neural_network.ipynb

{
  "cells": [
    {
      "attachments": {},
      "cell_type": "markdown",
      "metadata": {
        "colab_type": "text",
        "id": "view-in-github"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/aaanthonyyy/3020-project/blob/master/notebooks/neural_network.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 35
        },
        "id": "Y31ihkv9OXcE",
        "outputId": "24a75a36-afc9-4033-bd00-0fdd774c9b26"
      },
      "outputs": [
        {
          "data": {
            "application/vnd.google.colaboratory.intrinsic+json": {
              "type": "string"
            },
            "text/plain": [
              "'2.3.0'"
            ]
          },
          "execution_count": 4,
          "metadata": {
            "tags": []
          },
          "output_type": "execute_result"
        }
      ],
      "source": [
        "import tensorflow as tf\n",
        "from tensorflow import keras\n",
        "import numpy as np\n",
        "import matplotlib.pyplot as plt\n",
        "\n",
        "tf.__version__"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# Data Parameters\n",
        "batch_size = 512\n",
        "\n",
        "# Image Parameters\n",
        "img_height, img_width = 128, 128"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "import pathlib\n",
        "\n",
        "data_dir = pathlib.Path('/content/drive/My Drive/3020_drive/Colab Notebooks/Data/Dataset/Devset_12')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 9556 files belonging to 5 classes.\n",
            "Using 7645 files for training.\n"
          ]
        }
      ],
      "source": [
        "# Loading training dataset from Drive\n",
        "train_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
        "  data_dir,\n",
        "  validation_split=0.2,\n",
        "  subset=\"training\",\n",
        "  seed=123,\n",
        "  color_mode='grayscale',\n",
        "  image_size=(img_height, img_width),\n",
        "  batch_size=batch_size)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Found 9556 files belonging to 5 classes.\n",
            "Using 1911 files for validation.\n"
          ]
        }
      ],
      "source": [
        "# Loading validation dataset\n",
        "val_ds = tf.keras.preprocessing.image_dataset_from_directory(\n",
        "  data_dir,\n",
        "  validation_split=0.2,\n",
        "  subset=\"validation\",\n",
        "  seed=123,\n",
        "  color_mode='grayscale',\n",
        "  image_size=(img_height, img_width),\n",
        "  label_mode=\"int\",\n",
        "  batch_size=512)\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Number of Classes: 5\n",
            "['0_diode', '1_resistor', '2_inductor', '3_capacitor', '5_power']\n"
          ]
        }
      ],
      "source": [
        "class_names = train_ds.class_names\n",
        "print('Number of Classes:',len(class_names))\n",
        "print(class_names)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "normalization_layer = tf.keras.layers.experimental.preprocessing.Rescaling(1./255)\n",
        "\n",
        "AUTOTUNE = tf.data.experimental.AUTOTUNE\n",
        "\n",
        "train_ds = train_ds.cache().prefetch(buffer_size=AUTOTUNE)\n",
        "val_ds = val_ds.cache().prefetch(buffer_size=AUTOTUNE)\n",
        "\n",
        "regularization = 0.0005"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "from tensorflow.keras import layers\n",
        "from tensorflow.keras import regularizers\n",
        "\n",
        "model = tf.keras.Sequential([\n",
        "  layers.experimental.preprocessing.Rescaling(1./255),\n",
        "\n",
        "  layers.Conv2D(32, 4, activation='relu', kernel_regularizer=\"l2\"),\n",
        "  layers.MaxPooling2D(),\n",
        "\n",
        "  layers.Conv2D(64, 5, activation='relu', kernel_regularizer=\"l2\"),\n",
        "  layers.MaxPooling2D(),\n",
        "\n",
        "  layers.Conv2D(64, 5, activation='relu', kernel_regularizer=\"l2\"),\n",
        "  layers.MaxPooling2D(),\n",
        "\n",
        "  layers.Conv2D(128, 6, activation='relu', kernel_regularizer=\"l2\"),\n",
        "  layers.MaxPooling2D(),\n",
        "\n",
        "  layers.Flatten(),\n",
        "\n",
        "  layers.Dense(128, activation='relu', kernel_regularizer=\"l2\"),\n",
        "\n",
        "  layers.Dense(len(class_names), activation='softmax', kernel_regularizer=\"l2\")\n",
        "])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "model.compile(\n",
        "  optimizer='adam',\n",
        "  loss=tf.losses.SparseCategoricalCrossentropy(from_logits=True),\n",
        "  metrics=['accuracy'])"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Epoch 1/100\n",
            "15/15 [==============================] - 6s 312ms/step - loss: 5.2150 - accuracy: 0.2150 - val_loss: 3.4076 - val_accuracy: 0.2036\n",
            "Epoch 2/100\n",
            "15/15 [==============================] - 4s 244ms/step - loss: 3.0682 - accuracy: 0.2334 - val_loss: 2.4413 - val_accuracy: 0.2255\n",
            "Epoch 3/100\n",
            "15/15 [==============================] - 4s 244ms/step - loss: 2.2339 - accuracy: 0.2737 - val_loss: 1.7380 - val_accuracy: 0.4286\n",
            "Epoch 4/100\n",
            "15/15 [==============================] - 4s 244ms/step - loss: 1.6854 - accuracy: 0.4803 - val_loss: 1.5214 - val_accuracy: 0.4778\n",
            "Epoch 5/100\n",
            "15/15 [==============================] - 4s 244ms/step - loss: 1.4056 - accuracy: 0.5161 - val_loss: 1.2569 - val_accuracy: 0.5406\n",
            "Epoch 6/100\n",
            "15/15 [==============================] - 4s 245ms/step - loss: 1.2581 - accuracy: 0.5602 - val_loss: 1.1345 - val_accuracy: 0.5997\n",
            "Epoch 7/100\n",
            "15/15 [==============================] - 4s 245ms/step - loss: 1.1299 - accuracy: 0.6067 - val_loss: 1.0637 - val_accuracy: 0.6321\n",
            "Epoch 8/100\n",
            "15/15 [==============================] - 4s 244ms/step - loss: 1.0688 - accuracy: 0.6216 - val_loss: 1.0150 - val_accuracy: 0.6572\n",
            "Epoch 9/100\n",
            "15/15 [==============================] - 4s 245ms/step - loss: 1.0175 - accuracy: 0.6464 - val_loss: 0.9670 - val_accuracy: 0.6850\n",
