Train_CNN_Analog-Readout_Version-Small1-very-long.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN Training\n",
    "\n",
    "Target of this code is to train a CNN network to extract the needle position of an analog needle device.\n",
    "\n",
    "### Preparing the training\n",
    "* First all libraries are loaded\n",
    "    * It is assumed, that they are installed during the Python setup\n",
    "* matplotlib is set to print the output inline in the jupyter notebook"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "########### Basic Parameters for Running: ################################\n",
    "    \n",
    "TFliteNamingAndVersion = \"ana0910s1_long\"   # Used for tflite Filename\n",
    "Training_Percentage = 0.2              # 0.0 = Use all Images for Training\n",
    "Epoch_Anz = 100\n",
    "\n",
    "##########################################################################\n",
    "\n",
    "\n",
    "\n",
    "import os\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "import glob\n",
    "import numpy as np\n",
    "from sklearn.utils import shuffle\n",
    "from tensorflow import keras\n",
    "from keras import Sequential\n",
    "from keras.layers import Dense, InputLayer, Conv2D, MaxPool2D, Flatten, BatchNormalization\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from sklearn.model_selection import train_test_split\n",
    "import keras.backend as K\n",
    "from keras.callbacks import History \n",
    "import math\n",
    "from PIL import Image \n",
    "\n",
    "loss_ges = np.array([])\n",
    "val_loss_ges = np.array([])\n",
    "\n",
    "%matplotlib inline\n",
    "np.set_printoptions(precision=4)\n",
    "np.set_printoptions(suppress=True)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Load training data\n",
    "* The data is expected in the \"Input_dir\"\n",
    "* Picture size must be 32x32 with 3 color channels (RGB)\n",
    "* The filename contains the informations needed for training in the first 3 digits::\n",
    "* Typical filename: \n",
    "    * x.y-zzzz.jpg \n",
    "    * e.g. \"4.6_Lfd-1406_zeiger3_2019-06-02T050011\"\n",
    "\n",
    "|Place holder | Meaning                     | Usage        |\n",
    "|------------- |-----------------------------|--------------|\n",
    "| **x.y**          | readout value               | **to be learned** |\n",
    "| zzzz        | additional information              | not needed   |\n",
    "\n",
    "* The images are stored in the x_data[]\n",
    "* The expected output for each image in the corresponding y_data[]\n",
    "    * The periodic nature is reflected in a **sin/cos coding**, which allows to restore the angle/counter value with an arctan later on.\n",
    "\n",
    "* The last step is a shuffle (from sklearn.utils) as the filenames are on order due to the encoding of the expected analog readout in the filename "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(5761, 32, 32, 3)\n",
      "(5761, 2)\n"
     ]
    }
   ],
   "source": [
    "Input_dir='data_resize_all'\n",
    "\n",
    "files = glob.glob(Input_dir + '/*.*')\n",
    "x_data = []\n",
    "y_data = []\n",
    "\n",
    "for aktfile in files:\n",
    "    test_image = Image.open(aktfile)\n",
    "    test_image = np.array(test_image, dtype=\"float32\")\n",
    "    test_image = np.reshape(test_image, (32,32,3))\n",
    "    base = os.path.basename(aktfile)\n",
    "    target_number = (float(base[0:3])) / 10\n",
    "    target_sin = math.sin(target_number * math.pi * 2)\n",
    "    target_cos = math.cos(target_number * math.pi * 2)\n",
    "\n",
    "    x_data.append(test_image)\n",
    "    zw = np.array([target_sin, target_cos])\n",
    "    y_data.append(zw)\n",
    "\n",
    "x_data = np.array(x_data)\n",
    "y_data = np.array(y_data)\n",
    "print(x_data.shape)\n",
    "print(y_data.shape)\n",
    "\n",
    "x_data, y_data = shuffle(x_data, y_data)\n",
    "\n",
    "if (Training_Percentage > 0):\n",
    "    X_train, X_test, y_train, y_test = train_test_split(x_data, y_data, test_size=Training_Percentage)\n",
    "else:\n",
    "    X_train = x_data\n",
    "    y_train = y_data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Define the model\n",
    "\n",
    "The layout of the network ist a typcial CNN network with alternating **Conv2D** and **MaxPool2D** layers. Finished after **flattening** with additional **Dense** layer.\n",
    "\n",
    "#### Important\n",
    "* Shape of the input layer: (32, 32, 3)\n",
    "* Shape of the output layer: (2) - sin and cos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " batch_normalization (BatchN  (None, 32, 32, 3)        12        \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 32, 32, 32)        2432      \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 8, 8, 32)         0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 8, 8, 32)          25632     \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 2, 2, 32)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 2, 2, 32)          9248      \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 1, 1, 32)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 32)                0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 128)               4224      \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 64)                8256      \n",
      "                                                                 \n",
      " dense_2 (Dense)             (None, 2)                 130       \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 49,934\n",
      "Trainable params: 49,928\n",
      "Non-trainable params: 6\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = Sequential()\n",
    "model.add(BatchNormalization(input_shape=(32,32,3)))\n",
    "model.add(Conv2D(32, (5, 5), input_shape=(32,32,3), padding='same', activation=\"relu\"))\n",
    "model.add(MaxPool2D(pool_size=(4,4)))\n",
    "model.add(Conv2D(32, (5, 5), padding='same', activation=\"relu\"))\n",
    "model.add(MaxPool2D(pool_size=(4,4)))\n",
    "model.add(Conv2D(32, (3, 3), padding='same', activation=\"relu\"))\n",
    "model.add(MaxPool2D(pool_size=(2,2)))\n",
    "model.add(Flatten())\n",
    "model.add(Dense(128,activation=\"relu\"))\n",
    "model.add(Dense(64,activation=\"relu\"))\n",
    "model.add(Dense(2))\n",
    "model.summary()\n",
    "\n",
    "model.compile(loss=keras.losses.mean_squared_error, optimizer=tf.keras.optimizers.Adadelta(learning_rate=1.0, rho=0.95), metrics = [\"accuracy\"])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Training\n",
    "The input pictures are randomly scattered for brightness and pixel shift variations. These is implemented with a ImageDataGenerator.\n",
    "\n",
    "\n",
    "The training is splitted into two steps:\n",
    "1. Variation of the brightness only\n",
    "2. Variation of brightness and Pixel Shift"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 1: Brigthness scattering only"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/100\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/ipykernel_84415/65716224.py:11: UserWarning: `Model.fit_generator` is deprecated and will be removed in a future version. Please use `Model.fit`, which supports generators.\n",
      "  history = model.fit_generator(train_iterator, validation_data = validation_iterator, epochs = Epoch_Anz)\n",
      "2022-07-15 20:27:24.668376: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "721/721 [==============================] - 3s 4ms/step - loss: 0.1653 - accuracy: 0.8474 - val_loss: 0.0191 - val_accuracy: 0.9705\n",
      "Epoch 2/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0179 - accuracy: 0.9660 - val_loss: 0.0119 - val_accuracy: 0.9792\n",
