2D_CNN+LSTM_No_Splint.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/alexsalman/CSE247/blob/main/2D_CNN%2BLSTM_No_Splint.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "Fk8TLrPiv-ab"
      },
      "source": [
        "####**Convolutional Neural Network + Long Short Term Memory**\n",
        "######*I am using a Convolution Neural Network (CNN) + Long Short Term Memory (LSTM) Network to extract general representation while utilizing the Spatial-temporal aspect of the videos.*"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 1,
      "metadata": {
        "id": "M8ibtd5HKtZk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "df8a103f-31ac-4ee5-9901-58b7e5f34e11"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "2.8.0\n"
          ]
        }
      ],
      "source": [
        "# required libraries\n",
        "import os\n",
        "import cv2\n",
        "import random\n",
        "import numpy as np\n",
        "import datetime as dt\n",
        "import matplotlib.pyplot as plt\n",
        "from google.colab import drive\n",
        "from sklearn.model_selection import train_test_split\n",
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "from keras.models import Sequential\n",
        "from keras.layers import Conv2D, Flatten, Dense, Dropout, Activation, BatchNormalization\n",
        "from keras.layers import TimeDistributed, MaxPooling2D, GlobalAveragePooling2D\n",
        "from keras.layers.recurrent import LSTM\n",
        "from keras import regularizers\n",
        "%matplotlib inline\n",
        "print(tf.version.VERSION)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 2,
      "metadata": {
        "id": "iffdFOf1CEAN"
      },
      "outputs": [],
      "source": [
        "# set Numpy, Python, and Tensorflow seeds to get consistent results on every execution\n",
        "seed_constant = 27\n",
        "np.random.seed(seed_constant)\n",
        "random.seed(seed_constant)\n",
        "tf.random.set_seed(seed_constant)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 3,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 53
        },
        "id": "mcLh22LiOHyn",
        "outputId": "c1571292-6dfa-4a7e-bb88-ef04fa162b86"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Mounted at /content/gdrive/\n"
          ]
        },
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "'/content/gdrive/My Drive/247'"
            ],
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            }
          },
          "metadata": {},
          "execution_count": 3
        }
      ],
      "source": [
        "# mount dataset from google drive\n",
        "drive.mount('/content/gdrive/', force_remount=True)\n",
        "gdrive_path = '/content/gdrive' + '/My Drive/247/'\n",
        "os.chdir(gdrive_path)\n",
        "os.getcwd()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 4,
      "metadata": {
        "id": "oeDK8SzumZ1Q"
      },
      "outputs": [],
      "source": [
        "# frame dimention\n",
        "IMAGE_HEIGHT, IMAGE_WIDTH = 96, 96\n",
        "# frame number for each video (depth)\n",
        "SEQUENCE_LENGTH = 16\n",
        "DATASET_DIR = gdrive_path + 'Circle_Cropped_videos_labeled'\n",
        "# labels of classes\n",
        "CLASSES_LIST = ['hemostasis', 'inflammatory', 'proliferative', 'maturation']"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 5,
      "metadata": {
        "id": "3mFB5qD6b3Kd"
      },
      "outputs": [],
      "source": [
        "# image cropping\n",
        "def crop_center_square(frame):\n",
        "    y, x = frame.shape[0:2]\n",
        "    min_dim = min(y, x)\n",
        "    start_x = (x // 2) - (min_dim // 2)\n",
        "    start_y = (y // 2) - (min_dim // 2)\n",
        "    return frame[start_y:start_y+min_dim,start_x:start_x+min_dim]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 6,
      "metadata": {
        "id": "sqeexMUjaLzJ"
      },
      "outputs": [],
      "source": [
        "def load_video(path, resize=(96, 96)):\n",
        "    video_reader = cv2.VideoCapture(path)\n",
        "    frames = []\n",
        "    try:\n",
        "        while True:\n",
        "            ret, frame = video_reader.read()\n",
        "            if not ret:\n",
        "                  break\n",
        "            frame = crop_center_square(frame)\n",
        "            frame = cv2.resize(frame, resize)\n",
        "            frame = frame[:, :, [2, 1, 0]]\n",
        "            frames.append(frame)\n",
        "    finally:\n",
        "        video_reader.release()\n",
        "    return np.array(frames) / 255.0"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 7,
      "metadata": {
        "id": "ljUWHW6Jqzu-"
      },
      "outputs": [],
      "source": [
        "def create_dataset(state):\n",
        "    # Declared Empty Lists to store the features, labels and video file path values.\n",
        "    features = []\n",
        "    labels = []\n",
        "    video_files_paths = []\n",
        "    # Iterating through all the classes mentioned in the classes list\n",
        "    for class_index, class_name in enumerate(CLASSES_LIST):\n",
        "        # Display the name of the class whose data is being extracted.\n",
        "        print(f'Extracting Data of Class: {class_name} {state}')\n",
        "        # Get the list of video files present in the specific class name directory.\n",
        "        files_list = os.listdir(os.path.join(DATASET_DIR, class_name))\n",
        "        # Iterate through all the files present in the files list.\n",
        "        for file_name in files_list:\n",
        "            # Get the complete video path.\n",
        "            video_file_path = os.path.join(DATASET_DIR, class_name, file_name)\n",
        "            # create testing data\n",
        "            if state == 'test':\n",
        "                # get the mouse number\n",
        "                mouse_number = int(video_file_path.split(' ')[2].split('_')[1].split('-')[1])\n",
        "                # get the mouse side (L or R)\n",
        "                mouse_side = video_file_path.split(' ')[2].split('_')[1].split('-')[2]\n",
        "                if mouse_number == 4 and mouse_side == 'L':\n",
        "                    frames = load_video(video_file_path)\n",
        "                    features.append(frames)\n",
        "                    labels.append(class_index)\n",
        "                    video_files_paths.append(video_file_path)\n",
        "            # create validation data\n",
        "            elif state == 'valid':\n",
        "                # get the mouse number\n",
        "                mouse_number = int(video_file_path.split(' ')[2].split('_')[1].split('-')[1])\n",
        "                # get the mouse side (L or R)\n",
        "                mouse_side = video_file_path.split(' ')[2].split('_')[1].split('-')[2]\n",
        "                if mouse_number == 4 and mouse_side == 'R':\n",