            "Epoch 10/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.9791 - accuracy: 0.6597 - val_loss: 0.9615 - val_accuracy: 0.6761\n",
            "Epoch 11/100\n",
            "15/15 [==============================] - 4s 246ms/step - loss: 0.9660 - accuracy: 0.6571 - val_loss: 0.9344 - val_accuracy: 0.6719\n",
            "Epoch 12/100\n",
            "15/15 [==============================] - 4s 246ms/step - loss: 0.9401 - accuracy: 0.6774 - val_loss: 0.9461 - val_accuracy: 0.6546\n",
            "Epoch 13/100\n",
            "15/15 [==============================] - 4s 246ms/step - loss: 0.9372 - accuracy: 0.6655 - val_loss: 0.8987 - val_accuracy: 0.6782\n",
            "Epoch 14/100\n",
            "15/15 [==============================] - 4s 246ms/step - loss: 0.8833 - accuracy: 0.7120 - val_loss: 0.8507 - val_accuracy: 0.7190\n",
            "Epoch 15/100\n",
            "15/15 [==============================] - 4s 245ms/step - loss: 0.8936 - accuracy: 0.7037 - val_loss: 0.8590 - val_accuracy: 0.7007\n",
            "Epoch 16/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.8498 - accuracy: 0.7264 - val_loss: 0.7972 - val_accuracy: 0.7619\n",
            "Epoch 17/100\n",
            "15/15 [==============================] - 4s 246ms/step - loss: 0.8105 - accuracy: 0.7591 - val_loss: 0.7805 - val_accuracy: 0.7650\n",
            "Epoch 18/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.7730 - accuracy: 0.7796 - val_loss: 0.7909 - val_accuracy: 0.7609\n",
            "Epoch 19/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.7756 - accuracy: 0.7749 - val_loss: 0.7434 - val_accuracy: 0.7917\n",
            "Epoch 20/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.7192 - accuracy: 0.8160 - val_loss: 0.7248 - val_accuracy: 0.8142\n",
            "Epoch 21/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.7176 - accuracy: 0.8161 - val_loss: 0.7190 - val_accuracy: 0.8231\n",
            "Epoch 22/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.7094 - accuracy: 0.8193 - val_loss: 0.7351 - val_accuracy: 0.8174\n",
            "Epoch 23/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.6957 - accuracy: 0.8368 - val_loss: 0.6701 - val_accuracy: 0.8566\n",
            "Epoch 24/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.6563 - accuracy: 0.8588 - val_loss: 0.6662 - val_accuracy: 0.8629\n",
            "Epoch 25/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.6439 - accuracy: 0.8653 - val_loss: 0.6622 - val_accuracy: 0.8550\n",
            "Epoch 26/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.6343 - accuracy: 0.8673 - val_loss: 0.6566 - val_accuracy: 0.8587\n",
            "Epoch 27/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.6364 - accuracy: 0.8714 - val_loss: 0.6353 - val_accuracy: 0.8838\n",
            "Epoch 28/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.6322 - accuracy: 0.8768 - val_loss: 0.6121 - val_accuracy: 0.8917\n",
            "Epoch 29/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.6136 - accuracy: 0.8852 - val_loss: 0.6199 - val_accuracy: 0.8891\n",
            "Epoch 30/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5852 - accuracy: 0.8980 - val_loss: 0.6121 - val_accuracy: 0.8828\n",
            "Epoch 31/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5817 - accuracy: 0.9025 - val_loss: 0.5946 - val_accuracy: 0.8943\n",
            "Epoch 32/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5789 - accuracy: 0.9039 - val_loss: 0.6131 - val_accuracy: 0.8849\n",
            "Epoch 33/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5838 - accuracy: 0.8993 - val_loss: 0.5749 - val_accuracy: 0.9173\n",
            "Epoch 34/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5500 - accuracy: 0.9214 - val_loss: 0.5611 - val_accuracy: 0.9173\n",
            "Epoch 35/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5413 - accuracy: 0.9262 - val_loss: 0.5579 - val_accuracy: 0.9236\n",
            "Epoch 36/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5373 - accuracy: 0.9251 - val_loss: 0.5603 - val_accuracy: 0.9142\n",
            "Epoch 37/100\n",
            "15/15 [==============================] - 4s 247ms/step - loss: 0.5586 - accuracy: 0.9119 - val_loss: 0.5872 - val_accuracy: 0.9048\n",
            "Epoch 38/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5443 - accuracy: 0.9256 - val_loss: 0.6662 - val_accuracy: 0.8629\n",
            "Epoch 39/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5863 - accuracy: 0.8993 - val_loss: 0.5961 - val_accuracy: 0.8990\n",
            "Epoch 40/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5503 - accuracy: 0.9207 - val_loss: 0.5583 - val_accuracy: 0.9053\n",
            "Epoch 41/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.5148 - accuracy: 0.9421 - val_loss: 0.5866 - val_accuracy: 0.8938\n",
            "Epoch 42/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5202 - accuracy: 0.9330 - val_loss: 0.5590 - val_accuracy: 0.9178\n",
            "Epoch 43/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5048 - accuracy: 0.9420 - val_loss: 0.5114 - val_accuracy: 0.9403\n",
            "Epoch 44/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4851 - accuracy: 0.9505 - val_loss: 0.5051 - val_accuracy: 0.9440\n",
            "Epoch 45/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4760 - accuracy: 0.9563 - val_loss: 0.5097 - val_accuracy: 0.9367\n",
            "Epoch 46/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4898 - accuracy: 0.9451 - val_loss: 0.5206 - val_accuracy: 0.9267\n",
            "Epoch 47/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.5046 - accuracy: 0.9385 - val_loss: 0.5034 - val_accuracy: 0.9466\n",
            "Epoch 48/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4803 - accuracy: 0.9482 - val_loss: 0.5113 - val_accuracy: 0.9414\n",
            "Epoch 49/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4646 - accuracy: 0.9587 - val_loss: 0.5149 - val_accuracy: 0.9341\n",
            "Epoch 50/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4783 - accuracy: 0.9515 - val_loss: 0.4958 - val_accuracy: 0.9461\n",
            "Epoch 51/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.4745 - accuracy: 0.9524 - val_loss: 0.4962 - val_accuracy: 0.9424\n",