      "Epoch 3/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0087 - accuracy: 0.9773 - val_loss: 0.0077 - val_accuracy: 0.9714\n",
      "Epoch 4/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0059 - accuracy: 0.9797 - val_loss: 0.0032 - val_accuracy: 0.9896\n",
      "Epoch 5/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0040 - accuracy: 0.9847 - val_loss: 0.0055 - val_accuracy: 0.9931\n",
      "Epoch 6/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0032 - accuracy: 0.9839 - val_loss: 0.0032 - val_accuracy: 0.9887\n",
      "Epoch 7/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0026 - accuracy: 0.9892 - val_loss: 0.0030 - val_accuracy: 0.9748\n",
      "Epoch 8/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0022 - accuracy: 0.9887 - val_loss: 0.0019 - val_accuracy: 0.9853\n",
      "Epoch 9/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0019 - accuracy: 0.9882 - val_loss: 0.0019 - val_accuracy: 0.9913\n",
      "Epoch 10/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0017 - accuracy: 0.9896 - val_loss: 0.0022 - val_accuracy: 0.9983\n",
      "Epoch 11/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0015 - accuracy: 0.9932 - val_loss: 0.0017 - val_accuracy: 0.9931\n",
      "Epoch 12/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0014 - accuracy: 0.9905 - val_loss: 0.0010 - val_accuracy: 0.9931\n",
      "Epoch 13/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0012 - accuracy: 0.9896 - val_loss: 9.7861e-04 - val_accuracy: 0.9879\n",
      "Epoch 14/100\n",
      "721/721 [==============================] - 3s 4ms/step - loss: 0.0011 - accuracy: 0.9903 - val_loss: 0.0011 - val_accuracy: 0.9948\n",
      "Epoch 15/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0011 - accuracy: 0.9915 - val_loss: 8.5100e-04 - val_accuracy: 0.9983\n",
      "Epoch 16/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 9.7015e-04 - accuracy: 0.9925 - val_loss: 8.8702e-04 - val_accuracy: 0.9965\n",
      "Epoch 17/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 9.3515e-04 - accuracy: 0.9932 - val_loss: 7.8986e-04 - val_accuracy: 0.9905\n",
      "Epoch 18/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 8.6538e-04 - accuracy: 0.9934 - val_loss: 6.0358e-04 - val_accuracy: 0.9965\n",
      "Epoch 19/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 8.4921e-04 - accuracy: 0.9932 - val_loss: 6.2831e-04 - val_accuracy: 0.9965\n",
      "Epoch 20/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 7.4088e-04 - accuracy: 0.9927 - val_loss: 8.2021e-04 - val_accuracy: 0.9775\n",
      "Epoch 21/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 7.1562e-04 - accuracy: 0.9934 - val_loss: 8.6123e-04 - val_accuracy: 0.9931\n",
      "Epoch 22/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 6.9945e-04 - accuracy: 0.9936 - val_loss: 9.5598e-04 - val_accuracy: 0.9974\n",
      "Epoch 23/100\n",
      "721/721 [==============================] - 8s 11ms/step - loss: 6.4982e-04 - accuracy: 0.9936 - val_loss: 6.1769e-04 - val_accuracy: 0.9957\n",
      "Epoch 24/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 6.4566e-04 - accuracy: 0.9939 - val_loss: 6.5928e-04 - val_accuracy: 0.9957\n",
      "Epoch 25/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 5.9244e-04 - accuracy: 0.9957 - val_loss: 4.8299e-04 - val_accuracy: 0.9957\n",
      "Epoch 26/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 5.8317e-04 - accuracy: 0.9951 - val_loss: 6.0501e-04 - val_accuracy: 1.0000\n",
      "Epoch 27/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.7585e-04 - accuracy: 0.9943 - val_loss: 6.6379e-04 - val_accuracy: 0.9905\n",
      "Epoch 28/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 5.5514e-04 - accuracy: 0.9953 - val_loss: 4.7728e-04 - val_accuracy: 0.9939\n",
      "Epoch 29/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 5.4949e-04 - accuracy: 0.9950 - val_loss: 4.5739e-04 - val_accuracy: 0.9965\n",
      "Epoch 30/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 5.1545e-04 - accuracy: 0.9969 - val_loss: 4.0397e-04 - val_accuracy: 1.0000\n",
      "Epoch 31/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.8464e-04 - accuracy: 0.9964 - val_loss: 4.2719e-04 - val_accuracy: 1.0000\n",
      "Epoch 32/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.9806e-04 - accuracy: 0.9951 - val_loss: 4.3515e-04 - val_accuracy: 0.9896\n",
      "Epoch 33/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.6516e-04 - accuracy: 0.9951 - val_loss: 4.1623e-04 - val_accuracy: 0.9965\n",
      "Epoch 34/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 4.5973e-04 - accuracy: 0.9955 - val_loss: 5.8366e-04 - val_accuracy: 0.9870\n",
      "Epoch 35/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.3946e-04 - accuracy: 0.9951 - val_loss: 4.1379e-04 - val_accuracy: 0.9939\n",
      "Epoch 36/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.2917e-04 - accuracy: 0.9958 - val_loss: 4.8120e-04 - val_accuracy: 0.9922\n",
      "Epoch 37/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.2447e-04 - accuracy: 0.9964 - val_loss: 4.9930e-04 - val_accuracy: 0.9922\n",
      "Epoch 38/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0227e-04 - accuracy: 0.9967 - val_loss: 4.5200e-04 - val_accuracy: 0.9983\n",
      "Epoch 39/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0334e-04 - accuracy: 0.9981 - val_loss: 3.3120e-04 - val_accuracy: 0.9870\n",
      "Epoch 40/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.9910e-04 - accuracy: 0.9965 - val_loss: 2.6507e-04 - val_accuracy: 0.9965\n",
      "Epoch 41/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.9317e-04 - accuracy: 0.9969 - val_loss: 2.9854e-04 - val_accuracy: 1.0000\n",
      "Epoch 42/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0164e-04 - accuracy: 0.9958 - val_loss: 4.0828e-04 - val_accuracy: 0.9948\n",
      "Epoch 43/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.7179e-04 - accuracy: 0.9965 - val_loss: 3.9169e-04 - val_accuracy: 0.9974\n",
      "Epoch 44/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.8270e-04 - accuracy: 0.9964 - val_loss: 3.9406e-04 - val_accuracy: 1.0000\n",
      "Epoch 45/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.7649e-04 - accuracy: 0.9960 - val_loss: 2.9570e-04 - val_accuracy: 0.9974\n",
      "Epoch 46/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.5854e-04 - accuracy: 0.9960 - val_loss: 4.1451e-04 - val_accuracy: 0.9974\n",
      "Epoch 47/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.4766e-04 - accuracy: 0.9974 - val_loss: 3.0112e-04 - val_accuracy: 0.9974\n",
      "Epoch 48/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.5750e-04 - accuracy: 0.9972 - val_loss: 2.9534e-04 - val_accuracy: 0.9983\n",
      "Epoch 49/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.3317e-04 - accuracy: 0.9970 - val_loss: 3.4168e-04 - val_accuracy: 0.9965\n",
      "Epoch 50/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.2510e-04 - accuracy: 0.9974 - val_loss: 3.2234e-04 - val_accuracy: 0.9948\n",
      "Epoch 51/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.3471e-04 - accuracy: 0.9974 - val_loss: 2.6108e-04 - val_accuracy: 0.9983\n",
      "Epoch 52/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.3474e-04 - accuracy: 0.9974 - val_loss: 2.4432e-04 - val_accuracy: 0.9974\n",
      "Epoch 53/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.1436e-04 - accuracy: 0.9979 - val_loss: 2.4262e-04 - val_accuracy: 0.9974\n",
      "Epoch 54/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.1263e-04 - accuracy: 0.9979 - val_loss: 2.9653e-04 - val_accuracy: 0.9974\n",
      "Epoch 55/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.0626e-04 - accuracy: 0.9967 - val_loss: 2.3528e-04 - val_accuracy: 0.9983\n",