        "                    frames = load_video(video_file_path)\n",
        "                    features.append(frames)\n",
        "                    labels.append(class_index)\n",
        "                    video_files_paths.append(video_file_path)\n",
        "            # create training data\n",
        "            else:\n",
        "                # get the mouse number\n",
        "                mouse_number = int(video_file_path.split(' ')[2].split('_')[1].split('-')[1])\n",
        "                if mouse_number != 4:\n",
        "                    frames = load_video(video_file_path)\n",
        "                    features.append(frames)\n",
        "                    labels.append(class_index)\n",
        "                    video_files_paths.append(video_file_path)\n",
        "    # Converting the list to numpy arrays\n",
        "    features = np.asarray(features)\n",
        "    # print(features)\n",
        "    labels = np.array(labels)\n",
        "    # Return the frames, class index, and video file path.\n",
        "    return features, labels, video_files_paths"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 8,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "a8rpanz9rASe",
        "outputId": "4f13739c-da7a-4d9b-9363-b3c2584007ea"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Extracting Data of Class: hemostasis train\n",
            "Extracting Data of Class: inflammatory train\n",
            "Extracting Data of Class: proliferative train\n",
            "Extracting Data of Class: maturation train\n",
            "Extracting Data of Class: hemostasis test\n",
            "Extracting Data of Class: inflammatory test\n",
            "Extracting Data of Class: proliferative test\n",
            "Extracting Data of Class: maturation test\n",
            "Extracting Data of Class: hemostasis valid\n",
            "Extracting Data of Class: inflammatory valid\n",
            "Extracting Data of Class: proliferative valid\n",
            "Extracting Data of Class: maturation valid\n"
          ]
        }
      ],
      "source": [
        "# 6 mice for training, 2 mice for test and validation (one wound on each mice for test one for validation)\n",
        "features_train, labels_train, video_files_paths_train = create_dataset('train')\n",
        "features_test, labels_test, video_files_paths_test = create_dataset('test')\n",
        "features_valid, labels_valid, video_files_paths_valid = create_dataset('valid')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 9,
      "metadata": {
        "id": "dtJkK4qTAulC"
      },
      "outputs": [],
      "source": [
        "# labels to catogorical\n",
        "labels_train = keras.utils.to_categorical(labels_train)\n",
        "labels_test = keras.utils.to_categorical(labels_test)\n",
        "labels_valid = keras.utils.to_categorical(labels_valid)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 23,
      "metadata": {
        "id": "Bi-NDol3DHRV"
      },
      "outputs": [],
      "source": [
        "def create_convlstm_model():\n",
        "    # TimeDistributed is a wrapper to handle input of size five to maintain frames number for LSTM\n",
        "    # A Conv2D layer requires four dimensions: (batch_size, height, width, channels)\n",
        "    # TimeDistributed will require an additional dimension: (batch_size, frames, height, width, channels)\n",
        "\n",
        "    model = Sequential()\n",
        "\n",
        "    model.add(TimeDistributed(Conv2D(8, (3,3), activation='relu',\n",
        "                                     kernel_regularizer=regularizers.L2(l2=1e-4)),\n",
        "                              input_shape=(16, 96, 96, 3)))\n",
        "    model.add(TimeDistributed(MaxPooling2D((2,2), strides=(2,2))))\n",
        "    model.add(Dropout(0.3))\n",
        "\n",
        "    model.add(TimeDistributed(Conv2D(16, (3,3), activation='relu',\n",
        "                                     kernel_regularizer=regularizers.L2(l2=1e-4))))\n",
        "    model.add(TimeDistributed(MaxPooling2D((2,2), strides=(2,2))))\n",
        "    model.add(Dropout(0.3))\n",
        "\n",
        "    model.add(TimeDistributed(Conv2D(32, (3,3), activation='relu',\n",
        "                                     kernel_regularizer=regularizers.L2(l2=1e-4))))\n",
        "    model.add(TimeDistributed(MaxPooling2D((2,2), strides=(2,2))))\n",
        "    model.add(Dropout(0.3))\n",
        "\n",
        "    model.add(TimeDistributed(Conv2D(16, (3,3), activation='relu',\n",
        "                                     kernel_regularizer=regularizers.L2(l2=1e-4))))\n",
        "    model.add(TimeDistributed(MaxPooling2D((2,2), strides=(2,2))))\n",
        "    model.add(Dropout(0.3))\n",
        "  \n",
        "    model.add(TimeDistributed(GlobalAveragePooling2D()))\n",
        "    model.add(Dropout(0.35))\n",
        "\n",
        "    model.add(LSTM(4, activation='relu', return_sequences=False))\n",
        "    model.add(Dropout(0.4))\n",
        "\n",
        "    model.add(Dense(16, activation='relu', kernel_initializer='he_uniform',\n",
        "                    kernel_regularizer=regularizers.L2(l2=1e-4)))\n",
        "    model.add(Dropout(0.4))\n",
        "\n",
        "    model.add(Dense(8, activation='relu', kernel_initializer='he_uniform',\n",
        "                    kernel_regularizer=regularizers.L2(l2=1e-4)))\n",
        "    model.add(Dropout(0.4))\n",
        "\n",
        "    model.add(Dense(4, activation='relu', kernel_initializer='he_uniform',\n",
        "                    kernel_regularizer=regularizers.L2(l2=1e-4)))\n",
        "\n",
        "    model.add(Dense(len(CLASSES_LIST), activation='softmax'))\n",
        "    model.summary(line_length = 100)\n",
        "    return model"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 24,
      "metadata": {
        "id": "FbjNYI-0DY_v",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "2004fb5d-b462-4600-af75-3c75cd6efa38"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Model: \"sequential_3\"\n",
            "____________________________________________________________________________________________________\n",
            " Layer (type)                                Output Shape                            Param #        \n",
            "====================================================================================================\n",
            " time_distributed_27 (TimeDistributed)       (None, 16, 94, 94, 8)                   224            \n",
            "                                                                                                    \n",
            " time_distributed_28 (TimeDistributed)       (None, 16, 47, 47, 8)                   0              \n",
            "                                                                                                    \n",
            " dropout_24 (Dropout)                        (None, 16, 47, 47, 8)                   0              \n",