            "Epoch 52/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4602 - accuracy: 0.9629 - val_loss: 0.4819 - val_accuracy: 0.9540\n",
            "Epoch 53/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4516 - accuracy: 0.9650 - val_loss: 0.4916 - val_accuracy: 0.9487\n",
            "Epoch 54/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4490 - accuracy: 0.9658 - val_loss: 0.4936 - val_accuracy: 0.9419\n",
            "Epoch 55/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4525 - accuracy: 0.9616 - val_loss: 0.4787 - val_accuracy: 0.9503\n",
            "Epoch 56/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4560 - accuracy: 0.9578 - val_loss: 0.4843 - val_accuracy: 0.9466\n",
            "Epoch 57/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.4433 - accuracy: 0.9692 - val_loss: 0.4796 - val_accuracy: 0.9534\n",
            "Epoch 58/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4493 - accuracy: 0.9641 - val_loss: 0.4776 - val_accuracy: 0.9529\n",
            "Epoch 59/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4438 - accuracy: 0.9668 - val_loss: 0.4839 - val_accuracy: 0.9492\n",
            "Epoch 60/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4527 - accuracy: 0.9574 - val_loss: 0.5254 - val_accuracy: 0.9168\n",
            "Epoch 61/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4774 - accuracy: 0.9427 - val_loss: 0.5461 - val_accuracy: 0.9205\n",
            "Epoch 62/100\n",
            "15/15 [==============================] - 4s 248ms/step - loss: 0.4990 - accuracy: 0.9374 - val_loss: 0.5012 - val_accuracy: 0.9409\n",
            "Epoch 63/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4629 - accuracy: 0.9565 - val_loss: 0.4724 - val_accuracy: 0.9513\n",
            "Epoch 64/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4446 - accuracy: 0.9658 - val_loss: 0.4995 - val_accuracy: 0.9320\n",
            "Epoch 65/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4584 - accuracy: 0.9571 - val_loss: 0.5066 - val_accuracy: 0.9283\n",
            "Epoch 66/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4465 - accuracy: 0.9612 - val_loss: 0.5045 - val_accuracy: 0.9314\n",
            "Epoch 67/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4392 - accuracy: 0.9661 - val_loss: 0.4749 - val_accuracy: 0.9430\n",
            "Epoch 68/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4213 - accuracy: 0.9766 - val_loss: 0.4576 - val_accuracy: 0.9545\n",
            "Epoch 69/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4067 - accuracy: 0.9813 - val_loss: 0.4569 - val_accuracy: 0.9513\n",
            "Epoch 70/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4080 - accuracy: 0.9785 - val_loss: 0.4558 - val_accuracy: 0.9576\n",
            "Epoch 71/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4112 - accuracy: 0.9748 - val_loss: 0.4597 - val_accuracy: 0.9513\n",
            "Epoch 72/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4139 - accuracy: 0.9759 - val_loss: 0.4515 - val_accuracy: 0.9534\n",
            "Epoch 73/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4105 - accuracy: 0.9778 - val_loss: 0.4606 - val_accuracy: 0.9445\n",
            "Epoch 74/100\n",
            "15/15 [==============================] - 4s 251ms/step - loss: 0.4080 - accuracy: 0.9768 - val_loss: 0.4671 - val_accuracy: 0.9456\n",
            "Epoch 75/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4137 - accuracy: 0.9728 - val_loss: 0.4752 - val_accuracy: 0.9409\n",
            "Epoch 76/100\n",
            "15/15 [==============================] - 4s 251ms/step - loss: 0.4175 - accuracy: 0.9679 - val_loss: 0.4607 - val_accuracy: 0.9451\n",
            "Epoch 77/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4139 - accuracy: 0.9746 - val_loss: 0.4538 - val_accuracy: 0.9534\n",
            "Epoch 78/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4024 - accuracy: 0.9762 - val_loss: 0.4737 - val_accuracy: 0.9424\n",
            "Epoch 79/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4101 - accuracy: 0.9729 - val_loss: 0.5059 - val_accuracy: 0.9267\n",
            "Epoch 80/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4256 - accuracy: 0.9644 - val_loss: 0.5596 - val_accuracy: 0.9053\n",
            "Epoch 81/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4367 - accuracy: 0.9620 - val_loss: 0.4538 - val_accuracy: 0.9613\n",
            "Epoch 82/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4070 - accuracy: 0.9779 - val_loss: 0.4439 - val_accuracy: 0.9550\n",
            "Epoch 83/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.3888 - accuracy: 0.9862 - val_loss: 0.4335 - val_accuracy: 0.9618\n",
            "Epoch 84/100\n",
            "15/15 [==============================] - 4s 251ms/step - loss: 0.3912 - accuracy: 0.9804 - val_loss: 0.4429 - val_accuracy: 0.9592\n",
            "Epoch 85/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.3928 - accuracy: 0.9788 - val_loss: 0.4330 - val_accuracy: 0.9639\n",
            "Epoch 86/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.3979 - accuracy: 0.9774 - val_loss: 0.4436 - val_accuracy: 0.9498\n",
            "Epoch 87/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4034 - accuracy: 0.9755 - val_loss: 0.4776 - val_accuracy: 0.9341\n",
            "Epoch 88/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4309 - accuracy: 0.9598 - val_loss: 0.4898 - val_accuracy: 0.9414\n",
            "Epoch 89/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4593 - accuracy: 0.9509 - val_loss: 0.4892 - val_accuracy: 0.9341\n",
            "Epoch 90/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4349 - accuracy: 0.9598 - val_loss: 0.5034 - val_accuracy: 0.9236\n",
            "Epoch 91/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4283 - accuracy: 0.9649 - val_loss: 0.5248 - val_accuracy: 0.9220\n",
            "Epoch 92/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4564 - accuracy: 0.9505 - val_loss: 0.4780 - val_accuracy: 0.9440\n",
            "Epoch 93/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4376 - accuracy: 0.9653 - val_loss: 0.5014 - val_accuracy: 0.9362\n",
            "Epoch 94/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4155 - accuracy: 0.9713 - val_loss: 0.4837 - val_accuracy: 0.9419\n",