      "Epoch 56/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.2474e-04 - accuracy: 0.9976 - val_loss: 2.8379e-04 - val_accuracy: 0.9957\n",
      "Epoch 57/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.9510e-04 - accuracy: 0.9972 - val_loss: 2.7009e-04 - val_accuracy: 0.9991\n",
      "Epoch 58/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.0450e-04 - accuracy: 0.9974 - val_loss: 4.0689e-04 - val_accuracy: 0.9957\n",
      "Epoch 59/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.0401e-04 - accuracy: 0.9979 - val_loss: 3.4688e-04 - val_accuracy: 0.9983\n",
      "Epoch 60/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8844e-04 - accuracy: 0.9977 - val_loss: 2.7699e-04 - val_accuracy: 0.9991\n",
      "Epoch 61/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8801e-04 - accuracy: 0.9981 - val_loss: 2.3533e-04 - val_accuracy: 1.0000\n",
      "Epoch 62/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8216e-04 - accuracy: 0.9976 - val_loss: 2.8513e-04 - val_accuracy: 0.9957\n",
      "Epoch 63/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8622e-04 - accuracy: 0.9974 - val_loss: 2.5090e-04 - val_accuracy: 0.9957\n",
      "Epoch 64/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.9022e-04 - accuracy: 0.9974 - val_loss: 2.0969e-04 - val_accuracy: 0.9983\n",
      "Epoch 65/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8334e-04 - accuracy: 0.9970 - val_loss: 2.3426e-04 - val_accuracy: 0.9991\n",
      "Epoch 66/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.8067e-04 - accuracy: 0.9976 - val_loss: 2.9741e-04 - val_accuracy: 0.9991\n",
      "Epoch 67/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.7888e-04 - accuracy: 0.9981 - val_loss: 4.0404e-04 - val_accuracy: 0.9991\n",
      "Epoch 68/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.6710e-04 - accuracy: 0.9979 - val_loss: 2.0654e-04 - val_accuracy: 0.9965\n",
      "Epoch 69/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.5976e-04 - accuracy: 0.9988 - val_loss: 1.7365e-04 - val_accuracy: 1.0000\n",
      "Epoch 70/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.5763e-04 - accuracy: 0.9984 - val_loss: 1.5744e-04 - val_accuracy: 1.0000\n",
      "Epoch 71/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.5842e-04 - accuracy: 0.9981 - val_loss: 2.5499e-04 - val_accuracy: 1.0000\n",
      "Epoch 72/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.5399e-04 - accuracy: 0.9979 - val_loss: 2.1241e-04 - val_accuracy: 1.0000\n",
      "Epoch 73/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.5139e-04 - accuracy: 0.9977 - val_loss: 2.4198e-04 - val_accuracy: 0.9991\n",
      "Epoch 74/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.4725e-04 - accuracy: 0.9970 - val_loss: 2.3294e-04 - val_accuracy: 1.0000\n",
      "Epoch 75/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.4751e-04 - accuracy: 0.9983 - val_loss: 2.1376e-04 - val_accuracy: 1.0000\n",
      "Epoch 76/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.6807e-04 - accuracy: 0.9974 - val_loss: 2.2357e-04 - val_accuracy: 1.0000\n",
      "Epoch 77/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.3914e-04 - accuracy: 0.9972 - val_loss: 1.7993e-04 - val_accuracy: 1.0000\n",
      "Epoch 78/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 2.4396e-04 - accuracy: 0.9986 - val_loss: 3.2215e-04 - val_accuracy: 0.9974\n",
      "Epoch 79/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 2.4879e-04 - accuracy: 0.9974 - val_loss: 1.7600e-04 - val_accuracy: 1.0000\n",
      "Epoch 80/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.4385e-04 - accuracy: 0.9979 - val_loss: 3.0405e-04 - val_accuracy: 0.9965\n",
      "Epoch 81/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.2670e-04 - accuracy: 0.9984 - val_loss: 3.4274e-04 - val_accuracy: 1.0000\n",
      "Epoch 82/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.3691e-04 - accuracy: 0.9983 - val_loss: 1.8461e-04 - val_accuracy: 0.9991\n",
      "Epoch 83/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.3878e-04 - accuracy: 0.9979 - val_loss: 2.1786e-04 - val_accuracy: 0.9983\n",
      "Epoch 84/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.3516e-04 - accuracy: 0.9981 - val_loss: 2.7165e-04 - val_accuracy: 0.9991\n",
      "Epoch 85/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.2439e-04 - accuracy: 0.9974 - val_loss: 2.0529e-04 - val_accuracy: 0.9974\n",
      "Epoch 86/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.3136e-04 - accuracy: 0.9983 - val_loss: 1.3773e-04 - val_accuracy: 1.0000\n",
      "Epoch 87/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1856e-04 - accuracy: 0.9984 - val_loss: 1.9887e-04 - val_accuracy: 1.0000\n",
      "Epoch 88/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1963e-04 - accuracy: 0.9979 - val_loss: 1.7239e-04 - val_accuracy: 0.9991\n",
      "Epoch 89/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1522e-04 - accuracy: 0.9993 - val_loss: 2.2618e-04 - val_accuracy: 1.0000\n",
      "Epoch 90/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1424e-04 - accuracy: 0.9986 - val_loss: 2.4754e-04 - val_accuracy: 0.9983\n",
      "Epoch 91/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1681e-04 - accuracy: 0.9984 - val_loss: 1.7167e-04 - val_accuracy: 1.0000\n",
      "Epoch 92/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1343e-04 - accuracy: 0.9984 - val_loss: 3.0368e-04 - val_accuracy: 0.9991\n",
      "Epoch 93/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1274e-04 - accuracy: 0.9995 - val_loss: 2.5911e-04 - val_accuracy: 0.9991\n",
      "Epoch 94/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1990e-04 - accuracy: 0.9983 - val_loss: 2.9019e-04 - val_accuracy: 0.9905\n",
      "Epoch 95/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.0586e-04 - accuracy: 0.9981 - val_loss: 2.1886e-04 - val_accuracy: 0.9983\n",
      "Epoch 96/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.1379e-04 - accuracy: 0.9979 - val_loss: 1.8157e-04 - val_accuracy: 0.9991\n",
      "Epoch 97/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 2.0444e-04 - accuracy: 0.9983 - val_loss: 1.6057e-04 - val_accuracy: 0.9983\n",
      "Epoch 98/100\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 2.0931e-04 - accuracy: 0.9977 - val_loss: 2.3336e-04 - val_accuracy: 1.0000\n",
      "Epoch 99/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 2.0991e-04 - accuracy: 0.9981 - val_loss: 2.1578e-04 - val_accuracy: 1.0000\n",
      "Epoch 100/100\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 2.0505e-04 - accuracy: 0.9986 - val_loss: 1.9647e-04 - val_accuracy: 0.9983\n"
     ]
    }
   ],
   "source": [
    "Batch_Size = 8\n",
    "Epoch_Anz = 100\n",
    "Shift_Range = 0\n",
    "Brightness_Range = 0.3\n",
    "\n",
    "datagen = ImageDataGenerator(width_shift_range=[-Shift_Range,Shift_Range], height_shift_range=[-Shift_Range,Shift_Range],brightness_range=[1-Brightness_Range,1+Brightness_Range])\n",
    "\n",
    "if (Training_Percentage > 0):\n",
    "    train_iterator = datagen.flow(x_data, y_data, batch_size=Batch_Size)\n",
    "    validation_iterator = datagen.flow(X_test, y_test, batch_size=Batch_Size)\n",
    "    history = model.fit_generator(train_iterator, validation_data = validation_iterator, epochs = Epoch_Anz)\n",
    "else:\n",
    "    train_iterator = datagen.flow(x_data, y_data, batch_size=Batch_Size)\n",
    "    history = model.fit_generator(train_iterator, epochs = Epoch_Anz)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 1: Learing result\n",
    " \n",
    "* Visualization of the training and validation results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss_ges = np.append(loss_ges, history.history['loss'])\n",