            "                                                                                                    \n",
            " time_distributed_29 (TimeDistributed)       (None, 16, 45, 45, 16)                  1168           \n",
            "                                                                                                    \n",
            " time_distributed_30 (TimeDistributed)       (None, 16, 22, 22, 16)                  0              \n",
            "                                                                                                    \n",
            " dropout_25 (Dropout)                        (None, 16, 22, 22, 16)                  0              \n",
            "                                                                                                    \n",
            " time_distributed_31 (TimeDistributed)       (None, 16, 20, 20, 32)                  4640           \n",
            "                                                                                                    \n",
            " time_distributed_32 (TimeDistributed)       (None, 16, 10, 10, 32)                  0              \n",
            "                                                                                                    \n",
            " dropout_26 (Dropout)                        (None, 16, 10, 10, 32)                  0              \n",
            "                                                                                                    \n",
            " time_distributed_33 (TimeDistributed)       (None, 16, 8, 8, 16)                    4624           \n",
            "                                                                                                    \n",
            " time_distributed_34 (TimeDistributed)       (None, 16, 4, 4, 16)                    0              \n",
            "                                                                                                    \n",
            " dropout_27 (Dropout)                        (None, 16, 4, 4, 16)                    0              \n",
            "                                                                                                    \n",
            " time_distributed_35 (TimeDistributed)       (None, 16, 16)                          0              \n",
            "                                                                                                    \n",
            " dropout_28 (Dropout)                        (None, 16, 16)                          0              \n",
            "                                                                                                    \n",
            " lstm_3 (LSTM)                               (None, 4)                               336            \n",
            "                                                                                                    \n",
            " dropout_29 (Dropout)                        (None, 4)                               0              \n",
            "                                                                                                    \n",
            " dense_12 (Dense)                            (None, 16)                              80             \n",
            "                                                                                                    \n",
            " dropout_30 (Dropout)                        (None, 16)                              0              \n",
            "                                                                                                    \n",
            " dense_13 (Dense)                            (None, 8)                               136            \n",
            "                                                                                                    \n",
            " dropout_31 (Dropout)                        (None, 8)                               0              \n",
            "                                                                                                    \n",
            " dense_14 (Dense)                            (None, 4)                               36             \n",
            "                                                                                                    \n",
            " dense_15 (Dense)                            (None, 4)                               20             \n",
            "                                                                                                    \n",
            "====================================================================================================\n",
            "Total params: 11,264\n",
            "Trainable params: 11,264\n",
            "Non-trainable params: 0\n",
            "____________________________________________________________________________________________________\n",
            "Model Created Successfully!\n"
          ]
        }
      ],
      "source": [
        "# Construct the required convlstm model.\n",
        "convlstm_model = create_convlstm_model()\n",
        " \n",
        "# Display the success message. \n",
        "print(\"Model Created Successfully!\")"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 12,
      "metadata": {
        "id": "nMYwQOwyF3bd"
      },
      "outputs": [],
      "source": [
        "# keras.utils.plot_model(convlstm_model,\n",
        "#                          to_file = 'convlstm_model_structure_plot.png',\n",
        "#                          show_shapes = True,\n",
        "#                          show_layer_names = True)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 25,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "vYC_6H0uGqW9",
        "outputId": "9d5f1c63-c396-4f95-e62b-9e42adafa362"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Epoch 1/200\n",
            "26/26 [==============================] - 7s 147ms/step - loss: 1.3816 - accuracy: 0.3844 - val_loss: 1.3676 - val_accuracy: 0.3971\n",
            "Epoch 2/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.3419 - accuracy: 0.4508 - val_loss: 1.3202 - val_accuracy: 0.4559\n",
            "Epoch 3/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.2913 - accuracy: 0.4373 - val_loss: 1.2816 - val_accuracy: 0.3971\n",
            "Epoch 4/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 1.2336 - accuracy: 0.4723 - val_loss: 1.1666 - val_accuracy: 0.4890\n",
            "Epoch 5/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 1.2063 - accuracy: 0.4490 - val_loss: 1.1691 - val_accuracy: 0.3971\n",
            "Epoch 6/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 1.1752 - accuracy: 0.4809 - val_loss: 1.1531 - val_accuracy: 0.4706\n",
            "Epoch 7/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 1.1597 - accuracy: 0.4606 - val_loss: 1.1301 - val_accuracy: 0.4706\n",
            "Epoch 8/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.1323 - accuracy: 0.4828 - val_loss: 1.0870 - val_accuracy: 0.4706\n",