            "Epoch 95/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4030 - accuracy: 0.9771 - val_loss: 0.4722 - val_accuracy: 0.9435\n",
            "Epoch 96/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.4099 - accuracy: 0.9714 - val_loss: 0.4649 - val_accuracy: 0.9471\n",
            "Epoch 97/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4063 - accuracy: 0.9747 - val_loss: 0.4709 - val_accuracy: 0.9440\n",
            "Epoch 98/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.4063 - accuracy: 0.9781 - val_loss: 0.4649 - val_accuracy: 0.9461\n",
            "Epoch 99/100\n",
            "15/15 [==============================] - 4s 250ms/step - loss: 0.3997 - accuracy: 0.9774 - val_loss: 0.4433 - val_accuracy: 0.9592\n",
            "Epoch 100/100\n",
            "15/15 [==============================] - 4s 249ms/step - loss: 0.3944 - accuracy: 0.9797 - val_loss: 0.4328 - val_accuracy: 0.9597\n"
          ]
        }
      ],
      "source": [
        "history = model.fit(\n",
        "            train_ds,\n",
        "            validation_data=val_ds,\n",
        "            epochs=100)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXgAAAEWCAYAAABsY4yMAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADh0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uMy4yLjIsIGh0dHA6Ly9tYXRwbG90bGliLm9yZy+WH4yJAAAgAElEQVR4nO3dd3hb5d3/8fdXw5Jtea8MO3H2IjtkQAAHKGXThhnKCLRw0aelpQMeoLRAF7Tw8KPt08UoUKCkPFBSRtIyikkgrITs6QScxHES723ZknX//jiycYLt2I4VO0ff13X5inQknfO9Jfjo1n3OuY8YY1BKKWU/jv4uQCmlVGRowCullE1pwCullE1pwCullE1pwCullE1pwCullE1pwCt1GBHJE5GiCKw3V0SMiLj6et1KdUQDXvUpESkUkTP7u45IEZFtInJ9B8u/KyKr+3A7+mWgjpoGvFI98xRwTQfLrw4/ptSAoQGvjgkR8YjIwyJSHP57WEQ84cfSReRVEakSkQoRWSkijvBj/y0i+0SkVkS2i8gZnaz/PBFZKyI1IrJXRO5p91hrb/haEdkjImUi8qN2j8eKyJMiUikiW4ATu2jK08B8ERne7vUTgSnAc13V0VdEZIiIvBx+r3aKyA3tHpstIqvD2z8oIg+Fl3tF5BkRKQ+/zx+LSFZf16YGFv35p46VHwFzgWmAAf4J3AX8GPgBUARkhJ87FzAiMg74NnCiMaZYRHIBZyfrr8fqWW8GTgDeEJF1xpil7Z4zHxgHjAU+EpF/GGO2AncDo8J/8cDyzhphjCkSkbexeuw/Dy++GlhmjCkTke7UcbSWAJuAIcD48DZ2GWP+A/wG+I0x5mkR8YVrALgWSAJygCasz6GxD2tSA5D24NWx8jXgp8aYEmNMKXAvVjACBIDBwHBjTMAYs9JYkyS1AB5gooi4jTGFxphdHa3cGJNvjNlojAkZYzYAzwGnHfa0e40xjcaY9cB6YGp4+WXAL4wxFcaYvcBvj9CWp1prD//S+Fp4WXfr6DURyQFOBv7bGOM3xqwDHuPzYaMAMFpE0o0xdcaYD9otTwNGG2NajDFrjDE1fVWXGpg04NWxMgTY3e7+7vAygAeAncDrIvKpiNwOYIzZCdwC3AOUiMgSERlCB0Rkjoi8LSKlIlIN3ASkH/a0A+1uNwC+drXtPay2rvwDGCwic4E8IA54rQd1HI0hQIUxpvaweoeGb38d6xfKtvAwzPnh5U8D/waWhIfIfi0i7j6sSw1AGvDqWCkGhre7Pyy8DGNMrTHmB8aYkcCFwPdbx9qNMX8zxswPv9YAv+pk/X8DXgZyjDFJwJ8A6WZt+7GGLtrX1iljTAPwAlav+WpgiTGmuQ/q6I5iIFVEEg6rd1+4tgJjzCIgE+u9ekFE4sO/jO41xkwETgLOp+OdxcpGNOBVJLjDO/Va/1xYQxV3iUiGiKQDPwGeARCR80VktIgIUI01NBMSkXEicnp4Z6wfa8w41Mk2E7B6tn4RmQ1c2YN6nwfuEJEUEckGbu7Ga54CLgcu5tCjZ46mjo542r+XWEG+CrgvvGwKVq+99b28SkQyjDEhoCq8jpCILBCRySLiBGqwhmw6ey+VTWjAq0hYhhXGrX/3YO2QXA1sADYCn/D5TsoxwJtAHfA+8AdjzNtY4+/3A2VYwyuZwB2dbPO/gJ+KSC3Wl8fzPaj3Xqxhjs+A17GGM45kBdaXUZEx5uPe1CEifxKRPx1hO3Uc+l6eDiwCcrF68y8Bdxtj3gw//2xgs4jUYe1wvcIY0wgMwvrVUQNsBd7pZjvVcUz0gh9KKWVP2oNXSimb0oBXSimb0oBXSimb0oBXSimbGlBTFaSnp5vc3Nxevba+vp74+Pi+LWiAi8Y2Q3S2OxrbDNHZ7p62ec2aNWXGmIyOHhtQAZ+bm8vq1b2bcTU/P5+8vLy+LWiAi8Y2Q3S2OxrbDNHZ7p62WUQ6PfNah2iUUsqmNOCVUsqmNOCVUsqmBtQYvFLKPgKBAEVFRfj9/l6vIykpia1bt/ZhVQNfZ232er1kZ2fjdnd/ElANeKVURBQVFZGQkEBubi7WPHI9V1tbS0JCwpGfaCMdtdkYQ3l5OUVFRYwYMaLb69IhGqVURPj9ftLS0nod7upzIkJaWlqPfw1pwCulIkbDve/05r20RcD/9q0CNpYG+7sMpZQaUGwR8H9+Zxebylr6uwyl1ABSXl7OtGnTmDZtGoMGDWLo0KFt95ubm4+8gk6ce+65VFVVHfmJA4AtdrJ63U6aQ3pxGqXU59LS0li3bh0A99xzDz6fjx/+8IdtjweDQVyunkfgsmXL+qzGSLNFD97rdtKsHXil1BEsXryYm266iTlz5nDbbbfx0UcfMW/ePKZPn85JJ53E9u3bAXjyySdZuHAhZ599NmPGjOG2225rW0dubi5lZWUUFhYyYcIEbrjhBiZNmsRZZ51FY2MjAB9//DFTpkxh2rRp3HrrrZxwwgn90l6b9OAdNIf0ylRKDVT3vrKZLcU1PX5dS0sLTqezw8cmDknk7gsm9XidRUVFrFq1CqfTSU1NDStXrsTlcvHmm29y55138uKLLwKwbt061q5di8fjYdy4cdx8883k5OQcsq6CggKee+45Hn30US677DJefPFFrrrqKq677joeffRR5s2bx+23397jGvuKLQI+NsbJUQypKaWiyKWXXtr2pVFdXc21115LQUEBIkIgEGh73hlnnEFSUhIAEydOZPfu3V8I+BEjRjBt2jQAZs6cSWFhIVVVVdTW1jJv3jwArrzySl599dVj0bQvsEXAe11O6hu1B6/UQNWbnjZE5kSn9lPx/vjHP2bBggW89NJLFBYWHjKLo8fjabvtdDoJBr94pN7hz2kdohkobDEGHxujY/BKqZ6rrq5m6NChgDXu3heSk5NJSEjgww8/BGDJkiV9st7esEXAe1xOmvUgGqVUD912223ccccdTJ8+vcMeem89/vjj3HDDDUybNo36+vq2oZ5jzRZDNFYPXodolFIdu+eeezpcPm/ePHbs2NF2/+c//zlgHW2zePHituXtx9ALCwsBSE9PZ9OmTW3L2x+COWnSJDZs2ADA/fffz6xZs462Cb1ii4D3uhwEtAevlBogXnvtNe677z6CwSDDhw/vs+GfnrJFwMfGOGnSHrxSaoC4/PLLufzyy/u7DHuMwXvdTgK6k1UppQ5hm4BvDllzJiullLLYJOCtZjQFdSBeKaVa2SLgY93WWWmNejC8Ukq1sUXAe8MB7w9qwCulLAsWLODf//73IcsefvhhvvnNb3b4/Ly8PFavXg10PiXwPffcw4MPPtjldpcuXcqWLVva7v/kJz/hzTff7Gn5fcIWAa89eKXU4RYtWvSFs0iXLFnCokWLjvjaZcuWkZyc3KvtHh7wP/3pTznzzDN7ta6jZYuAbx2D9+vB8EqpsEsuuYTXXnut7eIehYWFFBcX89xzzzFr1iwmTZrE3Xff3eFrW6cEBvjFL37B2LFjmT9/ftt0wgCPPvooJ554IlOnTuXiiy+moaGBVatW8fLLL3Prrbcybdo0du3axeLFi3nhhRcAeOutt5g+fTqTJ0/m+uuvp6mpqW17d999NzNmzGDu3Lls27atT94DWxwH3zpE06jHSio1MC2/HQ5s7PHLYluC4OwkpgZNhnPu7/S1qampzJ49m+XLl3PRRRexZMkSLrvsMu68805SU1NpaWnhjDPOYMOGDUyZMqXDdaxZs4YlS5awbt06gsEgM2bMYObMmQAsXLiQG264AYC77rqLxx9/nJtvvpkLL7yQ888/n0suueSQdfn9fhYvXsxbb73F2LFjueaaa/jjH//ILbfcAlhnxn7yySc89NBDPPjggzz22GM9fbu+wCY9eCvgmzTglVLttB+maR2eef7555kxYwbTp09n8+bNhwynHG7lypV89atfJS4ujsTERC688MK2xzZt2sQpp5zC5MmTefbZZ9m8eXOXtWzfvp0RI0YwduxYAK699lpWrFjR9vjChQsBmDZtWtt0CEfLFj34WO3BKzWwddHT7krjUU4XfNFFF/G9732PTz75hIaGBlJTU3nwwQf5+OOPSUlJYfHixfj9/l6te/HixSxdupSpU6fy5JNPkp+f3+s64fOphzubmrg3bNWD14BXSrXn8/lYsGAB119/PYsWLaKmpob4+HiSkpI4ePAgy5cv7/L1p556KkuXLqWxsZHa2lpeeeWVtsdqa2sZPHgwgUCAZ599tm15QkICtbW1X1jXuHHjKCwsZOfOnQA8/fTTnHbaaX3U0o7ZIuBbe/C6k1UpdbhFixaxfv16Fi1axNSpU5k+fTrjx4/nyiuv5OSTT+7ytTNmzODyyy9n6tSpnHPOOZx44oltj/3sZz9jzpw5nHzyyYwfP75t+RVXXMEDDzzA9OnT2bVrV9tyr9fLE088waWXXsrkyZNxOBzcdNNNfd/gdmQgnd4/a9Ys03ocak+U1PiZ/cu3+NlXTuDqucMjUNnAlJ+ff8gVaKJFNLb7eGzz1q1bmTBhwlGtIxJXdBroumpzR++piKwxxnQ4H7EtevDeGN3JqpRSh7NHwLv0RCellDqcLQLe7RQcolMVKDXQDKQh4ONdb97LiAe8iDhFZK2IvHrkZ/d6G8Q4oFEvzKrUgOH1eikvL9eQ7wPGGMrLy/F6vT163bE4Dv67wFYgMZIbiXFqD16pgSQ7O5uioiJKS0t7vQ6/39/jUDveddZmr9dLdnZ2j9YV0YAXkWzgPOAXwPcjua0Yp+DXMXilBgy3282IESOOah35+flMnz69jyo6PvRlmyN6mKSIvADcByQAPzTGnN/Bc24EbgTIysqaefjsb911+zt1ZCe6+Pb06Pm2r6urw+fz9XcZx1w0tjsa2wzR2e6etnnBggWdHiYZsR68iJwPlBhj1ohIXmfPM8Y8AjwC1nHwvT3W17tqOQnJqeTlze7V649Hx+Ox0X0hGtsdjW2G6Gx3X7Y5kjtZTwYuFJFCYAlwuog8E6mNxTj1TFallGovYgFvjLnDGJNtjMkFrgD