Swedish_Leaf_CNN_k10.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 0,
  "metadata": {
    "colab": {
      "name": "Swedish_Leaf_CNN_k10",
      "provenance": [],
      "collapsed_sections": []
    },
    "kernelspec": {
      "name": "python3",
      "display_name": "Python 3"
    }
  },
  "cells": [
    {
      "cell_type": "code",
      "metadata": {
        "id": "pZPLL4iKc0EM",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "outputId": "53761758-7b2e-4c14-beb1-9b344dfa19fe"
      },
      "source": [
        "import tensorflow as tf\n",
        "import matplotlib.pyplot as plt\n",
        "import numpy as np\n",
        "import glob\n",
        "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
        "import os\n",
        "from sklearn.model_selection import KFold,StratifiedKFold\n",
        "import random\n",
        "%matplotlib inline\n",
        "tf.__version__"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'2.3.0'"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 1
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ylGsSqpgAaBl",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "87259f4e-0b27-4545-e1fe-9d375b634a86"
      },
      "source": [
        "from google.colab import drive\n",
        "drive.mount('/content/drive')"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "svtiI-4yAaYQ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "97e75785-5603-4f3b-bb94-7012cc8c39f8"
      },
      "source": [
        "%cd 'drive/My Drive/Swedish Leaf'"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "/content/drive/My Drive/Swedish Leaf\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "bssV66ChuEu4",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "import glob\n",
        "Leaf1 = glob.glob('./Leaf1/*.*')\n",
        "Leaf2 = glob.glob('./Leaf2/*.*')\n",
        "Leaf3 = glob.glob('./Leaf3/*.*')\n",
        "Leaf4 = glob.glob('./Leaf4/*.*')\n",
        "Leaf5 = glob.glob('./Leaf5/*.*')\n",
        "Leaf6 = glob.glob('./Leaf6/*.*')\n",
        "Leaf7 = glob.glob('./Leaf7/*.*')\n",
        "Leaf8 = glob.glob('./Leaf8/*.*')\n",
        "Leaf9 = glob.glob('./Leaf9/*.*')\n",
        "Leaf10 = glob.glob('./Leaf10/*.*')\n",
        "Leaf11= glob.glob('./Leaf11/*.*')\n",
        "Leaf12= glob.glob('./Leaf12/*.*')\n",
        "Leaf13= glob.glob('./Leaf13/*.*')\n",
        "Leaf14= glob.glob('./Leaf14/*.*')\n",
        "Leaf15= glob.glob('./Leaf15/*.*')\n",
        "\n",
        "data = []\n",
        "labels = []"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "oi_ijhEPv2mU",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from tensorflow.keras.preprocessing.image import load_img,img_to_array"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "494uI5OduzYM",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf1:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(0)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nOa8gIvwxbAK",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf2:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(1)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mawLIf3uxaag",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf3:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(2)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VfM8Ztozx5iL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf4:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(3)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xKk5nGsXx6ZI",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf5:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(4)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "k2kJwQVWx6wB",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf6:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(5)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xa5kbHKOx6Id",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf7:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(6)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KLSHxx5Y2EDL",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf8:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(7)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y62fvkUG2Evm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf9:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(8)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "FfInGUQL2Fer",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf10:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(9)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "OiOFqNlE2FHO",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf11:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(10)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "2oN7p4f22Edh",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf12:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(11)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MpK4ZoxLx5TY",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf13:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(12)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "PlVqxuyHxpdE",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf14:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(13)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "kPDu8lzV2spr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "for i in Leaf15:   \n",