            "Epoch 9/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.1261 - accuracy: 0.5000 - val_loss: 1.0869 - val_accuracy: 0.4779\n",
            "Epoch 10/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0985 - accuracy: 0.5031 - val_loss: 1.0360 - val_accuracy: 0.5000\n",
            "Epoch 11/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0925 - accuracy: 0.4975 - val_loss: 1.0072 - val_accuracy: 0.5000\n",
            "Epoch 12/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0676 - accuracy: 0.5141 - val_loss: 1.0269 - val_accuracy: 0.4779\n",
            "Epoch 13/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.0647 - accuracy: 0.4963 - val_loss: 1.0160 - val_accuracy: 0.4669\n",
            "Epoch 14/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.0413 - accuracy: 0.5080 - val_loss: 1.0080 - val_accuracy: 0.4890\n",
            "Epoch 15/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.0532 - accuracy: 0.5160 - val_loss: 0.9797 - val_accuracy: 0.5000\n",
            "Epoch 16/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0570 - accuracy: 0.4902 - val_loss: 0.9626 - val_accuracy: 0.5000\n",
            "Epoch 17/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0352 - accuracy: 0.4926 - val_loss: 0.9779 - val_accuracy: 0.4890\n",
            "Epoch 18/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 1.0344 - accuracy: 0.4859 - val_loss: 0.9649 - val_accuracy: 0.4706\n",
            "Epoch 19/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0195 - accuracy: 0.5025 - val_loss: 0.9962 - val_accuracy: 0.4412\n",
            "Epoch 20/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 1.0314 - accuracy: 0.5031 - val_loss: 0.9920 - val_accuracy: 0.4779\n",
            "Epoch 21/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9854 - accuracy: 0.5375 - val_loss: 0.9395 - val_accuracy: 0.5000\n",
            "Epoch 22/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.9992 - accuracy: 0.5320 - val_loss: 0.9738 - val_accuracy: 0.4596\n",
            "Epoch 23/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9878 - accuracy: 0.5178 - val_loss: 0.9875 - val_accuracy: 0.4706\n",
            "Epoch 24/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9854 - accuracy: 0.5461 - val_loss: 0.9739 - val_accuracy: 0.4926\n",
            "Epoch 25/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 0.9463 - accuracy: 0.5713 - val_loss: 0.9060 - val_accuracy: 0.5956\n",
            "Epoch 26/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9169 - accuracy: 0.5916 - val_loss: 1.0073 - val_accuracy: 0.4081\n",
            "Epoch 27/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9044 - accuracy: 0.6113 - val_loss: 0.8952 - val_accuracy: 0.6103\n",
            "Epoch 28/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.9342 - accuracy: 0.5929 - val_loss: 0.8495 - val_accuracy: 0.6397\n",
            "Epoch 29/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.8916 - accuracy: 0.6150 - val_loss: 0.9265 - val_accuracy: 0.5846\n",
            "Epoch 30/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8762 - accuracy: 0.6224 - val_loss: 0.8905 - val_accuracy: 0.6324\n",
            "Epoch 31/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.8656 - accuracy: 0.6279 - val_loss: 0.8444 - val_accuracy: 0.6434\n",
            "Epoch 32/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8801 - accuracy: 0.6175 - val_loss: 0.8169 - val_accuracy: 0.6544\n",
            "Epoch 33/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8605 - accuracy: 0.6285 - val_loss: 0.8295 - val_accuracy: 0.6434\n",
            "Epoch 34/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8575 - accuracy: 0.6248 - val_loss: 0.8296 - val_accuracy: 0.6544\n",
            "Epoch 35/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8879 - accuracy: 0.6125 - val_loss: 0.8208 - val_accuracy: 0.6434\n",
            "Epoch 36/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 0.8610 - accuracy: 0.6255 - val_loss: 0.8112 - val_accuracy: 0.6507\n",
            "Epoch 37/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.8690 - accuracy: 0.6205 - val_loss: 0.8808 - val_accuracy: 0.6213\n",
            "Epoch 38/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.8387 - accuracy: 0.6433 - val_loss: 0.7944 - val_accuracy: 0.6544\n",
            "Epoch 39/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8389 - accuracy: 0.6433 - val_loss: 0.8212 - val_accuracy: 0.6397\n",
            "Epoch 40/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8625 - accuracy: 0.6205 - val_loss: 0.8090 - val_accuracy: 0.6544\n",
            "Epoch 41/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.8374 - accuracy: 0.6316 - val_loss: 0.8603 - val_accuracy: 0.6287\n",
            "Epoch 42/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.8158 - accuracy: 0.6482 - val_loss: 0.7834 - val_accuracy: 0.6544\n",
            "Epoch 43/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.8260 - accuracy: 0.6378 - val_loss: 0.9409 - val_accuracy: 0.5956\n",
            "Epoch 44/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8299 - accuracy: 0.6378 - val_loss: 0.8347 - val_accuracy: 0.6324\n",
            "Epoch 45/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 0.8249 - accuracy: 0.6328 - val_loss: 0.7826 - val_accuracy: 0.6544\n",
            "Epoch 46/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8292 - accuracy: 0.6488 - val_loss: 0.8472 - val_accuracy: 0.6287\n",
            "Epoch 47/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.8338 - accuracy: 0.6267 - val_loss: 0.7616 - val_accuracy: 0.6544\n",
            "Epoch 48/200\n",
            "26/26 [==============================] - 3s 115ms/step - loss: 0.8328 - accuracy: 0.6341 - val_loss: 0.7389 - val_accuracy: 0.6838\n",
            "Epoch 49/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8354 - accuracy: 0.6279 - val_loss: 0.7440 - val_accuracy: 0.6544\n",
            "Epoch 50/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7948 - accuracy: 0.6464 - val_loss: 0.7327 - val_accuracy: 0.6801\n",
            "Epoch 51/200\n",
            "26/26 [==============================] - 3s 115ms/step - loss: 0.8029 - accuracy: 0.6501 - val_loss: 0.7405 - val_accuracy: 0.6544\n",
            "Epoch 52/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7887 - accuracy: 0.6550 - val_loss: 0.8003 - val_accuracy: 0.6360\n",
            "Epoch 53/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7960 - accuracy: 0.6415 - val_loss: 0.8116 - val_accuracy: 0.6397\n",
            "Epoch 54/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.8110 - accuracy: 0.6507 - val_loss: 0.7487 - val_accuracy: 0.6838\n",
            "Epoch 55/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7833 - accuracy: 0.6519 - val_loss: 0.7145 - val_accuracy: 0.6838\n",
            "Epoch 56/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7981 - accuracy: 0.6624 - val_loss: 0.6808 - val_accuracy: 0.6949\n",