    "\n",
    "if (Training_Percentage > 0):\n",
    "    val_loss_ges = np.append(val_loss_ges, history.history['val_loss'])\n",
    "    plt.semilogy(val_loss_ges)\n",
    "\n",
    "plt.semilogy(history.history['loss'])\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train','eval'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Step 2: Brigthness and Pixel Shift scattering\n",
    "Here a higher number of epochs is used to reach the minimum loss function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/400\n",
      " 36/721 [>.............................] - ETA: 3s - loss: 0.1832 - accuracy: 0.8611"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/ipykernel_84415/3361052463.py:11: UserWarning: `Model.fit_generator` is deprecated and will be removed in a future version. Please use `Model.fit`, which supports generators.\n",
      "  history = model.fit_generator(train_iterator, validation_data = validation_iterator, epochs = Epoch_Anz)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0348 - accuracy: 0.9472 - val_loss: 0.0075 - val_accuracy: 0.9792\n",
      "Epoch 2/400\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0077 - accuracy: 0.9759 - val_loss: 0.0035 - val_accuracy: 0.9818\n",
      "Epoch 3/400\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0047 - accuracy: 0.9780 - val_loss: 0.0029 - val_accuracy: 0.9827\n",
      "Epoch 4/400\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0033 - accuracy: 0.9799 - val_loss: 0.0029 - val_accuracy: 0.9896\n",
      "Epoch 5/400\n",
      "721/721 [==============================] - 3s 5ms/step - loss: 0.0030 - accuracy: 0.9868 - val_loss: 0.0025 - val_accuracy: 0.9887\n",
      "Epoch 6/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0023 - accuracy: 0.9856 - val_loss: 0.0025 - val_accuracy: 0.9913\n",
      "Epoch 7/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0021 - accuracy: 0.9861 - val_loss: 0.0026 - val_accuracy: 0.9896\n",
      "Epoch 8/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0020 - accuracy: 0.9889 - val_loss: 0.0023 - val_accuracy: 0.9931\n",
      "Epoch 9/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0018 - accuracy: 0.9892 - val_loss: 0.0018 - val_accuracy: 0.9913\n",
      "Epoch 10/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0016 - accuracy: 0.9894 - val_loss: 0.0016 - val_accuracy: 0.9887\n",
      "Epoch 11/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0016 - accuracy: 0.9896 - val_loss: 0.0012 - val_accuracy: 0.9922\n",
      "Epoch 12/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0014 - accuracy: 0.9906 - val_loss: 0.0016 - val_accuracy: 0.9887\n",
      "Epoch 13/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0014 - accuracy: 0.9908 - val_loss: 0.0012 - val_accuracy: 0.9905\n",
      "Epoch 14/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0013 - accuracy: 0.9903 - val_loss: 0.0012 - val_accuracy: 0.9887\n",
      "Epoch 15/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0012 - accuracy: 0.9911 - val_loss: 8.4824e-04 - val_accuracy: 0.9931\n",
      "Epoch 16/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0012 - accuracy: 0.9905 - val_loss: 9.5230e-04 - val_accuracy: 0.9931\n",
      "Epoch 17/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0011 - accuracy: 0.9899 - val_loss: 0.0011 - val_accuracy: 0.9913\n",
      "Epoch 18/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0011 - accuracy: 0.9924 - val_loss: 0.0011 - val_accuracy: 0.9905\n",
      "Epoch 19/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 0.0011 - accuracy: 0.9911 - val_loss: 8.9363e-04 - val_accuracy: 0.9957\n",
      "Epoch 20/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 9.9489e-04 - accuracy: 0.9932 - val_loss: 0.0013 - val_accuracy: 0.9870\n",
      "Epoch 21/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 9.6413e-04 - accuracy: 0.9939 - val_loss: 7.0439e-04 - val_accuracy: 0.9896\n",
      "Epoch 22/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 8.9930e-04 - accuracy: 0.9943 - val_loss: 7.7669e-04 - val_accuracy: 0.9948\n",
      "Epoch 23/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 9.2884e-04 - accuracy: 0.9915 - val_loss: 8.3889e-04 - val_accuracy: 0.9965\n",
      "Epoch 24/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 8.8661e-04 - accuracy: 0.9924 - val_loss: 7.5151e-04 - val_accuracy: 0.9870\n",
      "Epoch 25/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 8.4028e-04 - accuracy: 0.9922 - val_loss: 8.7860e-04 - val_accuracy: 0.9957\n",
      "Epoch 26/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 8.1882e-04 - accuracy: 0.9927 - val_loss: 8.2390e-04 - val_accuracy: 0.9922\n",
      "Epoch 27/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 8.2154e-04 - accuracy: 0.9934 - val_loss: 5.6785e-04 - val_accuracy: 0.9913\n",
      "Epoch 28/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 7.7026e-04 - accuracy: 0.9943 - val_loss: 8.4133e-04 - val_accuracy: 0.9957\n",
      "Epoch 29/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 7.8378e-04 - accuracy: 0.9931 - val_loss: 6.3655e-04 - val_accuracy: 0.9931\n",
      "Epoch 30/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 7.4481e-04 - accuracy: 0.9924 - val_loss: 7.6909e-04 - val_accuracy: 0.9853\n",
      "Epoch 31/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 7.4997e-04 - accuracy: 0.9934 - val_loss: 6.8074e-04 - val_accuracy: 0.9974\n",
      "Epoch 32/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 7.6152e-04 - accuracy: 0.9932 - val_loss: 5.8394e-04 - val_accuracy: 0.9974\n",
      "Epoch 33/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.8395e-04 - accuracy: 0.9924 - val_loss: 6.4391e-04 - val_accuracy: 0.9974\n",
      "Epoch 34/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.8388e-04 - accuracy: 0.9939 - val_loss: 8.3435e-04 - val_accuracy: 0.9827\n",
      "Epoch 35/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.8207e-04 - accuracy: 0.9927 - val_loss: 7.0036e-04 - val_accuracy: 0.9974\n",
      "Epoch 36/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.6410e-04 - accuracy: 0.9938 - val_loss: 6.3090e-04 - val_accuracy: 0.9965\n",
      "Epoch 37/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.5870e-04 - accuracy: 0.9931 - val_loss: 5.8090e-04 - val_accuracy: 0.9948\n",
      "Epoch 38/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.6529e-04 - accuracy: 0.9931 - val_loss: 7.1263e-04 - val_accuracy: 0.9939\n",
      "Epoch 39/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.9069e-04 - accuracy: 0.9932 - val_loss: 5.3367e-04 - val_accuracy: 0.9939\n",
      "Epoch 40/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.1469e-04 - accuracy: 0.9938 - val_loss: 5.6235e-04 - val_accuracy: 0.9974\n",
      "Epoch 41/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 6.0779e-04 - accuracy: 0.9948 - val_loss: 5.8505e-04 - val_accuracy: 0.9974\n",
      "Epoch 42/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.2869e-04 - accuracy: 0.9948 - val_loss: 5.1915e-04 - val_accuracy: 0.9905\n",
      "Epoch 43/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 6.2646e-04 - accuracy: 0.9932 - val_loss: 5.5589e-04 - val_accuracy: 0.9948\n",
      "Epoch 44/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.9876e-04 - accuracy: 0.9939 - val_loss: 5.3766e-04 - val_accuracy: 0.9965\n",
      "Epoch 45/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 6.0501e-04 - accuracy: 0.9953 - val_loss: 7.9114e-04 - val_accuracy: 0.9853\n",
      "Epoch 46/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.6645e-04 - accuracy: 0.9962 - val_loss: 5.3610e-04 - val_accuracy: 0.9965\n",