        "    image=load_img(i)\n",
        "    img_arr=img_to_array(image)\n",
        "    data.append(img_arr)\n",
        "    labels.append(14)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "jZQnJDma_qqJ",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "skf=StratifiedKFold(n_splits=5,shuffle=True,random_state=random.randint(0,100))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "TQ8UHUXb_rPw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "idg=ImageDataGenerator(rescale=1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "wNaGq2Ik_q_G",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "def get_model_name(k):\n",
        "    return 'model_'+str(k)+'.h5'"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "afKGr-LieVNw",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from keras.utils.np_utils import to_categorical   \n",
        "\n",
        "categorical_labels = to_categorical(labels, num_classes=15)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "_IeZdGfYeUVR",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 34
        },
        "outputId": "c82fe083-12ef-4855-cbe7-cf1449f0aa7f"
      },
      "source": [
        "data1 = np.array(data)\n",
        "print(data1.shape)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "(6000, 128, 128, 3)\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mDwpf5jYDBt1",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "train_datagen=ImageDataGenerator(rescale=1./255)\n",
        "val_datagen = ImageDataGenerator(rescale=1./255)\n",
        "test_datagen = ImageDataGenerator(rescale=1./255)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab_type": "code",
        "id": "OVtFc-bwIlYZ",
        "colab": {}
      },
      "source": [
        "model=tf.keras.models.Sequential()\n",
        "model.add(tf.keras.layers.Conv2D(32,(5,5),strides=(1,1),padding='same',activation='relu',input_shape=[128,128,3]))\n",
        "model.add(tf.keras.layers.MaxPool2D(pool_size=(2,2),padding='same'))\n",
        "  \n",
        "model.add(tf.keras.layers.Conv2D(64,(5,5),activation='relu',padding='same',strides=(1,1)))\n",
        "model.add(tf.keras.layers.MaxPool2D(pool_size=(2,2),padding='same'))\n",
        "model.add(tf.keras.layers.BatchNormalization())\n",
        "\n",
        "model.add(tf.keras.layers.Conv2D(96,(3,3),activation='relu',strides=(2,2),padding='same'))\n",
        "model.add(tf.keras.layers.MaxPool2D(pool_size=(2,2),padding='valid'))\n",
        "model.add(tf.keras.layers.BatchNormalization())\n",
        "\n",
        "model.add(tf.keras.layers.Flatten())\n",
        "model.add(tf.keras.layers.Dense(units=100,activation='relu'))\n",
        "model.add(tf.keras.layers.Dense(units=100,activation='relu'))\n",
        "model.add(tf.keras.layers.Dropout(0.25))\n",
        "\n",
        "model.add(tf.keras.layers.Dense(15,activation='softmax'))\n",
        "  \n",
        "model.compile(loss='categorical_crossentropy',optimizer='adam',metrics=['accuracy'])\n",
        "\n",
        "\n",
        "\n",
        "\n",
        " "
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-Musb_hTWr11",
        "colab_type": "text"
      },
      "source": [
        "K-Fold Cross Validation\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ChR5qNGyW0J3",
        "colab_type": "text"
      },
      "source": [
        "For K=10 Folds\n"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "5QPP_QNuYBCa",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "validation_accuracy_k10=[]\n",
        "validation_loss_k10=[]\n",
        "fold_var=1"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "96TYXxgmDBKe",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from sklearn.model_selection import train_test_split\n",
        "X_train1, X_test0, ytrain1, ytest0 = train_test_split(data1, categorical_labels, test_size=0.1,\n",
        "                                                    random_state=random.randint(0,100))"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "x2akFS2-hPvY",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 1000
        },
        "outputId": "7c9dda24-ad2c-4613-845f-c390d21ee54c"
      },
      "source": [
        "for index in range(10):#readings change for every epoch\n",
        "    \n",
        "    X_train, X_test1, ytrain, ytest1 = train_test_split(X_train1, ytrain1, test_size=0.2,\n",
        "                                                    random_state=random.randint(0,100))\n",
        "    \n",
        "    X_val, X_test, yval, ytest = train_test_split(X_test1, ytest1, test_size=0.5,\n",
        "                                                    random_state=random.randint(0,100))\n",
        "    \n",
        "    \n",
        "    \n",
        "    \n",
        "   \n",
        "  \n",
        "    \n",
        "    training_set = train_datagen.flow(\n",
        "            X_train, ytrain,\n",
        "            batch_size=64)\n",
        "    \n",
        "    val_set = val_datagen.flow(\n",
        "            X_val, yval, \n",
        "            batch_size=64)\n",
        "    \n",
        "    test_set = test_datagen.flow(\n",
        "            X_test, ytest,\n",