            "Epoch 57/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7755 - accuracy: 0.6574 - val_loss: 0.7506 - val_accuracy: 0.6765\n",
            "Epoch 58/200\n",
            "26/26 [==============================] - 3s 116ms/step - loss: 0.7878 - accuracy: 0.6630 - val_loss: 0.7860 - val_accuracy: 0.6397\n",
            "Epoch 59/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7868 - accuracy: 0.6544 - val_loss: 0.7165 - val_accuracy: 0.6691\n",
            "Epoch 60/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 0.7656 - accuracy: 0.6611 - val_loss: 0.6753 - val_accuracy: 0.6985\n",
            "Epoch 61/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.7586 - accuracy: 0.6587 - val_loss: 0.6922 - val_accuracy: 0.6949\n",
            "Epoch 62/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.7978 - accuracy: 0.6519 - val_loss: 0.7217 - val_accuracy: 0.6838\n",
            "Epoch 63/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7626 - accuracy: 0.6599 - val_loss: 0.7312 - val_accuracy: 0.6544\n",
            "Epoch 64/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7756 - accuracy: 0.6562 - val_loss: 0.7378 - val_accuracy: 0.6544\n",
            "Epoch 65/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7656 - accuracy: 0.6562 - val_loss: 0.8001 - val_accuracy: 0.6434\n",
            "Epoch 66/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7524 - accuracy: 0.6581 - val_loss: 0.7404 - val_accuracy: 0.6434\n",
            "Epoch 67/200\n",
            "26/26 [==============================] - 3s 114ms/step - loss: 0.7562 - accuracy: 0.6513 - val_loss: 0.7474 - val_accuracy: 0.6949\n",
            "Epoch 68/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7547 - accuracy: 0.6648 - val_loss: 0.7475 - val_accuracy: 0.6544\n",
            "Epoch 69/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7459 - accuracy: 0.6673 - val_loss: 0.6966 - val_accuracy: 0.6949\n",
            "Epoch 70/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7468 - accuracy: 0.6611 - val_loss: 0.7006 - val_accuracy: 0.6765\n",
            "Epoch 71/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7475 - accuracy: 0.6630 - val_loss: 0.7190 - val_accuracy: 0.6544\n",
            "Epoch 72/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7586 - accuracy: 0.6476 - val_loss: 0.7323 - val_accuracy: 0.6544\n",
            "Epoch 73/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7299 - accuracy: 0.6667 - val_loss: 0.6949 - val_accuracy: 0.6949\n",
            "Epoch 74/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.7813 - accuracy: 0.6494 - val_loss: 0.7667 - val_accuracy: 0.6471\n",
            "Epoch 75/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7347 - accuracy: 0.6691 - val_loss: 0.6840 - val_accuracy: 0.6949\n",
            "Epoch 76/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7409 - accuracy: 0.6679 - val_loss: 0.6687 - val_accuracy: 0.6949\n",
            "Epoch 77/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.7264 - accuracy: 0.6704 - val_loss: 0.7317 - val_accuracy: 0.6544\n",
            "Epoch 78/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7196 - accuracy: 0.6753 - val_loss: 0.7220 - val_accuracy: 0.6544\n",
            "Epoch 79/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7151 - accuracy: 0.6796 - val_loss: 0.7153 - val_accuracy: 0.6728\n",
            "Epoch 80/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7120 - accuracy: 0.6716 - val_loss: 0.6559 - val_accuracy: 0.6985\n",
            "Epoch 81/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7152 - accuracy: 0.6765 - val_loss: 0.6512 - val_accuracy: 0.6985\n",
            "Epoch 82/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7082 - accuracy: 0.6784 - val_loss: 0.6787 - val_accuracy: 0.6985\n",
            "Epoch 83/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.6982 - accuracy: 0.6833 - val_loss: 0.6452 - val_accuracy: 0.6985\n",
            "Epoch 84/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.7104 - accuracy: 0.6759 - val_loss: 0.7141 - val_accuracy: 0.6838\n",
            "Epoch 85/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7176 - accuracy: 0.6765 - val_loss: 0.8329 - val_accuracy: 0.5478\n",
            "Epoch 86/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7206 - accuracy: 0.6747 - val_loss: 0.6393 - val_accuracy: 0.6985\n",
            "Epoch 87/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7190 - accuracy: 0.6771 - val_loss: 0.6897 - val_accuracy: 0.6985\n",
            "Epoch 88/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.7043 - accuracy: 0.6753 - val_loss: 0.6694 - val_accuracy: 0.6985\n",
            "Epoch 89/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.6972 - accuracy: 0.6857 - val_loss: 0.6711 - val_accuracy: 0.6949\n",
            "Epoch 90/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.7061 - accuracy: 0.6820 - val_loss: 0.7233 - val_accuracy: 0.6691\n",
            "Epoch 91/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.7048 - accuracy: 0.6716 - val_loss: 0.6799 - val_accuracy: 0.6949\n",
            "Epoch 92/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.6789 - accuracy: 0.6863 - val_loss: 0.6448 - val_accuracy: 0.6985\n",
            "Epoch 93/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.6757 - accuracy: 0.6919 - val_loss: 0.6945 - val_accuracy: 0.6875\n",
            "Epoch 94/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 11.1586 - accuracy: 0.6790 - val_loss: 1.4915 - val_accuracy: 0.3971\n",
            "Epoch 95/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.2740 - accuracy: 0.4496 - val_loss: 1.2947 - val_accuracy: 0.3971\n",
            "Epoch 96/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.1425 - accuracy: 0.4502 - val_loss: 1.1042 - val_accuracy: 0.3971\n",
            "Epoch 97/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.1027 - accuracy: 0.4582 - val_loss: 1.0697 - val_accuracy: 0.3971\n",
            "Epoch 98/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0862 - accuracy: 0.4668 - val_loss: 1.0741 - val_accuracy: 0.3971\n",
            "Epoch 99/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0900 - accuracy: 0.4477 - val_loss: 1.0219 - val_accuracy: 0.3971\n",
            "Epoch 100/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.0596 - accuracy: 0.4625 - val_loss: 1.0060 - val_accuracy: 0.3971\n",
            "Epoch 101/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0639 - accuracy: 0.4643 - val_loss: 0.9937 - val_accuracy: 0.4154\n",
            "Epoch 102/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0405 - accuracy: 0.4588 - val_loss: 0.9681 - val_accuracy: 0.4522\n",
            "Epoch 103/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0494 - accuracy: 0.4643 - val_loss: 0.9660 - val_accuracy: 0.4375\n",
            "Epoch 104/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0326 - accuracy: 0.4619 - val_loss: 0.9570 - val_accuracy: 0.4779\n",