      "Epoch 47/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.9963e-04 - accuracy: 0.9951 - val_loss: 4.6689e-04 - val_accuracy: 0.9939\n",
      "Epoch 48/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.8673e-04 - accuracy: 0.9946 - val_loss: 5.5363e-04 - val_accuracy: 0.9965\n",
      "Epoch 49/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.6963e-04 - accuracy: 0.9941 - val_loss: 5.9713e-04 - val_accuracy: 0.9965\n",
      "Epoch 50/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 5.7444e-04 - accuracy: 0.9967 - val_loss: 4.4383e-04 - val_accuracy: 0.9974\n",
      "Epoch 51/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.6154e-04 - accuracy: 0.9955 - val_loss: 7.7096e-04 - val_accuracy: 0.9896\n",
      "Epoch 52/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.5947e-04 - accuracy: 0.9941 - val_loss: 4.5794e-04 - val_accuracy: 0.9983\n",
      "Epoch 53/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.3789e-04 - accuracy: 0.9953 - val_loss: 4.1213e-04 - val_accuracy: 0.9983\n",
      "Epoch 54/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.5009e-04 - accuracy: 0.9934 - val_loss: 5.4642e-04 - val_accuracy: 0.9879\n",
      "Epoch 55/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.4915e-04 - accuracy: 0.9953 - val_loss: 4.2711e-04 - val_accuracy: 0.9939\n",
      "Epoch 56/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.4220e-04 - accuracy: 0.9950 - val_loss: 5.1548e-04 - val_accuracy: 0.9991\n",
      "Epoch 57/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.0840e-04 - accuracy: 0.9957 - val_loss: 5.9403e-04 - val_accuracy: 0.9957\n",
      "Epoch 58/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.2599e-04 - accuracy: 0.9941 - val_loss: 4.4854e-04 - val_accuracy: 0.9983\n",
      "Epoch 59/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.1513e-04 - accuracy: 0.9941 - val_loss: 4.7938e-04 - val_accuracy: 0.9983\n",
      "Epoch 60/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 5.0449e-04 - accuracy: 0.9946 - val_loss: 5.5818e-04 - val_accuracy: 0.9974\n",
      "Epoch 61/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.9279e-04 - accuracy: 0.9950 - val_loss: 5.8403e-04 - val_accuracy: 0.9896\n",
      "Epoch 62/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.8723e-04 - accuracy: 0.9948 - val_loss: 4.6186e-04 - val_accuracy: 0.9991\n",
      "Epoch 63/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.8394e-04 - accuracy: 0.9946 - val_loss: 4.1865e-04 - val_accuracy: 0.9939\n",
      "Epoch 64/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.9710e-04 - accuracy: 0.9950 - val_loss: 4.1869e-04 - val_accuracy: 0.9974\n",
      "Epoch 65/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.8699e-04 - accuracy: 0.9953 - val_loss: 5.0716e-04 - val_accuracy: 0.9991\n",
      "Epoch 66/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.9485e-04 - accuracy: 0.9951 - val_loss: 6.7406e-04 - val_accuracy: 0.9957\n",
      "Epoch 67/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.7853e-04 - accuracy: 0.9946 - val_loss: 6.0909e-04 - val_accuracy: 0.9801\n",
      "Epoch 68/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.9175e-04 - accuracy: 0.9948 - val_loss: 4.8111e-04 - val_accuracy: 0.9905\n",
      "Epoch 69/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.5727e-04 - accuracy: 0.9957 - val_loss: 4.7523e-04 - val_accuracy: 0.9965\n",
      "Epoch 70/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.6555e-04 - accuracy: 0.9958 - val_loss: 4.2664e-04 - val_accuracy: 0.9957\n",
      "Epoch 71/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.6642e-04 - accuracy: 0.9962 - val_loss: 3.9258e-04 - val_accuracy: 0.9983\n",
      "Epoch 72/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.8484e-04 - accuracy: 0.9950 - val_loss: 4.4770e-04 - val_accuracy: 0.9991\n",
      "Epoch 73/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.5285e-04 - accuracy: 0.9951 - val_loss: 3.7027e-04 - val_accuracy: 0.9983\n",
      "Epoch 74/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.4581e-04 - accuracy: 0.9955 - val_loss: 2.9909e-04 - val_accuracy: 0.9965\n",
      "Epoch 75/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.5823e-04 - accuracy: 0.9958 - val_loss: 4.8683e-04 - val_accuracy: 0.9965\n",
      "Epoch 76/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3578e-04 - accuracy: 0.9965 - val_loss: 4.5574e-04 - val_accuracy: 0.9983\n",
      "Epoch 77/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.5296e-04 - accuracy: 0.9957 - val_loss: 3.2910e-04 - val_accuracy: 0.9991\n",
      "Epoch 78/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3091e-04 - accuracy: 0.9948 - val_loss: 3.9475e-04 - val_accuracy: 0.9974\n",
      "Epoch 79/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.4507e-04 - accuracy: 0.9953 - val_loss: 3.8599e-04 - val_accuracy: 0.9974\n",
      "Epoch 80/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.4619e-04 - accuracy: 0.9957 - val_loss: 3.9583e-04 - val_accuracy: 0.9983\n",
      "Epoch 81/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3347e-04 - accuracy: 0.9953 - val_loss: 3.3747e-04 - val_accuracy: 0.9974\n",
      "Epoch 82/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.2619e-04 - accuracy: 0.9969 - val_loss: 3.8441e-04 - val_accuracy: 0.9948\n",
      "Epoch 83/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3821e-04 - accuracy: 0.9955 - val_loss: 3.0592e-04 - val_accuracy: 0.9948\n",
      "Epoch 84/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.1845e-04 - accuracy: 0.9962 - val_loss: 3.0280e-04 - val_accuracy: 0.9965\n",
      "Epoch 85/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3963e-04 - accuracy: 0.9955 - val_loss: 4.1351e-04 - val_accuracy: 0.9896\n",
      "Epoch 86/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.3199e-04 - accuracy: 0.9955 - val_loss: 4.6017e-04 - val_accuracy: 0.9957\n",
      "Epoch 87/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.3351e-04 - accuracy: 0.9955 - val_loss: 3.8124e-04 - val_accuracy: 0.9983\n",
      "Epoch 88/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.2513e-04 - accuracy: 0.9957 - val_loss: 3.2275e-04 - val_accuracy: 0.9965\n",
      "Epoch 89/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.1409e-04 - accuracy: 0.9953 - val_loss: 4.8234e-04 - val_accuracy: 0.9965\n",
      "Epoch 90/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0131e-04 - accuracy: 0.9977 - val_loss: 3.4899e-04 - val_accuracy: 0.9931\n",
      "Epoch 91/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0689e-04 - accuracy: 0.9970 - val_loss: 4.7960e-04 - val_accuracy: 0.9974\n",
      "Epoch 92/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.2349e-04 - accuracy: 0.9957 - val_loss: 3.3098e-04 - val_accuracy: 0.9957\n",
      "Epoch 93/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.8409e-04 - accuracy: 0.9960 - val_loss: 4.0612e-04 - val_accuracy: 0.9991\n",
      "Epoch 94/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.1687e-04 - accuracy: 0.9953 - val_loss: 4.7934e-04 - val_accuracy: 0.9922\n",
      "Epoch 95/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.9256e-04 - accuracy: 0.9969 - val_loss: 3.5945e-04 - val_accuracy: 0.9948\n",
      "Epoch 96/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.8106e-04 - accuracy: 0.9950 - val_loss: 3.8777e-04 - val_accuracy: 0.9948\n",
      "Epoch 97/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.0562e-04 - accuracy: 0.9944 - val_loss: 2.9803e-04 - val_accuracy: 0.9948\n",
      "Epoch 98/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 4.0597e-04 - accuracy: 0.9950 - val_loss: 3.6683e-04 - val_accuracy: 0.9965\n",