        "            batch_size=64)\n",
        "    \n",
        "    X, y = test_set.next()\n",
        "    history=model.fit_generator(\n",
        "            training_set,\n",
        "            steps_per_epoch=20,\n",
        "            epochs=10,\n",
        "            validation_data=val_set,\n",
        "            validation_steps=100\n",
        "            )\n",
        "     \n",
        "\n",
        "    \n",
        "    \n",
        "    results = model.evaluate(X,y)\n",
        "    results1=dict(zip(model.metrics_names,results))\n",
        "    validation_accuracy_k10.append(results1['accuracy'])\n",
        "    validation_loss_k10.append(results1['loss'])\n",
        "\n",
        "    tf.keras.backend.clear_session()\n",
        "    fold_var+=1\n",
        "    \n",
        "    "
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 2.4416 - accuracy: 0.1953WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 59s 3s/step - loss: 2.4416 - accuracy: 0.1953 - val_loss: 2.6226 - val_accuracy: 0.1056\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 2.2362 - accuracy: 0.2398\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 2.0988 - accuracy: 0.2883\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 1.8336 - accuracy: 0.3742\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 52s 3s/step - loss: 1.7363 - accuracy: 0.4119\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 1.6462 - accuracy: 0.4484\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 1.6473 - accuracy: 0.4352\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 1.5020 - accuracy: 0.4820\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 52s 3s/step - loss: 1.4590 - accuracy: 0.4896\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 1.3510 - accuracy: 0.5234\n",
            "2/2 [==============================] - 0s 183ms/step - loss: 2.7942 - accuracy: 0.1562\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 1.1916 - accuracy: 0.5758WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 61s 3s/step - loss: 1.1916 - accuracy: 0.5758 - val_loss: 3.1702 - val_accuracy: 0.0963\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 1.2028 - accuracy: 0.5672\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 1.1997 - accuracy: 0.5844\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 1.0960 - accuracy: 0.6164\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 1.0310 - accuracy: 0.6450\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.8848 - accuracy: 0.6945\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.8167 - accuracy: 0.7023\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.8560 - accuracy: 0.7067\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.7642 - accuracy: 0.7289\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.7500 - accuracy: 0.7359\n",
            "2/2 [==============================] - 0s 187ms/step - loss: 2.2125 - accuracy: 0.3125\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.7808 - accuracy: 0.7289WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 62s 3s/step - loss: 0.7808 - accuracy: 0.7289 - val_loss: 1.6667 - val_accuracy: 0.4667\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.7356 - accuracy: 0.7484\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.6377 - accuracy: 0.7836\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.6360 - accuracy: 0.7766\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.6292 - accuracy: 0.7711\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.5314 - accuracy: 0.8188\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 57s 3s/step - loss: 0.5033 - accuracy: 0.8242\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.4921 - accuracy: 0.8317\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.4926 - accuracy: 0.8253\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.4844 - accuracy: 0.8344\n",
            "2/2 [==============================] - 0s 181ms/step - loss: 0.6240 - accuracy: 0.8281\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.5246 - accuracy: 0.8156WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 61s 3s/step - loss: 0.5246 - accuracy: 0.8156 - val_loss: 1.2955 - val_accuracy: 0.6315\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.4420 - accuracy: 0.8574\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.4283 - accuracy: 0.8648\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.4444 - accuracy: 0.8617\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.3742 - accuracy: 0.8734\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.3413 - accuracy: 0.8846\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.3307 - accuracy: 0.8969\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.3156 - accuracy: 0.8977\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.3111 - accuracy: 0.9006\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.2655 - accuracy: 0.9164\n",