            "Epoch 105/200\n",
            "26/26 [==============================] - 3s 108ms/step - loss: 1.0389 - accuracy: 0.4785 - val_loss: 0.9517 - val_accuracy: 0.4522\n",
            "Epoch 106/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0344 - accuracy: 0.4828 - val_loss: 0.9350 - val_accuracy: 0.4559\n",
            "Epoch 107/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0293 - accuracy: 0.4822 - val_loss: 0.9237 - val_accuracy: 0.4816\n",
            "Epoch 108/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0139 - accuracy: 0.4994 - val_loss: 0.9092 - val_accuracy: 0.4743\n",
            "Epoch 109/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0100 - accuracy: 0.5025 - val_loss: 0.9094 - val_accuracy: 0.4963\n",
            "Epoch 110/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0173 - accuracy: 0.4920 - val_loss: 0.8992 - val_accuracy: 0.4779\n",
            "Epoch 111/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0043 - accuracy: 0.5031 - val_loss: 0.8914 - val_accuracy: 0.4669\n",
            "Epoch 112/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0019 - accuracy: 0.5135 - val_loss: 0.8804 - val_accuracy: 0.4632\n",
            "Epoch 113/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 1.0147 - accuracy: 0.5105 - val_loss: 0.8783 - val_accuracy: 0.4816\n",
            "Epoch 114/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9942 - accuracy: 0.5185 - val_loss: 0.8706 - val_accuracy: 0.5110\n",
            "Epoch 115/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0066 - accuracy: 0.5111 - val_loss: 0.8704 - val_accuracy: 0.5147\n",
            "Epoch 116/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9944 - accuracy: 0.5191 - val_loss: 0.8679 - val_accuracy: 0.5221\n",
            "Epoch 117/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0094 - accuracy: 0.5141 - val_loss: 0.8689 - val_accuracy: 0.5147\n",
            "Epoch 118/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9953 - accuracy: 0.5166 - val_loss: 0.8624 - val_accuracy: 0.5588\n",
            "Epoch 119/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9939 - accuracy: 0.5308 - val_loss: 0.8556 - val_accuracy: 0.5662\n",
            "Epoch 120/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 1.0220 - accuracy: 0.5062 - val_loss: 0.8578 - val_accuracy: 0.5588\n",
            "Epoch 121/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 1.0016 - accuracy: 0.5185 - val_loss: 0.8634 - val_accuracy: 0.5699\n",
            "Epoch 122/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9866 - accuracy: 0.5351 - val_loss: 0.8545 - val_accuracy: 0.5551\n",
            "Epoch 123/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9930 - accuracy: 0.5363 - val_loss: 0.8511 - val_accuracy: 0.5551\n",
            "Epoch 124/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 1.0041 - accuracy: 0.5234 - val_loss: 0.8513 - val_accuracy: 0.5551\n",
            "Epoch 125/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9938 - accuracy: 0.5363 - val_loss: 0.8477 - val_accuracy: 0.5588\n",
            "Epoch 126/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9844 - accuracy: 0.5492 - val_loss: 0.8569 - val_accuracy: 0.5809\n",
            "Epoch 127/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9777 - accuracy: 0.5363 - val_loss: 0.8527 - val_accuracy: 0.5809\n",
            "Epoch 128/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9902 - accuracy: 0.5400 - val_loss: 0.8489 - val_accuracy: 0.5735\n",
            "Epoch 129/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0040 - accuracy: 0.5264 - val_loss: 0.8557 - val_accuracy: 0.5809\n",
            "Epoch 130/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9776 - accuracy: 0.5375 - val_loss: 0.8547 - val_accuracy: 0.5956\n",
            "Epoch 131/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9908 - accuracy: 0.5547 - val_loss: 0.8563 - val_accuracy: 0.5846\n",
            "Epoch 132/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9764 - accuracy: 0.5437 - val_loss: 0.8677 - val_accuracy: 0.5993\n",
            "Epoch 133/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9847 - accuracy: 0.5431 - val_loss: 0.8505 - val_accuracy: 0.5846\n",
            "Epoch 134/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9784 - accuracy: 0.5498 - val_loss: 0.8699 - val_accuracy: 0.5993\n",
            "Epoch 135/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 1.0046 - accuracy: 0.5474 - val_loss: 0.8759 - val_accuracy: 0.5993\n",
            "Epoch 136/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9831 - accuracy: 0.5467 - val_loss: 0.8690 - val_accuracy: 0.5956\n",
            "Epoch 137/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9900 - accuracy: 0.5541 - val_loss: 0.8664 - val_accuracy: 0.5993\n",
            "Epoch 138/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9944 - accuracy: 0.5510 - val_loss: 0.8813 - val_accuracy: 0.5993\n",
            "Epoch 139/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9806 - accuracy: 0.5449 - val_loss: 0.8760 - val_accuracy: 0.5993\n",
            "Epoch 140/200\n",
            "26/26 [==============================] - 3s 108ms/step - loss: 0.9723 - accuracy: 0.5431 - val_loss: 0.8945 - val_accuracy: 0.5993\n",
            "Epoch 141/200\n",
            "26/26 [==============================] - 3s 108ms/step - loss: 0.9603 - accuracy: 0.5578 - val_loss: 0.8860 - val_accuracy: 0.5993\n",
            "Epoch 142/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9643 - accuracy: 0.5523 - val_loss: 0.9060 - val_accuracy: 0.5993\n",
            "Epoch 143/200\n",
            "26/26 [==============================] - 3s 108ms/step - loss: 0.9771 - accuracy: 0.5554 - val_loss: 0.8867 - val_accuracy: 0.5919\n",
            "Epoch 144/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9636 - accuracy: 0.5578 - val_loss: 0.9188 - val_accuracy: 0.5882\n",
            "Epoch 145/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9678 - accuracy: 0.5578 - val_loss: 0.9072 - val_accuracy: 0.5772\n",
            "Epoch 146/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9831 - accuracy: 0.5517 - val_loss: 0.8821 - val_accuracy: 0.6029\n",
            "Epoch 147/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9850 - accuracy: 0.5590 - val_loss: 0.8847 - val_accuracy: 0.5993\n",
            "Epoch 148/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 1.0002 - accuracy: 0.5394 - val_loss: 0.8813 - val_accuracy: 0.5993\n",
            "Epoch 149/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9645 - accuracy: 0.5670 - val_loss: 0.8953 - val_accuracy: 0.5919\n",
            "Epoch 150/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9679 - accuracy: 0.5707 - val_loss: 0.9141 - val_accuracy: 0.5772\n",
            "Epoch 151/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9736 - accuracy: 0.5584 - val_loss: 0.9155 - val_accuracy: 0.5735\n",