+Y4y5KlLbi3GIzkWjlFLt2OI4eAC3E/wa8Eop1eaYTBdsjMkH8iO5DY8TKvUoGqWUamOfHrxDdAxeKaXasU3Axzh1PnillGrPRgEvOgavlFLt2CfgHdAUDBEK6bwXSikFdgp4a8ZgmoI6Dq+UUmCrgBdAx+GVUqqVfQI+3BIdh1dKKYt9Al578EopdQgbBbz1r/bglVLKYp+A1yEapZQ6hH0CvnWIRi/bp5RSgK0C3vpXe/BKKWWxT8A7dCerUkq1Z5+A1x68UkodwjYB79aAV0qpQ9gm4D3hnaw6ZbBSSllsE/DucEt0DF4ppSy2CXiXQ3A5dMpgpZRqZZuAB/C6ndqDV0qpMNsFvI7BK6WU5fgPeGPg48dJqtqC1+3QIRqllAo7/gNeBN74Cellq4h1OzXglVIq7PgPeABvMu5APbExOgavlFKt7BHwsSm4grV4XdqDV0qpVjYJ+GRcwXq8MU4adSerUkoBNgp4d6AWr8tBk/bglVIKsEvAe5NxBet0DF4ppdqxR8DHpuAO1OkYvFJKtWOTgE/GYQIkuII0NmvAK6UU2CbgUwBIklr8Qd3JqpRSYJeA9yYDkEw9zcEQLSHTzwUppVT/s0fAh3vwidQB0BTUYRqllLJJwFs9+ARTD6Dj8EophW0C3urB+0I1ADoOr5RS2CXgw2PwcSFriEZ78EopZZeA9yRicBDbEu7B67HwSillk4B3OAi64okN1gIa8EopBREMeBHxishHIrJeRDaLyL2R2hZAwO3DEw54na5AKaXAFcF1NwGnG2PqRMQNvCsiy40xH0RiY0GXj5hANYBetk8ppYhgD95Y6sJ33eG/iJ2BFHT5cDdbAa89eKWU6kYPXkQuBf5ljKkVkbuAGcDPjTGfdOO1TmANMBr4vTHmww6ecyNwI0BWVhb5+fk9a0HYGPHiri4EYP3GzSRW7ujVeo4ndXV1vX6/jmfR2O5obDNEZ7v7tM3GmC7/gA3hf+cD+cB5wIdHet1h60gG3gZO6Op5M2fONL1V9OfLTct9uWb4f79qnlr1Wa/Xczx5++23+7uEfhGN7Y7GNhsTne3uaZuB1aaTTO3OEE3reMd5wCPGmNeAmB5+iVSFA/7snryuJ4IuH9JUhRDSo2iUUorujcHvE5E/A5cDy0TE053XiUiGiCSHb8cCXwK2HU2xXQm4ExATwoefxmbdyaqUUt0J+MuAfwNfDvfEU4Fbu/G6wcDbIrIB+Bh4wxjzaq8rPYKgKx6AdGc9fp1sTCmlunWY5GDgNWNMk4jkAVOAvx7pRcaYDcD0oyuv+wLuBAAyXI06VYFSStG9HvyLQIuIjAYeAXKAv0W0ql4IunwAZLgadAxeKaXoXsCHjDFBYCHwO2PMrVi9+gGlNeAz3X5q/IF+rkYppfpfdwI+ICKLgGuA1jF0d+RK6p2A2wr4wTGNlNU193M1SinV/7oT8NcB84BfGGM+E5ERwNORLavn2oZo3I1U1GvAK6XUEQPeGLMF+CGwUUROAIqMMb+KeGU9FHJ6wOUlzdFAeV1Tf5ejlFL9rjtTFeQBTwGFgAA5InKtMWZFZEvrBW8yKY56qhoDtIQMTof0d0VKKdVvunOY5P8AZxljtgOIyFjgOWBmJAvrldgUEk0dxkBlQzPpPk9/V6SUUv2mO2Pw7tZwBzDG7GAA7mQFIDaZ+PBl+8p1R6tSKsp1pwe/WkQeA54J3/8asDpyJR2F2BRiaz8DoLy+CUjo33qUUqofdSfgvwl8C/hO+P5K4PcRq+hoeJOJCVjXZdUevFIq2h0x4I0xTcBD4T8AROQ94OQI1tU7sSk4m6oA9FBJpVTU6+0VnYb1aRV9JTYZR6AetwT1UEmlVNTrbcBH7NJ7RyU2BYBhsQHKtQevlIpynQ7RiMjCzh4CYiNTzlHyJgOQE9ukQzRKqajX1Rj8BV08FrF53Y9KuAc/1OunQHeyKqWiXKcBb4y57lgW0idirR78EI+fD2p1DF4pFd16OwY/MIV78JmuRh2DV0pFPXsFfHgMPt3ZSFVDgGCLXptVKRW9bBbwSQCkOOoBqGjQXrxSKnp1GvAiclu725ce9tgvI1lUrzld4EkkiXDA6zCNUiqKddWDv6Ld7TsOe+zsCNTSN2JT8IWqAajQI2mUUlGsq4CXTm53dH/g8GURF6gAoEx78EqpKNZVwJtObnd0f+DwZeLxlwFQodMVKKWiWFcnOk0VkRrCZ66GbxO+7414Zb3ly8K5531E0EMllVJRrasTnZzHspA+48tCGsrJjHVowCulolpXR9GcKCLndLD8HBEZeJfra+XLBGBEXKPOKKmUimpdjcH/CtjSwfItwAORKacP+LIAGOGt08MklVJRrauATzDG7D58YXhZeuRKOkrhgM+JqdUhGqVUVOsq4FO6eCyurwvpM+EhmsHOGr1sn1IqqnUV8G+KyC9EpO2Yd7H8FPhP5EvrpXDAZzqqqW4MEND5aJRSUaqrwyR/ADwG7BSRdeFlU4HVwDciXVivuTzgTSYtZJ3sVFnfTGbiwD2qUymlIqWrwyTrgUUiMhKYFF682Rjz6TGp7Gj4skgKVQLWsfAa8EqpaNTVJfu+jLWj9QXg03bLLwGqjTFvHIP6eichi/i6ckAnHFNKRa+uxuB/ArzTwfJ84KcRqaav+LLwNlnTFZTpsfBKqSjVVcB7jDGlhy80xpQB8ZErqQ/4snA3WqVrD14pFa26CvhEEfnCEI6IuIHYyJXUB3yZSKABn/j1UEmlVNTqKuD/ATwqIm29dRHxAX8KPzZwhU92Gh1Xryc7KaWiVlcBfxdwENgtImtE5BPgM6A0/NjAFT4WfqS3ntJaHYNXSkWnrg6TDAK3i8i9wOjw4p3GmMburFhEcoC/AllY88c/Yoz5zVHW2z3hHvyY+AZeLK8/JptUSqmB5kizSQ4yxjQaYzYC04ElIvJbEUntxrqDwA+MMROBucC3RGRi35R9BG1DNA3sLq/Xs1mVUlGpqyGaPwPNACJyKnA/Vo+8GnjkSCs2xuw3xnwSvl0LbAWGHm3B3RKbCuIk211DoMWwt6LhmGxWKaUGEjGm46vvich6Y8zU8O3fA6XGmHvC99cZY6Z1eyMiucAK4ARjTM1hj90I3AiQlZU1c8mSJT1vBVBXV4fP52u7P2/VdRT6pvPl4q/znekeZmR1NSvD8enwNkeLaGx3NLYZorPdPW3zggUL1hhjZnX0WFep5xQRV3gs/gzCIdyN1x0ifOTNi8Ath4c7gDHmEcK/CGbNmmXy8vK6u+pD5Ofnc8hrt+cwMs4BxRCbNYK8vFG9Wu9A9oU2R4lobHc0thmis9192eaugvo54B0RKQMagZUAIjIaa5jmiMLHzL8IPGuMObaHVvqycNeVkJngYVdp3THdtFJKDQRdHUXzCxF5CxgMvG4+H8txADcfacXhaYYfB7YaYx7qi2J7xJcJBzYxKsPHzhINeKVU9OlyqMUY80EHy3Z0c90nA1cDG9tNN3ynMWZZz0rsJV8W1JcwemQsS9cfwBhDu6ntlVLK9iK259EY8y7Qf4nqy4JQkInJLTztD1Ja10Rmgk4brJSKHl0dJnl8C5/NOtZnnZe1q0RPeFJKRRcbB7x1stNwjzX+vlN3tCqlooztAz7NVBIX42SX7mhVSkUZGwe8NUQjdQcZleHTQyWVUlHHvgEf4wN3HNSVMCojXnvwSqmoY9+AF7F68bUHGJ3po7jaT31TsL+rUkqpY8a+AQ+QMQGKPmJUunXNks/K9EgapVT0sHfAjzsbqvYwwVUEoGe0KqWiir0DfuzZAGSX5OMQdEerUiqq2DvgEwbB0Jm4Cv7F8LR4DXilVFSxd8ADjDsH9q1hVloTW4q/MFuxUkrZVhQE/LkAnOtZT2F5A5X1zf1ckFJKHRv2D/jMiZA8nKn17wOwbm9VPxeklFLHhv0DXgTGnUvKwVXEi5+1GvBKqShh/4AHGHcOEvRzaeou1u6p7O9qlFLqmIiOgB9+EniSOD9mLev3VhEKdXyhcaWUspPoCHinG0YtYGLjGmr8AT7VM1qVUlEgOgIeYMQpxPkPMkxKdEerUioqRE/A554CQF7Mdh2HV0pFhegJ+PSxEJ/BWfEF2oNXSkWF6Al4Ecidz5TgRrYdqKGxuaW/K1JKqYiKnoAHyJ1PYnMJQ81BNu6r7u9qlFIqoqIs4K1x+LmOLazbq+PwSil7i66AD4/Dn+Hdwdo9Og6vlLK36Ar48Dj8HNnCyoJSyuua+rsipZSKmOgKeIDc+SQHS8kI7Od3/9nZ39UopVTERGHAW+PwN488wDMf7NbrtCqlbCv6Aj48Dn9uwk5iXA5+/a9t/V2RUkpFRPQFvAiMOA3vZ2/wnZMyWL7pAGt26xE1Sin7ib6AB5h/C/hruJ6lZCR4+OWyrRijM0wqpewlOgN+0GSYchkxqx/hJ6cksWZ3JS+t3dffVSmlVJ+KzoAHWPAjMCHOr3yKaTnJ/HLZNmr8gf6uSiml+kz0BnzKcDjxG8i6Z/n1qTGU1zfx/97Y0d9VKaVUn4negAc45YcQ42Ps+l9x1exsnlpVyNb9Nf1dlVJK9YnoDvj4NFhwJ+x8g7tcz5Ac6+ZHL23UmSaVUrYQ3QEPMOcmmPtfeNY8wt/GvsPavVUsevQDynQaA6XUcU4DXgTO+gVM+xrjt/0vy2Zb88V/9Q/vsbOkrr+rU0qpXotYwIvIX0SkREQ2RWobfcbhgAt+C+PPZ8L6+3h3wlJC/gYW/uE9/rXpQH9Xp5RSvRLJHvyTwNkRXH/fcrrg0idh/vdJ3/F38pN/yilJZdz0zBp+8s9N+AM6Lq+UOr5ELOCNMSuAikitPyKcbjjzbrj6H7ibKvjf+h/w+3Hr+ev7hXzl9+/xceHx1RylVHSTSJ6iLyK5wKvGmBO6eM6NwI0AWVlZM5csWdKrbdXV1eHz+Xr12o7ENFUyftvDpFauoyBhDl+v+QZ7muKZmeXksrExZMX3/+6Lvm7z8SIa2x2NbYbobHdP27xgwYI1xphZHT3W7wHf3qxZs8zq1at7ta38/Hzy8vJ69dpOhULw/v/CW/diPInsd+ewpTqGopYUVnlPpTJtBrnpPs6dMpjTxmTgcEjfbv8IItLm40A0tjsa2wzR2e6etllEOg14V18VZUsOB5z8Hcidj3zwB4bUHiDTXYKp3MTi4OvsKc/lxZL5bFlfQ0xMMeNiqwjNuI6MBf+FOPq/h6+Uim4a8N0xdAZc/BgQfsOa6mDTiwxb/Tjf2/8MIbeLfY6hFNc6mLLyR6x49yWWj7yLS06dyszhqf1aulIqekUs4EXkOSAPSBeRIuBuY8zjkdreMeXxwcxrYcY1UFOMIz6dHJcHyuv45K3fctLWh5i4azG3bP0mGVPP5vZzxpOV6CXYEqKktomsRC/OYzyco5SKPhELeGPMokite8AQgaShbXdz0nzkXHYnHLiA1Be+zjNl9/HIpk18afMVJMTFcaDGT0vIMGlIIn++eibZKXH9WLxSyu50iCYSBk3GcWM+/PtOblzzBGfHFlAQO43s+FKSAiU8XT6dC37bwO+/NpOTRqf3d7VKKZvSgI+UmDi44GEYtYBhr3yXYdUvQfIwiPVya81fyXXXcPVfgtx02ii+MX8kKfEx/V2xUspmNOAjbeJFMO48EId1VE4oBP/6by796BGyM5r52ttX8sR7hVw1dzjXnzyCQUne/q5YKWUTGvDHgrPd2+xwwDm/Bm8S81Y8wNZB2/mQKTz33nBufTeWswfVMj+lguzccThP+pZ1dq1SSvWCBnx/EIHT74LUUXg2Ps+pe/7Nqe4G67EKqCv34tzlZ9e7S/hk5gPMmTmTYWm6Q1Yp1TMa8P1p2iLrL9gMxWuhpYlg6hg+KhYOvr+EC/b8irPfu5QHV1xBwoyF3HDOPJJiO+nRh0LWrwOllArTgB8IXDEwbI51Ezg9CZhwC1QtpOn5r3Nv8ROw4Qm2bcjl08FzCMZl0BiTRkXJAQoLHiOzYjUxLfU4Ln0Sx5gz+rUpSqmBQwN+IEsehueG1+HARg588hrB9csYU/xPfOJve0qZSSQ/NJ6Rsp+Rz8fj+SUAABR0SURBVF7Oh3N+x5wvXU6MS3vzSkU7DfiBTgQGT2HQeVMYdN4d1DUFqW2uR+rL+OijDxl3ylcY09zCjsLduF6/mtkffJs71+xmyOyvcumsHHJSdexeqWilAX+c8Xlc4EmChCQciXsZGj4bdkzWCZjJb1H72AU8WHEfze/9mtr34jjg8hETn0JCSjruuGSIS4P4DPBlwvCTIHOi9SXS30q2wbIfWhdBH35Sf1ejlC1owNuIxKWSeOMyWPsMTeX72bdvPyUlB3FU1pJUdYBBnkJSTC2xwWqEEABlnhxWOOexLe0M0kbNYuqwFCYMTux8Z24k1JfD3y6Dqt3w7GVw7cvWBG9KqaOiAW833iSY9y0SgCmAMYbNxTW8sr6YN7YcpLC8HkyILCpZ4FzH+eZjLpIXWNjwPDt3D+HllpNYTjxjPZWM8lSTkZzA8JHjcKXkQNYJMGT658f1B5uh6CNweiDnxN7VG2yGv18FtQfgsqfh9R/BMxfDdcshc3xfvStKRSUNeJsTEU4YmsQJQ5O449wJNAdD7K1sYG9FA9kpCxmZHo/DXwlb/snwdc/z/aIXAAiYGA42pUNxM+xfBuEev/EkWPPj44DP3oHmOmtD486Fs34OaaO6X5wx8Nr3YM8quPhxmHghZE2CJ86Bv14EV/wNsmce+hp/NRLS6+Mq1R0a8FEmxuVgVIaPURntLgkWlwqzrsM96zqoKwXAHZ/OUGBlQRn/9a/NlBcXMtWxi/nBTZy8dQ1OB2zynsK+oSeRy37mFzyJc8cctmRdyJCpZ5I+bi6kjOh4fL+hAtY/B6v/AuU74dTbYPIlAIRSRvLWzD8zY8V1pD12OqEpV+A4826o+BQ+/BNse4257mRw3AQzF0NCVuTfNKWOUxrw6lC+jLabApw6NoP5o0/jza3jKSipoyxkeDFkqKhv4rOyej4rraeyYSyDHFO5RZ7jy/tfxnvgRfg3BB1ejCcRl9eHuDwQ9EOgERrKIRRgT/wJPBXzXYqLzuKUD/cwPC2Oh97YwZrdDYxK/B2XNPydb2x4Adn4PGJCEJsCc75J/fZVePJ/CSt+DbNvhDPuBreN5/CpLyO2YX9/VxF5xWth2a3WF/7Ys/q7GlvQgFdH5HAIZ00axFmTjvTMKygur2HFeyvYs+k9kht242tuJKHeT0qMgZhYHO5YKn0+fl82jYJALvNHp7N9Xw3Lt5QAkBYfw68vmcIlM7JZvmkWC5eezcLAayQOn8zoM65nSm4WG7zvkHdCNqz6LXzwB/j0HbjkL0ces28JgmkBl6dP3peIqi+DDX+Hba/BnveZY0Jg1sCX7oWY+P6uru9t+D94+dtWJ+Dlm+HbH1n7k/pDcwNU7ILMScf92eEa8KpPDUlL5IoLz8dccB67SuvZsr+GTcXV7Cqpp6K+ifL6ZoIthoV5Q3ly7nAyE70YY9hVWsfm4hryxmaSFGcdwXPelMHMHXkpD/x7Cv9cV0zjn9cwflACzkAj968rprz+Qs7z5fLDit8Q+6dT8U9YSHxajjVs43BBY6X1V7UXSrdDeYE17p8zh5ohJ1EZN4JhrkqkusjaydtUA021VkNy58OYL0P2LHA4j90baAysfRpe/zH4q6wd26feStHOTWR//CjsfBO+8kcYPu+Lr60rsdpasw/qS0Cc4PJaU1ePOA1ik49dO7qruQHyfwmrfgfDT4b537OOqHrrZ3Deg8eujpr9UPgubH3Zeo8DDTDqdOu9ThhkPccYqN4Lvqzjo5OABryKEBFhdKaP0Zk+Lpw6pBvPTWB0ZsIXHkvzebj/4inced4EXllfzNK1+yhvhJGZcUzNTmZn1cmct3cot4ceZdam14iVWhyYttcbpwdJGEQwfTyVg05hf1UD8ftWMaLwARLFel7QGYsjaQgObxJ4Eqxe5LsPw8r/IeiMBWcMTjGIw2XtBB46EwZNtvYlVOyCis+snc1BP7Q0Q+pIyJ4NObOtIKgvtQ4FdTitkI1NAU8iOGOsAA76rXCu3Q8f/hl2vwvDTrICLmsSB2v8PFW8kjEnXMAZ2+8h7Ymzqcw6ieQFNyOjFli9/I8ft3ZWd8YdDzOuhrnfhJTcXn2mfaqp1qr5/f+13p9ZX4dzfmXNnjr7Rut9mNpHF4UzBsp3wcGNcGCj9SUoYn0BNtVYQ0M1+6zn+gbBtCshYTCseBD+MA++/Evr8Y3/B6XbwB1nfRmNzLMO582cODC/PAExxhz5WcfIrFmzzOrVq3v12vz8fPLy8vq2oAEuGtsMX2x3S8hQUFLLO9tLeXvLfgr3FIIJUUkCcXHxxLgcHKxpanv+pCGJfHV8LEOknMc3NLOmVEiKjWFqTjJTs5NIinXz9voCkorfZaajAAchQMiIhcnOQrKbduI0QQBC7nhM6kgcsUmIy2uFRulWqNrTu8Z5k+BLP6PhhEWsKCjn+dVF5G8vIWSsTBqdCBcFXmNhy3KGSAUhceEwQWuH9oyrIXMSLQmDOdCSRLrPhccErC+P1X+BTS+ACUH2iVaPfuRpVpA5nOHrFbg+/ws2WeHnr7HmSkoeBt5kq4jGSutLrXov1B60vphq91vBWbUHAvUweJq1nSHTISkbEodY4b3rP7D9X7B9mfULZdQZcOqth/4i8dfA7+dAXBrvjLuX007v5fxKxsD25fD2L61wB6ttiUOtdoRCGFcMMnia9Uste7ZVb+uwTOkOQi9+A8eB9db9YfNg/PnW+Rq73rZ+EbZKHGp9BolDIHEwhFqs9731l5Qv0zrBMGsSjD27yy+Env5/LSJrjDGzOnxMA/74FY1thiO3u7K+mU3F1ew4WEfBwVoCLYaRGfGMyohn0pCkQ6ZvMMbw7s4yXl2/n/VFVRSU1NESMozLSmDhjKGcPj6TvZUNbCiqZmNRNZuLa6isqWGk7KfMJFFKEiDExzjJSvKSleClqjGAv2IfowPbEQzlJpF6VxJZCR4mp7YwNiFIvKMJf2MDTf4GmkIOGj3pNHoy+MwM4pMDAT4rq8cYyEr0cPGMbAY17+Pyc/PwuJw0B0O8/Ekh297+G8Nr1/JmaCY1Q05h6rBUCkpqWb+3mrqmIG6n9cto4uBERmf6mOCrY/L+f5C6fyWyf60V9j3hSbS+CPxVhy4XpzWMkTwMknKsL4TidVCypeNteJOs4a85N33xMNhWW1+Bv1+F35OBN3sKpI+BtNGQMQ7Sx1q/lCo+tf7EaT2WPgYQq5ddug3W/Q2KP7F+Uc37FmSfyLvVaSzdWM6npXV8WlZPQ1MLZ0zI5CvTh5I3LgOPy9n238VbW0v41WsbGF75PnvcI8ibM+vQi/LUFMOBTVY7S8Jf6jVF1nCPw2WFui/TCvv6Uqg7aNXtcFu9/9FnWu3JGGd90YaPONOA70A0hl00thki2+7G5hbK6prITolFOpnCoayuie0HaqlsaKbWH6S6MUBJTRMHa/wcqPGTFOsmJyWW7JQ4HA6hvilIrT/AnooGth+oZXdFA8aA2ymkxse0hXZzS4i4GCcTBicycXAiM4ancPKoNFxOR4dtDoUMa/ZUsrKgjJUFpWwurmF0ho8Zw5MZPyiRfVWNbCmuYcv+GkprP/8FkxTr5oxcDxek7GZsUgtZCTG4CFk7oFsCEApaw0feJIwngVCgESr3QNUexLTgSBtp9VaTh1nBFJcGDgctIUN5fROV9QHqm4M011fjrdzOEEclaaFynM11kHsy5Mw99CI4wM6SWl5Ys4/XNx/A7XQwKNHDRaE3GFv9PuPja3BV7LJ+GfRE0jA47TaYuohNB+q5f/k23t1ZRkqcm3GDEhiZ4cMh8K9NByirayYuxklOShxZSV4am4N8XFjJyIx4vnnaKFYWlPHqhmKcDuHMCVksnJFN3rgM3M4OdsK2Zurh//2EQtYXzpalsOVl65dAK98g+ME2EOnTgNcxeKXaiY1xHnGCtnSfh/TRvd/J1tjcQiAUIsHj6vRLpDscDuHE3FROzE3l+18a2+Vza/wBCsvqKThYxweflvPuzjL+sTUNsM6NGD8ogUSvm6ZgC83BEHVN1hdXdWOAQIsbGBX+A6/bgc/jwuMqQ6QMAH8gREV9E6EO+4sJOB2JZKfEMvKzGEZm7GBocizl9U0UVTay42AdW/fX4HQI80enE+NycKDaz88qZ1PZMB1nhTB7eArn5RrmJVWQSxFOt5fGhOHscwymuTlAYn0h8bWfEWxpocg1jILQELbWJ7B3YyN73nmPHQfrSI5zc9d5E7h63vC2njrAPRdM4t2dZeRvL6W4qpEDNX7qmoLcfcFErpo7HLfTwaWzcrj1y+N4clUhS9fuY/mmA6TGxzBxcCKZiR4yEjxU1DWzK/zLIBQypMbHkBwXQ4LXhdftxOt2Eh8Tg89zJQmTrmGwq5phoSIGBfaQ5PCTEoE5oTTglTrGYmOcxHIMj8wBEr1upmQnMyU7mYtnZmOMobC8gQ1FVWwurmFLcQ2NgRZinA7i4lxkp8SRGOsmKdZNrNuJ02HtDA+2GOqbg9Q1BfEHPj+j2ONykOGzgi413kOcx0mc22rj3spGdpfX82n4vIn3Py3HHwjhdAhDkr3kpMRx13kTuGjaUDISPv/iNMbw5Mv/ocw7lNc3H+Sut+sAwecZgdspVDY0ALvCz3bQ+gVkqSAuppphqXEMT4vngilDuOak3A7nWHI5HeSNyyRvXGaX72FOahw/Pn8it58znne2l/LKhmJ2lzfw6a46SuuaSI6LYXSGj/MmD8blECobAlTUW7/ySmub8AdaaGhuodYfpLHtvYsBRpMWH8OaCBz6rwGvVBQSEUakxzMiPZ6Lpg2N6LbmHHY/FDJUNjSTFOvG1dEQR/sak5xclzeeW788npJaPx9+WsGHn5UTMpCTEkdOaixxMU4amq3w9Lgc5KTGMSw1jrT4mKP6hdQZt9PBmROzOHPi52dRG2N6tK1gS4jKhgBldU1t4R8JGvBKqWPK4RDSfD0f4spM8HLB1CFccITDbvtDT79IXE4HGQnWL54JgyNUFNbvGqWUUjakAa+UUjalAa+UUjalAa+UUjalAa+UUjalAa+UUjalAa+UUjalAa+UUjY1oCYbE5FSYPcRn9ixdKCsD8s5HkRjmyE62x2NbYbobHdP2zzcGJPR0QMDKuCPhois7mxGNbuKxjZDdLY7GtsM0dnuvmyzDtEopZRNacArpZRN2SngH+nvAvpBNLYZorPd0dhmiM5291mbbTMGr5RS6lB26sErpZRqRwNeKaVs6rgPeBE5W0S2i8hOEbm9v+uJFBHJEZG3RWSLiGwWke+Gl6eKyBsiUhD+N6W/a+1rIuIUkbUi8mr4/ggR+TD8mf9dRGL6u8a+JiLJIvKCiGwTka0iMs/un7WIfC/83/YmEXlORLx2/KxF5C8iUiIim9ot6/CzFctvw+3fICIzerKt4zrgRcQJ/B44B5gILBKRif1bVcQEgR8YYyYCc4Fvhdt6O/CWMWYM8Fb4vt18F9ja7v6vgP9njBkNVAJf75eqIus3wL+MMeOBqVjtt+1nLSJDge8As4wxJwBO4Ars+Vk/CZx92LLOPttzgDHhvxuBP/ZkQ8d1wAOzgZ3GmE+NMc3AEuCifq4pIowx+40xn4Rv12L9Dz8Uq71PhZ/2FPCV/qkwMkQkGzgPeCx8X4DTgRfCT7Fjm5OAU4HHAYwxzcaYKmz+WWNdQjRWRFxAHLAfG37WxpgVQMVhizv7bC8C/mosHwDJItLti/wd7wE/FNjb7n5ReJmtiUguMB34EMgyxuwPP3QAyOrkZcerh4HbgFD4fhpQZYwJhu/b8TMfAZQCT4SHph4TkXhs/FkbY/YBDwJ7sIK9GliD/T/rVp19tkeVccd7wEcdEfEBLwK3GGNq2j9mrGNebXPcq4icD5QYY9b0dy3HmAuYAfzRGDMdqOew4RgbftYpWL3VEcAQIJ4vDmNEhb78bI/3gN8H5LS7nx1eZksi4sYK92eNMf8ILz7Y+pMt/G9Jf9UXAScDF4pIIdbw2+lYY9PJ4Z/xYM/PvAgoMsZ8GL7/Albg2/mzPhP4zBhTaowJAP/A+vzt/lm36uyzPaqMO94D/mNgTHhPewzWTpmX+7mmiAiPPT8ObDXGPNTuoZeBa8O3rwX+eaxrixRjzB3GmGxjTC7WZ/sfY8zXgLeBS8JPs1WbAYwxB4C9IjIuvOgMYAs2/qyxhmbmikhc+L/11jbb+rNup7PP9mXgmvDRNHOB6nZDOUdmjDmu/4BzgR3ALuBH/V1PBNs5H+tn2wZgXfjvXKwx6beAAuBNILW/a41Q+/OAV8O3RwIfATuB/wM8/V1fBNo7DVgd/ryXAil2/6yBe4FtwCbgacBjx88aeA5rP0MA69fa1zv7bAHBOlJwF7AR6yijbm9LpypQSimbOt6HaJRSSnVCA14ppWxKA14ppWxKA14ppWxKA14ppWxKA15FBRFpEZF17f5uDy/PD89Gul5E3ms99lxEYkTk4fAsfgUi8s/wvDit6xskIktEZJeIrBGRZSIyVkRy288SGH7uPSLyw2PbYqWsU6KVigaNxphpnTz2NWPMahG5EXgAuBD4JZAAjDPGtIjIdcA/RGRO+DUvAU8ZY64AEJGpWPOH7P3i6pXqHxrwSn1uBXCLiMQB1wEjjDEtAMaYJ0TkeqzpEgwQMMb8qfWFxpj10DYRnFIDgga8ihaxIrKu3f37jDF/P+w5F2CdLTga2GMOm8wN68zSSeHbXU2ANuqwbQ3CmilRqWNKA15Fi66GaJ4VkUagELgZa1qAo7Gr/bZE5J6jXJ9SvaIBr1R4DL71johUAMNEJMFYF1dpNRN4NXz7EpQa4PQoGqUOY4ypx7qqzkPhy0IiItdgXWXoP+E/T3inLOHHp4jIKf1Rr1Kd0YBX0SL2sMMk7z/C8+8A/MAOESkALgW+asKArwJnhg+T3Azch3UlHqUGDJ1NUimlbEp78EopZVMa8EopZVMa8EopZVMa8EopZVMa8EopZVMa8EopZVMa8EopZVP/H0zmp5cD3Jz2AAAAAElFTkSuQmCC",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "plt.subplots(1)\n",