            "2/2 [==============================] - 0s 182ms/step - loss: 0.5633 - accuracy: 0.8438\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.2805 - accuracy: 0.9087WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 59s 3s/step - loss: 0.2805 - accuracy: 0.9087 - val_loss: 0.2628 - val_accuracy: 0.9167\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.2242 - accuracy: 0.9258\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.2357 - accuracy: 0.9247\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.2212 - accuracy: 0.9289\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 57s 3s/step - loss: 0.2220 - accuracy: 0.9305\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.1822 - accuracy: 0.9352\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.1841 - accuracy: 0.9414\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.1509 - accuracy: 0.9495\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.1746 - accuracy: 0.9461\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 57s 3s/step - loss: 0.1693 - accuracy: 0.9453\n",
            "2/2 [==============================] - 0s 181ms/step - loss: 0.8729 - accuracy: 0.7500\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.1904 - accuracy: 0.9352WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 63s 3s/step - loss: 0.1904 - accuracy: 0.9352 - val_loss: 0.1658 - val_accuracy: 0.9537\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1958 - accuracy: 0.9367\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1761 - accuracy: 0.9484\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1519 - accuracy: 0.9539\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1496 - accuracy: 0.9516\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1685 - accuracy: 0.9406\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.1246 - accuracy: 0.9567\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1338 - accuracy: 0.9570\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.1331 - accuracy: 0.9539\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 58s 3s/step - loss: 0.1203 - accuracy: 0.9641\n",
            "2/2 [==============================] - 0s 184ms/step - loss: 0.3908 - accuracy: 0.8906\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.1224 - accuracy: 0.9648WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 62s 3s/step - loss: 0.1224 - accuracy: 0.9648 - val_loss: 0.2476 - val_accuracy: 0.9148\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.1277 - accuracy: 0.9602\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.1052 - accuracy: 0.9602\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0868 - accuracy: 0.9719\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0872 - accuracy: 0.9680\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0931 - accuracy: 0.9688\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0557 - accuracy: 0.9824\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 57s 3s/step - loss: 0.0717 - accuracy: 0.9789\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0563 - accuracy: 0.9781\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.0498 - accuracy: 0.9856\n",
            "2/2 [==============================] - 0s 182ms/step - loss: 0.1956 - accuracy: 0.9375\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.0572 - accuracy: 0.9820WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 65s 3s/step - loss: 0.0572 - accuracy: 0.9820 - val_loss: 0.5516 - val_accuracy: 0.8630\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 59s 3s/step - loss: 0.0922 - accuracy: 0.9719\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 57s 3s/step - loss: 0.0934 - accuracy: 0.9663\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0841 - accuracy: 0.9703\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0533 - accuracy: 0.9875\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0636 - accuracy: 0.9808\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0806 - accuracy: 0.9758\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0629 - accuracy: 0.9816\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0606 - accuracy: 0.9836\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0592 - accuracy: 0.9832\n",
            "2/2 [==============================] - 0s 187ms/step - loss: 0.0861 - accuracy: 0.9531\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.0686 - accuracy: 0.9797WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 63s 3s/step - loss: 0.0686 - accuracy: 0.9797 - val_loss: 0.0488 - val_accuracy: 0.9870\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0612 - accuracy: 0.9828\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0490 - accuracy: 0.9832\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0622 - accuracy: 0.9824\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0427 - accuracy: 0.9883\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0584 - accuracy: 0.9820\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0309 - accuracy: 0.9898\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0419 - accuracy: 0.9875\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0572 - accuracy: 0.9836\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 56s 3s/step - loss: 0.0652 - accuracy: 0.9766\n",
            "2/2 [==============================] - 0s 182ms/step - loss: 1.2307 - accuracy: 0.8125\n",
            "Epoch 1/10\n",
            "20/20 [==============================] - ETA: 0s - loss: 0.0532 - accuracy: 0.9805WARNING:tensorflow:Your input ran out of data; interrupting training. Make sure that your dataset or generator can generate at least `steps_per_epoch * epochs` batches (in this case, 100 batches). You may need to use the repeat() function when building your dataset.\n",