            "Epoch 152/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9570 - accuracy: 0.5720 - val_loss: 0.9404 - val_accuracy: 0.5919\n",
            "Epoch 153/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9573 - accuracy: 0.5664 - val_loss: 0.9454 - val_accuracy: 0.5919\n",
            "Epoch 154/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9676 - accuracy: 0.5535 - val_loss: 0.9276 - val_accuracy: 0.5735\n",
            "Epoch 155/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9680 - accuracy: 0.5646 - val_loss: 0.9437 - val_accuracy: 0.5956\n",
            "Epoch 156/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9639 - accuracy: 0.5541 - val_loss: 0.9092 - val_accuracy: 0.5772\n",
            "Epoch 157/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9709 - accuracy: 0.5615 - val_loss: 0.9361 - val_accuracy: 0.5735\n",
            "Epoch 158/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9591 - accuracy: 0.5830 - val_loss: 0.9676 - val_accuracy: 0.5993\n",
            "Epoch 159/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9723 - accuracy: 0.5652 - val_loss: 0.9352 - val_accuracy: 0.5735\n",
            "Epoch 160/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9737 - accuracy: 0.5584 - val_loss: 0.9742 - val_accuracy: 0.5993\n",
            "Epoch 161/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9829 - accuracy: 0.5633 - val_loss: 0.9325 - val_accuracy: 0.5772\n",
            "Epoch 162/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9765 - accuracy: 0.5615 - val_loss: 0.9636 - val_accuracy: 0.5956\n",
            "Epoch 163/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9632 - accuracy: 0.5738 - val_loss: 0.9889 - val_accuracy: 0.5956\n",
            "Epoch 164/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9871 - accuracy: 0.5535 - val_loss: 0.9817 - val_accuracy: 0.5956\n",
            "Epoch 165/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9769 - accuracy: 0.5744 - val_loss: 0.9894 - val_accuracy: 0.5956\n",
            "Epoch 166/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9707 - accuracy: 0.5781 - val_loss: 0.9811 - val_accuracy: 0.5956\n",
            "Epoch 167/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.9693 - accuracy: 0.5584 - val_loss: 1.0169 - val_accuracy: 0.5956\n",
            "Epoch 168/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9838 - accuracy: 0.5701 - val_loss: 0.9923 - val_accuracy: 0.5993\n",
            "Epoch 169/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9753 - accuracy: 0.5560 - val_loss: 1.0022 - val_accuracy: 0.5956\n",
            "Epoch 170/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9744 - accuracy: 0.5633 - val_loss: 0.9616 - val_accuracy: 0.5919\n",
            "Epoch 171/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9698 - accuracy: 0.5652 - val_loss: 0.9931 - val_accuracy: 0.5993\n",
            "Epoch 172/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9813 - accuracy: 0.5566 - val_loss: 0.9570 - val_accuracy: 0.5809\n",
            "Epoch 173/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9506 - accuracy: 0.5695 - val_loss: 0.9912 - val_accuracy: 0.5993\n",
            "Epoch 174/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9726 - accuracy: 0.5701 - val_loss: 0.9836 - val_accuracy: 0.5993\n",
            "Epoch 175/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9565 - accuracy: 0.5633 - val_loss: 0.9343 - val_accuracy: 0.5809\n",
            "Epoch 176/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9564 - accuracy: 0.5670 - val_loss: 1.0017 - val_accuracy: 0.5993\n",
            "Epoch 177/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9560 - accuracy: 0.5695 - val_loss: 0.9823 - val_accuracy: 0.5882\n",
            "Epoch 178/200\n",
            "26/26 [==============================] - 3s 108ms/step - loss: 0.9717 - accuracy: 0.5578 - val_loss: 0.9650 - val_accuracy: 0.5772\n",
            "Epoch 179/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9715 - accuracy: 0.5597 - val_loss: 0.9512 - val_accuracy: 0.5699\n",
            "Epoch 180/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9565 - accuracy: 0.5652 - val_loss: 1.0305 - val_accuracy: 0.5956\n",
            "Epoch 181/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9567 - accuracy: 0.5677 - val_loss: 0.9881 - val_accuracy: 0.5956\n",
            "Epoch 182/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9537 - accuracy: 0.5756 - val_loss: 1.0244 - val_accuracy: 0.5956\n",
            "Epoch 183/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9542 - accuracy: 0.5670 - val_loss: 0.9880 - val_accuracy: 0.5993\n",
            "Epoch 184/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9473 - accuracy: 0.5726 - val_loss: 1.0487 - val_accuracy: 0.5956\n",
            "Epoch 185/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9668 - accuracy: 0.5621 - val_loss: 1.0375 - val_accuracy: 0.6029\n",
            "Epoch 186/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9593 - accuracy: 0.5713 - val_loss: 1.0760 - val_accuracy: 0.5993\n",
            "Epoch 187/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9572 - accuracy: 0.5658 - val_loss: 1.0500 - val_accuracy: 0.5993\n",
            "Epoch 188/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9558 - accuracy: 0.5670 - val_loss: 1.0295 - val_accuracy: 0.5956\n",
            "Epoch 189/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9608 - accuracy: 0.5597 - val_loss: 1.0153 - val_accuracy: 0.5993\n",
            "Epoch 190/200\n",
            "26/26 [==============================] - 3s 109ms/step - loss: 0.9524 - accuracy: 0.5621 - val_loss: 1.0415 - val_accuracy: 0.5993\n",
            "Epoch 191/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9766 - accuracy: 0.5474 - val_loss: 1.0499 - val_accuracy: 0.5993\n",
            "Epoch 192/200\n",
            "26/26 [==============================] - 3s 113ms/step - loss: 0.9471 - accuracy: 0.5720 - val_loss: 1.0349 - val_accuracy: 0.5993\n",
            "Epoch 193/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9757 - accuracy: 0.5541 - val_loss: 0.9990 - val_accuracy: 0.5919\n",
            "Epoch 194/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9557 - accuracy: 0.5627 - val_loss: 1.0146 - val_accuracy: 0.5993\n",
            "Epoch 195/200\n",
            "26/26 [==============================] - 3s 112ms/step - loss: 0.9674 - accuracy: 0.5627 - val_loss: 1.0313 - val_accuracy: 0.5993\n",
            "Epoch 196/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9605 - accuracy: 0.5720 - val_loss: 1.0491 - val_accuracy: 0.5956\n",
            "Epoch 197/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9443 - accuracy: 0.5824 - val_loss: 1.1059 - val_accuracy: 0.5882\n",
            "Epoch 198/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9580 - accuracy: 0.5744 - val_loss: 1.0853 - val_accuracy: 0.5846\n",
            "Epoch 199/200\n",