      "Epoch 99/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.9914e-04 - accuracy: 0.9960 - val_loss: 3.0732e-04 - val_accuracy: 0.9974\n",
      "Epoch 100/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.9520e-04 - accuracy: 0.9958 - val_loss: 3.6151e-04 - val_accuracy: 0.9983\n",
      "Epoch 101/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 4.0396e-04 - accuracy: 0.9970 - val_loss: 3.3483e-04 - val_accuracy: 0.9983\n",
      "Epoch 102/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.9785e-04 - accuracy: 0.9965 - val_loss: 2.9154e-04 - val_accuracy: 0.9991\n",
      "Epoch 103/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.8252e-04 - accuracy: 0.9958 - val_loss: 4.1751e-04 - val_accuracy: 1.0000\n",
      "Epoch 104/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.7576e-04 - accuracy: 0.9957 - val_loss: 4.3956e-04 - val_accuracy: 0.9905\n",
      "Epoch 105/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.9031e-04 - accuracy: 0.9962 - val_loss: 3.6574e-04 - val_accuracy: 0.9965\n",
      "Epoch 106/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.8476e-04 - accuracy: 0.9969 - val_loss: 3.5843e-04 - val_accuracy: 0.9913\n",
      "Epoch 107/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.8033e-04 - accuracy: 0.9965 - val_loss: 3.1554e-04 - val_accuracy: 0.9957\n",
      "Epoch 108/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.8124e-04 - accuracy: 0.9958 - val_loss: 3.7767e-04 - val_accuracy: 0.9974\n",
      "Epoch 109/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.8962e-04 - accuracy: 0.9972 - val_loss: 2.9298e-04 - val_accuracy: 0.9974\n",
      "Epoch 110/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.6771e-04 - accuracy: 0.9962 - val_loss: 3.6238e-04 - val_accuracy: 0.9991\n",
      "Epoch 111/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.7789e-04 - accuracy: 0.9965 - val_loss: 3.0495e-04 - val_accuracy: 0.9931\n",
      "Epoch 112/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.7510e-04 - accuracy: 0.9948 - val_loss: 3.2091e-04 - val_accuracy: 0.9983\n",
      "Epoch 113/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.6857e-04 - accuracy: 0.9965 - val_loss: 3.0814e-04 - val_accuracy: 0.9974\n",
      "Epoch 114/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.5566e-04 - accuracy: 0.9955 - val_loss: 4.5446e-04 - val_accuracy: 0.9974\n",
      "Epoch 115/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.6398e-04 - accuracy: 0.9964 - val_loss: 4.2741e-04 - val_accuracy: 0.9922\n",
      "Epoch 116/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.7674e-04 - accuracy: 0.9958 - val_loss: 2.5909e-04 - val_accuracy: 0.9957\n",
      "Epoch 117/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.5475e-04 - accuracy: 0.9979 - val_loss: 3.7301e-04 - val_accuracy: 0.9922\n",
      "Epoch 118/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.7860e-04 - accuracy: 0.9960 - val_loss: 4.5067e-04 - val_accuracy: 0.9974\n",
      "Epoch 119/400\n",
      "721/721 [==============================] - 4s 5ms/step - loss: 3.5563e-04 - accuracy: 0.9970 - val_loss: 2.6503e-04 - val_accuracy: 1.0000\n",
      "Epoch 120/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.6082e-04 - accuracy: 0.9967 - val_loss: 3.9055e-04 - val_accuracy: 0.9991\n",
      "Epoch 121/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.6544e-04 - accuracy: 0.9965 - val_loss: 2.6801e-04 - val_accuracy: 0.9983\n",
      "Epoch 122/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.5437e-04 - accuracy: 0.9960 - val_loss: 3.1187e-04 - val_accuracy: 0.9983\n",
      "Epoch 123/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.6609e-04 - accuracy: 0.9977 - val_loss: 2.4773e-04 - val_accuracy: 0.9965\n",
      "Epoch 124/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.5170e-04 - accuracy: 0.9964 - val_loss: 3.0174e-04 - val_accuracy: 0.9965\n",
      "Epoch 125/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4743e-04 - accuracy: 0.9972 - val_loss: 3.4623e-04 - val_accuracy: 0.9991\n",
      "Epoch 126/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.5773e-04 - accuracy: 0.9969 - val_loss: 4.1943e-04 - val_accuracy: 0.9948\n",
      "Epoch 127/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.4401e-04 - accuracy: 0.9957 - val_loss: 3.0882e-04 - val_accuracy: 0.9991\n",
      "Epoch 128/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.5165e-04 - accuracy: 0.9957 - val_loss: 3.3104e-04 - val_accuracy: 0.9983\n",
      "Epoch 129/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4397e-04 - accuracy: 0.9967 - val_loss: 3.4479e-04 - val_accuracy: 0.9965\n",
      "Epoch 130/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.5255e-04 - accuracy: 0.9960 - val_loss: 3.4008e-04 - val_accuracy: 1.0000\n",
      "Epoch 131/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4811e-04 - accuracy: 0.9962 - val_loss: 3.5693e-04 - val_accuracy: 0.9974\n",
      "Epoch 132/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.3916e-04 - accuracy: 0.9962 - val_loss: 2.9540e-04 - val_accuracy: 1.0000\n",
      "Epoch 133/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.3570e-04 - accuracy: 0.9953 - val_loss: 3.2726e-04 - val_accuracy: 0.9965\n",
      "Epoch 134/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.2592e-04 - accuracy: 0.9964 - val_loss: 2.8868e-04 - val_accuracy: 1.0000\n",
      "Epoch 135/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.3350e-04 - accuracy: 0.9967 - val_loss: 2.9559e-04 - val_accuracy: 0.9991\n",
      "Epoch 136/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4183e-04 - accuracy: 0.9970 - val_loss: 2.6534e-04 - val_accuracy: 0.9974\n",
      "Epoch 137/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4339e-04 - accuracy: 0.9960 - val_loss: 3.0293e-04 - val_accuracy: 0.9965\n",
      "Epoch 138/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.3712e-04 - accuracy: 0.9964 - val_loss: 3.5744e-04 - val_accuracy: 0.9913\n",
      "Epoch 139/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2506e-04 - accuracy: 0.9965 - val_loss: 3.2558e-04 - val_accuracy: 0.9948\n",
      "Epoch 140/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.4190e-04 - accuracy: 0.9969 - val_loss: 3.5062e-04 - val_accuracy: 0.9974\n",
      "Epoch 141/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.3657e-04 - accuracy: 0.9970 - val_loss: 2.9088e-04 - val_accuracy: 0.9974\n",
      "Epoch 142/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2561e-04 - accuracy: 0.9962 - val_loss: 2.9172e-04 - val_accuracy: 1.0000\n",
      "Epoch 143/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.3131e-04 - accuracy: 0.9965 - val_loss: 3.7016e-04 - val_accuracy: 0.9957\n",
      "Epoch 144/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2567e-04 - accuracy: 0.9970 - val_loss: 2.9642e-04 - val_accuracy: 0.9991\n",
      "Epoch 145/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2180e-04 - accuracy: 0.9967 - val_loss: 3.3810e-04 - val_accuracy: 0.9991\n",
      "Epoch 146/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.0847e-04 - accuracy: 0.9977 - val_loss: 2.5922e-04 - val_accuracy: 0.9983\n",
      "Epoch 147/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.3037e-04 - accuracy: 0.9960 - val_loss: 2.7148e-04 - val_accuracy: 0.9983\n",
      "Epoch 148/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.2717e-04 - accuracy: 0.9965 - val_loss: 3.2752e-04 - val_accuracy: 0.9957\n",
      "Epoch 149/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.2103e-04 - accuracy: 0.9969 - val_loss: 2.9244e-04 - val_accuracy: 0.9983\n",
      "Epoch 150/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.1697e-04 - accuracy: 0.9969 - val_loss: 2.7861e-04 - val_accuracy: 0.9974\n",
