relu', kernel_initializer='uniform'))\n",
        "new_model.add(Dropout(rate = 0.5, noise_shape=None, seed=None))\n",
        "#adding a softmax layer with 5 classes\n",
        "new_model.add(Dense(units = 5))\n",
        "new_model.add(keras.layers.Activation('softmax'))\n",
        "#configures the model for training\n",
        "sgd = SGD(lr=0.000001, decay=1e-6, momentum=0.9,nesterov=True)\n",
        "new_model.compile(loss = 'categorical_crossentropy', optimizer = sgd, metrics=['accuracy'] )\n",
        "\n",
        "#train the model for given number of epochs\n",
        "training_history = {}\n",
        "print(dt.datetime.now())\n",
        "try:\n",
        "    training_history = new_model.fit(x = features_input, y = labels, batch_size = 25, epochs = 10, verbose = 2, validation_split = 0.05)\n",
        "except Exception as error:\n",
        "    print(error)\n",
        "finally:\n",
        "    print(training_history.history)  #validation accuracy and loss, training accuracy and loss\n",
        "    #new_model.save('my_model_train002_0_4000_128.h5')  # creates a HDF5 file 'my_model.h5'\n",
        "    #new_model.save_weights('my_model_weigths_train002_0_4000_128.h5') #creates a HDF5 'my_model_weights.h5' weight file\n",
        "#End of training and va;idation   \n",
        "print(dt.datetime.now())\n",
        "#plotting accuracy of training and validation\n",
        "plt.plot(training_history.history['acc'])\n",
        "plt.plot(training_history.history['val_acc'])\n",
        "plt.title('Model Accuracy')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n",
        "#plotting loss of training and validation\n",
        "plt.plot(training_history.history['loss'])\n",
        "plt.plot(training_history.history['val_loss'])\n",
        "plt.title('Model Loss')\n",
        "plt.ylabel('Loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.legend(['train', 'test'], loc='upper left')\n",
        "plt.show()\n",
        "\n"
      ],
      "execution_count": 15,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "             0  1\n",
            "0   19500_left  0\n",
            "1  19500_right  0\n",
            "2   19503_left  0\n",
            "3  19503_right  0\n",
            "4   19504_left  1\n",
            "(4000,)\n",
            "0\n",
            "2019-02-21 16:28:36.917516\n",
            "Train on 3800 samples, validate on 200 samples\n",
            "Epoch 1/10\n",
            " - 11s - loss: 9.8804 - acc: 0.2637 - val_loss: 3.4308 - val_acc: 0.7650\n",
            "Epoch 2/10\n",
            " - 11s - loss: 5.8440 - acc: 0.5489 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 3/10\n",
            " - 10s - loss: 4.8732 - acc: 0.6389 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 4/10\n",
            " - 10s - loss: 4.6264 - acc: 0.6797 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 5/10\n",
            " - 10s - loss: 4.5543 - acc: 0.6895 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 6/10\n",
            " - 10s - loss: 4.5168 - acc: 0.7005 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 7/10\n",
            " - 10s - loss: 4.4710 - acc: 0.7037 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 8/10\n",
            " - 10s - loss: 4.3982 - acc: 0.7108 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 9/10\n",
            " - 10s - loss: 4.4156 - acc: 0.7103 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "Epoch 10/10\n",
            " - 10s - loss: 4.4298 - acc: 0.7118 - val_loss: 3.6266 - val_acc: 0.7750\n",
            "{'val_loss': [3.430809184908867, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196, 3.6265714317560196], 'val_acc': [0.765000008046627, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451, 0.7750000134110451], 'loss': [9.880351599894071, 5.844030325350008, 4.873164972192363, 4.626403985839141, 4.5543055424564765, 4.51679132721926, 4.470955424991093, 4.3982252439385965, 4.415612292917151, 4.429845002136733], 'acc': [0.26368420997536496, 0.5489473662486202, 0.6389473706091705, 0.6797368449992255, 0.6894736876221079, 0.7005263202284512, 0.7036842137183014, 0.7107894753939227, 0.7102631624591978, 0.7118421102824964]}\n",
            "2019-02-21 16:30:20.098307\n"
          ],
          "name": "stdout"
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe8AAAFnCAYAAACPasF4AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIiIKMAxvIiKiAMPwJiIiCjD/H6rmykg1IYjqAAAAAElFTkSuQmCC\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe0AAAFnCAYAAACLnxFFAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEYQ2AAAhgtAGACBE/H88FD8xszw/dAAAAABJRU5ErkJggg==\n",
            "text/plain": [
              "<Figure size 576x396 with 1 Axes>"
            ]
          },
          "metadata": {
            "tags": []
          }
        }
      ]
    }
  ]
}