            "26/26 [==============================] - 3s 111ms/step - loss: 0.9541 - accuracy: 0.5590 - val_loss: 1.0949 - val_accuracy: 0.5919\n",
            "Epoch 200/200\n",
            "26/26 [==============================] - 3s 110ms/step - loss: 0.9473 - accuracy: 0.5652 - val_loss: 1.0904 - val_accuracy: 0.5993\n"
          ]
        }
      ],
      "source": [
        "# Compile the model and specify loss function, optimizer and metrics values to the model\n",
        "convlstm_model.compile(loss = 'categorical_crossentropy',\n",
        "                       optimizer= keras.optimizers.Adam(0.001, decay=1e-4),\n",
        "                       metrics = [\"accuracy\"])\n",
        "# Start training the model.\n",
        "convlstm_model_training_history = convlstm_model.fit(x = features_train,\n",
        "                                                     y = labels_train,\n",
        "                                                     epochs = 200,\n",
        "                                                     batch_size = 64,\n",
        "                                                     shuffle = True,\n",
        "                                                     validation_data = (features_valid, labels_valid))\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 14,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LcpKhGqvIenv",
        "outputId": "08fb3bfe-288b-4b3c-8f25-50b310741a59"
      },
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "9/9 [==============================] - 0s 21ms/step - loss: 1.2506 - accuracy: 0.6765\n",
            "\n",
            "\n",
            "Train accuracy: 77.491 % || Test accuracy: 67.647 % || Val accuracy: 72.794 %\n",
            "\n",
            "\n",
            "Train loss: 0.534 || Test loss: 1.251 || Val loss: 0.573\n"
          ]
        }
      ],
      "source": [
        "model_evaluation_history = convlstm_model.evaluate(features_test, labels_test)\n",
        "print('\\n')\n",
        "train_loss, train_acc = convlstm_model.evaluate(features_train, labels_train, verbose=0)\n",
        "test_loss, test_acc = convlstm_model.evaluate(features_test, labels_test, verbose=0)\n",
        "val_loss, val_acc = convlstm_model.evaluate(features_valid, labels_valid, verbose=0)\n",
        "\n",
        "print(f'Train accuracy: {train_acc*100:.3f} % || Test accuracy: {test_acc*100:.3f} % || Val accuracy: {val_acc*100:.3f} %')\n",
        "print('\\n')\n",
        "print(f'Train loss: {train_loss:.3f} || Test loss: {test_loss:.3f} || Val loss: {val_loss:.3f}')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 15,
      "metadata": {
        "id": "qsFw5KIzIum9"
      },
      "outputs": [],
      "source": [
        "# Get the loss and accuracy from model_evaluation_history.\n",
        "model_evaluation_loss, model_evaluation_accuracy = model_evaluation_history\n",
        " \n",
        "# Define the string date format.\n",
        "# Get the current Date and Time in a DateTime Object.\n",
        "# Convert the DateTime object to string according to the style mentioned in date_time_format string.\n",
        "date_time_format = '%Y_%m_%d__%H_%M_%S'\n",
        "current_date_time_dt = dt.datetime.now()\n",
        "current_date_time_string = dt.datetime.strftime(current_date_time_dt, date_time_format)\n",
        " \n",
        "# Define a useful name for our model to make it easy for us while navigating through multiple saved models.\n",
        "model_file_name = f'convlstm_model___Date_Time_{current_date_time_string}___Loss_{model_evaluation_loss}___Accuracy_{model_evaluation_accuracy}.h5'\n",
        " \n",
        "# Change dir\n",
        "gdrive_path = '/content/gdrive' + '/My Drive/247/Saved_models/'\n",
        "os.chdir(gdrive_path)\n",
        "# Create a floder for the model files\n",
        "!mkdir -p convlstm_{current_date_time_string}\n",
        "# Save your Model.\n",
        "convlstm_model.save('convlstm_' + str(current_date_time_string) + '/' + model_file_name)\n",
        "# Save model weights\n",
        "convlstm_model.save_weights('convlstm_' + str(current_date_time_string) + '/' + 'weights')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 16,
      "metadata": {
        "id": "Z0eAobhSIz_P"
      },
      "outputs": [],
      "source": [
        "def plot_metric(model_training_history, metric_name_1, metric_name_2, plot_name):\n",
        "    '''\n",
        "    This function will plot the metrics passed to it in a graph.\n",
        "    Args:\n",
        "        model_training_history: A history object containing a record of training and validation \n",
        "                                loss values and metrics values at successive epochs\n",
        "        metric_name_1:          The name of the first metric that needs to be plotted in the graph.\n",
        "        metric_name_2:          The name of the second metric that needs to be plotted in the graph.\n",
        "        plot_name:              The title of the graph.\n",
        "    '''\n",
        "    \n",
        "    # Get metric values using metric names as identifiers.\n",
        "    metric_value_1 = model_training_history.history[metric_name_1]\n",
        "    metric_value_2 = model_training_history.history[metric_name_2]\n",
        "    \n",
        "    # Construct a range object which will be used as x-axis (horizontal plane) of the graph.\n",
        "    epochs = range(len(metric_value_1))\n",
        "\n",
        "    # Plot the Graph.\n",
        "    plt.plot(epochs, metric_value_1, 'blue', label = metric_name_1)\n",
        "    plt.plot(epochs, metric_value_2, 'red', label = metric_name_2)\n",
        "\n",
        "    # Add title to the plot.\n",
        "    plt.title(str(plot_name))\n",
        "\n",
        "    # Add legend to the plot.\n",
        "    plt.legend()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 17,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "18RtBjI9I5lE",
        "outputId": "71f8e454-3f7b-4cff-9759-b8d2958e3665"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plot_metric(convlstm_model_training_history, 'loss', 'val_loss', 'Total Loss vs Total Validation Loss')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": 18,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 281
        },
        "id": "02amk7TXI_OS",
        "outputId": "6e09d5e1-eee2-4fc2-a941-02dbd018d954"
      },
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ],
            "image/png": "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\n"
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ],
      "source": [
        "plot_metric(convlstm_model_training_history, 'accuracy', 'val_accuracy', 'Total Accuracy vs Total Validation Accuracy') "
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "machine_shape": "hm",
      "name": "2D_CNN+LSTM_No_Splint.ipynb",
      "provenance": [],
      "authorship_tag": "ABX9TyNCE5Jchefo+MTsJ+4QGBZD",
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}