            "20/20 [==============================] - 62s 3s/step - loss: 0.0532 - accuracy: 0.9805 - val_loss: 1.5958 - val_accuracy: 0.7352\n",
            "Epoch 2/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0564 - accuracy: 0.9750\n",
            "Epoch 3/10\n",
            "20/20 [==============================] - 54s 3s/step - loss: 0.0609 - accuracy: 0.9808\n",
            "Epoch 4/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0910 - accuracy: 0.9719\n",
            "Epoch 5/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.0822 - accuracy: 0.9720\n",
            "Epoch 6/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0774 - accuracy: 0.9797\n",
            "Epoch 7/10\n",
            "20/20 [==============================] - 53s 3s/step - loss: 0.0858 - accuracy: 0.9752\n",
            "Epoch 8/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0578 - accuracy: 0.9766\n",
            "Epoch 9/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0291 - accuracy: 0.9891\n",
            "Epoch 10/10\n",
            "20/20 [==============================] - 55s 3s/step - loss: 0.0290 - accuracy: 0.9914\n",
            "2/2 [==============================] - 0s 191ms/step - loss: 0.3402 - accuracy: 0.9062\n"
          ],
          "name": "stdout"
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DbzYrWzWnfVN",
        "colab_type": "text"
      },
      "source": [
        "Plotting Training Accuracy and Loss "
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "C9YBzctKmgVr",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "acc = history.history['accuracy']\n",
        "\n",
        "loss = history.history['loss']"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QfpK-jhPnRst",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 513
        },
        "outputId": "d99458f2-6449-4985-cd2f-6b001cbbfc82"
      },
      "source": [
        "plt.figure(figsize=(8, 8))\n",
        "plt.subplot(2, 1, 1)\n",
        "plt.plot(acc, label='Training Accuracy')\n",
        "plt.legend(loc='lower right')\n",
        "plt.ylabel('Accuracy')\n",
        "plt.ylim([min(plt.ylim()),1])\n",
        "plt.title('Training Accuracy')\n",
        "\n",
        "plt.subplot(2, 1, 2)\n",
        "plt.plot(loss, label='Training Loss')\n",
        "plt.legend(loc='upper right')\n",
        "plt.ylabel('Cross Entropy')\n",
        "plt.ylim([0,4.0])\n",
        "plt.title('Training Loss')\n",
        "plt.xlabel('epoch')\n",
        "plt.show()"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 576x576 with 2 Axes>"
            ]
          },
          "metadata": {
            "tags": [],
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9v82wCzeyQTK",
        "colab_type": "text"
      },
      "source": [
        "Determining the Mean Accuracy and Loss on the Validation Set"
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zvJdMucbyFrH",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        },
        "outputId": "76825eb0-4e65-44ac-c76a-39ce86acee46"
      },
      "source": [
        "validation_accuracy_k10"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[0.15625,\n",
              " 0.3125,\n",
              " 0.828125,\n",
              " 0.84375,\n",
              " 0.75,\n",
              " 0.890625,\n",
              " 0.9375,\n",
              " 0.953125,\n",
              " 0.8125,\n",
              " 0.90625]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ymuogaajzjpZ",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 187
        },
        "outputId": "5c5eadc5-2fbf-4238-c373-1fecaef895b6"
      },
      "source": [
        "validation_loss_k10"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "[2.7942121028900146,\n",
              " 2.212526321411133,\n",
              " 0.6239731311798096,\n",
              " 0.563250720500946,\n",
              " 0.8728958368301392,\n",
              " 0.3907765746116638,\n",
              " 0.19561173021793365,\n",
              " 0.08605296909809113,\n",
              " 1.2306711673736572,\n",
              " 0.340165913105011]"
            ]
          },
          "metadata": {
            "tags": []
          },
          "execution_count": 40
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "JdGEpBEfzlQm",
        "colab_type": "code",
        "colab": {}
      },
      "source": [
        "from statistics import mean\n",
        "mean_accuracy=mean(validation_accuracy_k10)\n",
        "mean_loss=mean(validation_loss_k10)"
      ],
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "-aqt5CkyzqQU",
        "colab_type": "code",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 51
        },
        "outputId": "da1b1b1c-e4ea-463c-b2dd-962da6c2a4e8"
      },
      "source": [
        "print(\"Mean validation Accuracy value: \",mean_accuracy*100,\"%\")\n",
        "print(\"Mean validation loss value: \",mean_loss)"
      ],
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "text": [
            "Mean validation Accuracy value:  73.90625 %\n",
            "Mean validation loss value:  0.93101364672184\n"
          ],
          "name": "stdout"
        }
      ]
    }
  ]
}