      "Epoch 151/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2206e-04 - accuracy: 0.9962 - val_loss: 2.7599e-04 - val_accuracy: 1.0000\n",
      "Epoch 152/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.3465e-04 - accuracy: 0.9972 - val_loss: 3.4403e-04 - val_accuracy: 0.9974\n",
      "Epoch 153/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.1493e-04 - accuracy: 0.9965 - val_loss: 2.4587e-04 - val_accuracy: 0.9983\n",
      "Epoch 154/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.2041e-04 - accuracy: 0.9974 - val_loss: 3.1616e-04 - val_accuracy: 0.9974\n",
      "Epoch 155/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.1758e-04 - accuracy: 0.9974 - val_loss: 2.9431e-04 - val_accuracy: 0.9974\n",
      "Epoch 156/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.1809e-04 - accuracy: 0.9960 - val_loss: 3.0960e-04 - val_accuracy: 0.9991\n",
      "Epoch 157/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0867e-04 - accuracy: 0.9976 - val_loss: 2.5836e-04 - val_accuracy: 1.0000\n",
      "Epoch 158/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.2328e-04 - accuracy: 0.9962 - val_loss: 2.7346e-04 - val_accuracy: 0.9965\n",
      "Epoch 159/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.1687e-04 - accuracy: 0.9970 - val_loss: 2.9160e-04 - val_accuracy: 0.9983\n",
      "Epoch 160/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.2184e-04 - accuracy: 0.9965 - val_loss: 2.8773e-04 - val_accuracy: 0.9965\n",
      "Epoch 161/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 3.1201e-04 - accuracy: 0.9957 - val_loss: 3.3591e-04 - val_accuracy: 0.9991\n",
      "Epoch 162/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.0428e-04 - accuracy: 0.9981 - val_loss: 2.8312e-04 - val_accuracy: 0.9983\n",
      "Epoch 163/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.0971e-04 - accuracy: 0.9958 - val_loss: 2.9038e-04 - val_accuracy: 0.9948\n",
      "Epoch 164/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.1541e-04 - accuracy: 0.9964 - val_loss: 2.7074e-04 - val_accuracy: 1.0000\n",
      "Epoch 165/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.0570e-04 - accuracy: 0.9981 - val_loss: 2.5017e-04 - val_accuracy: 1.0000\n",
      "Epoch 166/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.1736e-04 - accuracy: 0.9979 - val_loss: 4.0373e-04 - val_accuracy: 0.9913\n",
      "Epoch 167/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.0679e-04 - accuracy: 0.9974 - val_loss: 2.9402e-04 - val_accuracy: 0.9983\n",
      "Epoch 168/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.1555e-04 - accuracy: 0.9962 - val_loss: 3.3366e-04 - val_accuracy: 0.9983\n",
      "Epoch 169/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 3.0943e-04 - accuracy: 0.9972 - val_loss: 3.0259e-04 - val_accuracy: 1.0000\n",
      "Epoch 170/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9962e-04 - accuracy: 0.9976 - val_loss: 3.0820e-04 - val_accuracy: 0.9974\n",
      "Epoch 171/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0533e-04 - accuracy: 0.9964 - val_loss: 2.8597e-04 - val_accuracy: 0.9974\n",
      "Epoch 172/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0280e-04 - accuracy: 0.9969 - val_loss: 2.5819e-04 - val_accuracy: 0.9974\n",
      "Epoch 173/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0374e-04 - accuracy: 0.9969 - val_loss: 2.4427e-04 - val_accuracy: 0.9983\n",
      "Epoch 174/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9611e-04 - accuracy: 0.9970 - val_loss: 2.4572e-04 - val_accuracy: 0.9991\n",
      "Epoch 175/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9787e-04 - accuracy: 0.9962 - val_loss: 2.8385e-04 - val_accuracy: 0.9965\n",
      "Epoch 176/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0497e-04 - accuracy: 0.9972 - val_loss: 2.8299e-04 - val_accuracy: 0.9983\n",
      "Epoch 177/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9374e-04 - accuracy: 0.9972 - val_loss: 2.7185e-04 - val_accuracy: 0.9991\n",
      "Epoch 178/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0762e-04 - accuracy: 0.9965 - val_loss: 2.9022e-04 - val_accuracy: 0.9931\n",
      "Epoch 179/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 3.0123e-04 - accuracy: 0.9970 - val_loss: 2.5188e-04 - val_accuracy: 0.9983\n",
      "Epoch 180/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9636e-04 - accuracy: 0.9964 - val_loss: 3.4911e-04 - val_accuracy: 0.9931\n",
      "Epoch 181/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.9515e-04 - accuracy: 0.9969 - val_loss: 4.0047e-04 - val_accuracy: 0.9939\n",
      "Epoch 182/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8848e-04 - accuracy: 0.9983 - val_loss: 2.4940e-04 - val_accuracy: 0.9948\n",
      "Epoch 183/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.9635e-04 - accuracy: 0.9970 - val_loss: 2.6695e-04 - val_accuracy: 0.9983\n",
      "Epoch 184/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.9536e-04 - accuracy: 0.9957 - val_loss: 2.0928e-04 - val_accuracy: 0.9991\n",
      "Epoch 185/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8557e-04 - accuracy: 0.9964 - val_loss: 2.2642e-04 - val_accuracy: 0.9974\n",
      "Epoch 186/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8804e-04 - accuracy: 0.9962 - val_loss: 2.5421e-04 - val_accuracy: 0.9974\n",
      "Epoch 187/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.9286e-04 - accuracy: 0.9976 - val_loss: 2.5127e-04 - val_accuracy: 1.0000\n",
      "Epoch 188/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.9911e-04 - accuracy: 0.9967 - val_loss: 3.7039e-04 - val_accuracy: 0.9983\n",
      "Epoch 189/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8630e-04 - accuracy: 0.9976 - val_loss: 2.3479e-04 - val_accuracy: 0.9983\n",
      "Epoch 190/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9068e-04 - accuracy: 0.9979 - val_loss: 2.8412e-04 - val_accuracy: 1.0000\n",
      "Epoch 191/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9173e-04 - accuracy: 0.9958 - val_loss: 2.4406e-04 - val_accuracy: 0.9965\n",
      "Epoch 192/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8671e-04 - accuracy: 0.9977 - val_loss: 2.6293e-04 - val_accuracy: 0.9991\n",
      "Epoch 193/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8699e-04 - accuracy: 0.9967 - val_loss: 2.8547e-04 - val_accuracy: 0.9983\n",
      "Epoch 194/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8594e-04 - accuracy: 0.9977 - val_loss: 2.6832e-04 - val_accuracy: 0.9974\n",
      "Epoch 195/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9147e-04 - accuracy: 0.9972 - val_loss: 3.4013e-04 - val_accuracy: 0.9948\n",
      "Epoch 196/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8723e-04 - accuracy: 0.9972 - val_loss: 2.9736e-04 - val_accuracy: 1.0000\n",
      "Epoch 197/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9739e-04 - accuracy: 0.9974 - val_loss: 2.8788e-04 - val_accuracy: 0.9991\n",
      "Epoch 198/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9247e-04 - accuracy: 0.9977 - val_loss: 2.7804e-04 - val_accuracy: 0.9991\n",
      "Epoch 199/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8154e-04 - accuracy: 0.9969 - val_loss: 2.4113e-04 - val_accuracy: 0.9991\n",
      "Epoch 200/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.7300e-04 - accuracy: 0.9977 - val_loss: 2.4650e-04 - val_accuracy: 0.9983\n",
      "Epoch 201/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7838e-04 - accuracy: 0.9976 - val_loss: 2.7232e-04 - val_accuracy: 0.9983\n",
      "Epoch 202/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.8601e-04 - accuracy: 0.9972 - val_loss: 4.1330e-04 - val_accuracy: 0.9965\n",