        "plt.plot(history.history['accuracy'])\n",
        "plt.plot(history.history['val_accuracy'])\n",
        "plt.title(\"Accuracy and Val. Accuracy\")\n",
        "plt.grid(b=True)\n",
        "plt.legend([\"Traning\", \"Validation\"])\n",
        "plt.ylabel(\"Accuracy (%)\")\n",
        "plt.xlabel(\"EPOCH\")\n",
        "\n",
        "\n",
        "plt.subplots(1)\n",
        "plt.plot(history.history['loss'])\n",
        "plt.plot(history.history['val_loss'])\n",
        "# plt.yscale('log')\n",
        "plt.legend([\"Traning\", \"Validation\"])\n",
        "plt.title(\"Loss and Val. Loss\")\n",
        "plt.ylabel(\"SCCE Loss\")\n",
        "plt.xlabel(\"EPOCH\")\n",
        "plt.grid(b=True, which='both')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "INFO:tensorflow:Assets written to: /content/drive/My Drive/3020_drive/Models/9000_1000_epocs_8_layers/assets\n"
          ]
        }
      ],
      "source": [
        "#@markdown\n",
        "# model.save('/content/drive/My Drive/3020_drive/Models/9000_1000_epocs_8_layers/')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "#@markdown\n",
        "import os\n",
        "\n",
        "test_dir = '/content/drive/My Drive/3020_drive/Colab Notebooks/Data/Test_set/'\n",
        "# test_dir = \"/content/drive/My Drive/3020_drive/Data/Devset_13/\"\n",
        "\n",
        "\n",
        "print('\\tFILE:\\t |\\tPREDICTIONS\\t |\\t' + str(class_names) )\n",
        "print('-----------------------------------------------------------------------------------------------------------------------')\n",
        "for folder in os.listdir(test_dir):\n",
        "    for file in os.listdir(test_dir + folder):\n",
        "        img = tf.keras.preprocessing.image.load_img(\n",
        "            f'{test_dir}{folder}/{file}',\n",
        "            target_size=(img_width, img_height),\n",
        "            color_mode='grayscale')\n",
        "        x = tf.keras.preprocessing.image.img_to_array(img)\n",
        "        x = np.expand_dims(x, axis=0)\n",
        "\n",
        "        try:\n",
        "            prediction = model.predict(x)\n",
        "            classes = np.argmax(prediction, axis=-1)\n",
        "\n",
        "            print(file + '\\t |\\t ' + class_names[int(classes)] + '\\t |\\t ' + str(prediction))\n",
        "            print('\\t\\t |\\t\\t\\t |')\n",
        "        except:\n",
        "            pass\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "model = tf.keras.models.load_model('/content/drive/My Drive/3020_drive/Models/9000_1000_epocs_8_layers/')"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "Model: \"sequential_15\"\n",
            "_________________________________________________________________\n",
            "Layer (type)                 Output Shape              Param #   \n",
            "=================================================================\n",
            "rescaling_18 (Rescaling)     (None, 128, 128, 1)       0         \n",
            "_________________________________________________________________\n",
            "conv2d_59 (Conv2D)           (None, 123, 123, 32)      1184      \n",
            "_________________________________________________________________\n",
            "max_pooling2d_59 (MaxPooling (None, 61, 61, 32)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_60 (Conv2D)           (None, 57, 57, 64)        51264     \n",
            "_________________________________________________________________\n",
            "max_pooling2d_60 (MaxPooling (None, 28, 28, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_61 (Conv2D)           (None, 24, 24, 64)        102464    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_61 (MaxPooling (None, 12, 12, 64)        0         \n",
            "_________________________________________________________________\n",
            "conv2d_62 (Conv2D)           (None, 9, 9, 128)         131200    \n",
            "_________________________________________________________________\n",
            "max_pooling2d_62 (MaxPooling (None, 4, 4, 128)         0         \n",
            "_________________________________________________________________\n",
            "flatten_15 (Flatten)         (None, 2048)              0         \n",
            "_________________________________________________________________\n",
            "dense_30 (Dense)             (None, 128)               262272    \n",
            "_________________________________________________________________\n",
            "dense_31 (Dense)             (None, 5)                 645       \n",
            "=================================================================\n",
            "Total params: 549,029\n",
            "Trainable params: 549,029\n",
            "Non-trainable params: 0\n",
            "_________________________________________________________________\n",
            "4/4 [==============================] - 3s 52ms/step - loss: 0.2016 - accuracy: 0.9649\n"
          ]
        },
        {
          "data": {
            "text/plain": [
              "[0.20160211622714996, 0.9649398326873779]"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "from sklearn import metrics\n",
        "\n",
        "model.summary()\n",
        "model.evaluate(val_ds,  batch_size=512, verbose=1)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "import itertools\n",
        "\n",
        "def plot_confusion_matrix(cm, classes,\n",
        "                        normalize=False,\n",
        "                        title='Confusion matrix',\n",
        "                        cmap=plt.cm.Blues):\n",
        "    \"\"\"\n",
        "    This function prints and plots the confusion matrix.\n",
        "    Normalization can be applied by setting `normalize=True`.\n",
        "    \"\"\"\n",
        "    plt.imshow(cm, interpolation='nearest', cmap=cmap)\n",
        "    plt.title(title)\n",
        "    plt.colorbar()\n",
        "    tick_marks = np.arange(len(classes))\n",
        "    plt.xticks(tick_marks, classes, rotation=45)\n",
        "    plt.yticks(tick_marks, classes)\n",
        "\n",
        "    if normalize:\n",
        "        cm = cm.astype('float') / cm.sum(axis=1)[:, np.newaxis]\n",
        "        print(\"Normalized confusion matrix\")\n",
        "    else:\n",
        "        print('Confusion matrix, without normalization')\n",
        "\n",
        "    thresh = cm.max() / 2.\n",
        "    for i, j in itertools.product(range(cm.shape[0]), range(cm.shape[1])):\n",
        "        plt.text(j, i, cm[i, j],\n",
        "            horizontalalignment=\"center\",\n",
        "            color=\"white\" if cm[i, j] > thresh else \"black\")\n",
        "\n",
        "    plt.tight_layout()\n",
        "    plt.ylabel('True label')\n",
        "    plt.xlabel('Predicted label')\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "predictions = np.array([])\n",
        "labels =  np.array([])\n",
        "for x, y in val_ds:\n",
        "    predictions = np.concatenate([predictions, np.argmax(model.predict(x), axis=-1)])\n",
        "    labels = np.concatenate([labels, np.array(y)])\n",
        "\n",
        "cm = tf.math.confusion_matrix(labels=labels, predictions=predictions).numpy()"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "ename": "NameError",
          "evalue": "ignored",
          "output_type": "error",
          "traceback": [
            "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m                                 Traceback (most recent call last)\n",
            "\u001b[0;32m<ipython-input-4-a2821ac77226>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n",
            "\u001b[1;32m      1\u001b[0m \u001b[0mclass_names\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;34m[\u001b[0m\u001b[0;34m'diode'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'resistor'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'inductor'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'capacitor'\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m'dc source'\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[1;32m      2\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0;32m----> 3\u001b[0;31m \u001b[0mplot_confusion_matrix\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mcm\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mcm\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mclasses\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mclass_names\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtitle\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0;34m'Confusion Matrix'\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
            "\u001b[0m\n",
            "\u001b[0;31mNameError\u001b[0m: name 'plot_confusion_matrix' is not defined"
          ]
        }
      ],
      "source": [
        "class_names = ['diode', 'resistor', 'inductor', 'capacitor', 'dc source']\n",
        "\n",
        "plot_confusion_matrix(cm=cm, classes=class_names, title='Confusion Matrix',)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "name": "stdout",
          "output_type": "stream",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "         0.0       0.99      0.99      0.99       377\n",
            "         1.0       0.97      0.90      0.93       407\n",
            "         2.0       0.92      0.97      0.95       393\n",
            "         3.0       0.98      0.97      0.98       366\n",
            "         4.0       0.97      0.99      0.98       368\n",
            "\n",
            "    accuracy                           0.96      1911\n",
            "   macro avg       0.97      0.97      0.97      1911\n",
            "weighted avg       0.97      0.96      0.96      1911\n",
            "\n"
          ]
        }
      ],
      "source": [
        "print(metrics.classification_report(y_true=labels, y_pred=predictions))"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "# metrics for previous work done\n",
        "class_names = ['diode', 'resistor', 'inductor', 'capacitor', 'dc source']\n",
        "f1 = lambda precision, recall: round(2*(precision*recall)/(precision+recall),2)"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "proposed = [0.99, 0.93,0.95,0.98,0.98,]\n",
        "gunay = [0, f1(0.81,0.72),f1(1,0.81),f1(0.65,0.81),f1(1, 0.87)]\n",
        "naika = [0.98, 0.99, 0, 0.99, 0.99]\n",
        "aksakalli = [1, f1(0.831,0.92),f1(0.91,0.812),0,0]"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [
        {
          "data": {
            "text/plain": [
              "<matplotlib.legend.Legend at 0x7f831a8a9310>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        },
        {
          "data": {
            "image/png": "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",
            "text/plain": [
              "<Figure size 648x360 with 1 Axes>"
            ]
          },
          "metadata": {},
          "output_type": "display_data"
        }
      ],
      "source": [
        "import pandas as pd\n",
        "\n",
        "df = pd.DataFrame({\n",
        "                   'Proposed Method': proposed,\n",
        "                   'Gunay et al.(2020)': gunay,\n",
        "                   'Naika et al. (2019)': naika,\n",
        "                   'Aksakalli (2018)': aksakalli\n",
        "                   },index=class_names)\n",
        "\n",
        "ax = df.plot.bar(rot=45,width=0.4,figsize=(9,5))\n",
        "ax.set_axisbelow(True)\n",
        "plt.grid(True)\n",
        "plt.ylim(0.65)\n",
        "plt.yscale(\"linear\")\n",
        "plt.ylabel(\"F1 Score\")\n",
        "plt.legend(loc=\"upper center\", ncol=(len(df.columns)),)\n"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "authorship_tag": "ABX9TyPtVBXCJXDDzygkQJ9/dGZR",
      "include_colab_link": true,
      "mount_file_id": "1P0IZhL9bRcth587Tak_F9jVDcLWUcSG-",
      "name": "neural-network.ipynb",
      "provenance": []
    },
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}