      "Epoch 203/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.8619e-04 - accuracy: 0.9965 - val_loss: 3.5335e-04 - val_accuracy: 0.9974\n",
      "Epoch 204/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9033e-04 - accuracy: 0.9981 - val_loss: 2.2854e-04 - val_accuracy: 0.9991\n",
      "Epoch 205/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7128e-04 - accuracy: 0.9974 - val_loss: 2.5594e-04 - val_accuracy: 0.9974\n",
      "Epoch 206/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.9280e-04 - accuracy: 0.9977 - val_loss: 2.4911e-04 - val_accuracy: 0.9965\n",
      "Epoch 207/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.9442e-04 - accuracy: 0.9976 - val_loss: 2.0561e-04 - val_accuracy: 0.9983\n",
      "Epoch 208/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7966e-04 - accuracy: 0.9969 - val_loss: 2.7694e-04 - val_accuracy: 0.9965\n",
      "Epoch 209/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8230e-04 - accuracy: 0.9972 - val_loss: 2.6854e-04 - val_accuracy: 1.0000\n",
      "Epoch 210/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8047e-04 - accuracy: 0.9972 - val_loss: 3.4839e-04 - val_accuracy: 1.0000\n",
      "Epoch 211/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8005e-04 - accuracy: 0.9977 - val_loss: 3.6746e-04 - val_accuracy: 0.9991\n",
      "Epoch 212/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.7513e-04 - accuracy: 0.9974 - val_loss: 2.9740e-04 - val_accuracy: 0.9974\n",
      "Epoch 213/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.7758e-04 - accuracy: 0.9977 - val_loss: 2.6574e-04 - val_accuracy: 0.9991\n",
      "Epoch 214/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.8071e-04 - accuracy: 0.9972 - val_loss: 2.2697e-04 - val_accuracy: 0.9983\n",
      "Epoch 215/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.6895e-04 - accuracy: 0.9970 - val_loss: 2.5036e-04 - val_accuracy: 0.9983\n",
      "Epoch 216/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.8373e-04 - accuracy: 0.9967 - val_loss: 2.4301e-04 - val_accuracy: 0.9991\n",
      "Epoch 217/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6955e-04 - accuracy: 0.9972 - val_loss: 2.6590e-04 - val_accuracy: 0.9957\n",
      "Epoch 218/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.8021e-04 - accuracy: 0.9976 - val_loss: 2.4807e-04 - val_accuracy: 1.0000\n",
      "Epoch 219/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7613e-04 - accuracy: 0.9969 - val_loss: 2.6702e-04 - val_accuracy: 0.9957\n",
      "Epoch 220/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6181e-04 - accuracy: 0.9988 - val_loss: 2.4818e-04 - val_accuracy: 0.9991\n",
      "Epoch 221/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6541e-04 - accuracy: 0.9974 - val_loss: 2.1559e-04 - val_accuracy: 0.9983\n",
      "Epoch 222/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.7714e-04 - accuracy: 0.9974 - val_loss: 2.0918e-04 - val_accuracy: 0.9983\n",
      "Epoch 223/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.6988e-04 - accuracy: 0.9972 - val_loss: 3.7391e-04 - val_accuracy: 0.9957\n",
      "Epoch 224/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.6628e-04 - accuracy: 0.9984 - val_loss: 1.9530e-04 - val_accuracy: 0.9965\n",
      "Epoch 225/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.7845e-04 - accuracy: 0.9983 - val_loss: 2.9355e-04 - val_accuracy: 0.9983\n",
      "Epoch 226/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5941e-04 - accuracy: 0.9962 - val_loss: 3.2418e-04 - val_accuracy: 0.9939\n",
      "Epoch 227/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7239e-04 - accuracy: 0.9965 - val_loss: 2.6510e-04 - val_accuracy: 0.9983\n",
      "Epoch 228/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6086e-04 - accuracy: 0.9972 - val_loss: 2.8780e-04 - val_accuracy: 0.9974\n",
      "Epoch 229/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.7492e-04 - accuracy: 0.9972 - val_loss: 2.0788e-04 - val_accuracy: 0.9974\n",
      "Epoch 230/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.7149e-04 - accuracy: 0.9974 - val_loss: 2.7311e-04 - val_accuracy: 0.9991\n",
      "Epoch 231/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6791e-04 - accuracy: 0.9974 - val_loss: 2.2570e-04 - val_accuracy: 0.9957\n",
      "Epoch 232/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6075e-04 - accuracy: 0.9977 - val_loss: 2.6851e-04 - val_accuracy: 0.9965\n",
      "Epoch 233/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.6483e-04 - accuracy: 0.9962 - val_loss: 2.2169e-04 - val_accuracy: 0.9957\n",
      "Epoch 234/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5716e-04 - accuracy: 0.9986 - val_loss: 1.9589e-04 - val_accuracy: 0.9991\n",
      "Epoch 235/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5586e-04 - accuracy: 0.9983 - val_loss: 2.1845e-04 - val_accuracy: 0.9974\n",
      "Epoch 236/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.5942e-04 - accuracy: 0.9969 - val_loss: 2.0215e-04 - val_accuracy: 1.0000\n",
      "Epoch 237/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.6129e-04 - accuracy: 0.9979 - val_loss: 4.1021e-04 - val_accuracy: 0.9991\n",
      "Epoch 238/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6558e-04 - accuracy: 0.9984 - val_loss: 2.6163e-04 - val_accuracy: 1.0000\n",
      "Epoch 239/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5435e-04 - accuracy: 0.9974 - val_loss: 3.1337e-04 - val_accuracy: 1.0000\n",
      "Epoch 240/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5691e-04 - accuracy: 0.9967 - val_loss: 2.7521e-04 - val_accuracy: 1.0000\n",
      "Epoch 241/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5797e-04 - accuracy: 0.9976 - val_loss: 2.5631e-04 - val_accuracy: 1.0000\n",
      "Epoch 242/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5894e-04 - accuracy: 0.9970 - val_loss: 1.9977e-04 - val_accuracy: 1.0000\n",
      "Epoch 243/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5105e-04 - accuracy: 0.9984 - val_loss: 2.4778e-04 - val_accuracy: 0.9974\n",
      "Epoch 244/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.6160e-04 - accuracy: 0.9983 - val_loss: 2.0207e-04 - val_accuracy: 1.0000\n",
      "Epoch 245/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5605e-04 - accuracy: 0.9977 - val_loss: 1.9920e-04 - val_accuracy: 0.9991\n",
      "Epoch 246/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5081e-04 - accuracy: 0.9981 - val_loss: 2.8723e-04 - val_accuracy: 0.9974\n",
      "Epoch 247/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5467e-04 - accuracy: 0.9960 - val_loss: 2.3482e-04 - val_accuracy: 0.9965\n",
      "Epoch 248/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5732e-04 - accuracy: 0.9983 - val_loss: 2.8215e-04 - val_accuracy: 0.9983\n",
      "Epoch 249/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5507e-04 - accuracy: 0.9974 - val_loss: 2.4249e-04 - val_accuracy: 1.0000\n",
      "Epoch 250/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5144e-04 - accuracy: 0.9981 - val_loss: 2.4224e-04 - val_accuracy: 0.9957\n",
      "Epoch 251/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5394e-04 - accuracy: 0.9981 - val_loss: 2.2204e-04 - val_accuracy: 0.9974\n",
      "Epoch 252/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5442e-04 - accuracy: 0.9969 - val_loss: 1.9340e-04 - val_accuracy: 1.0000\n",
      "Epoch 253/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5531e-04 - accuracy: 0.9969 - val_loss: 3.0067e-04 - val_accuracy: 0.9991\n",
      "Epoch 254/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5259e-04 - accuracy: 0.9974 - val_loss: 2.1386e-04 - val_accuracy: 1.0000\n",
      "Epoch 255/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5165e-04 - accuracy: 0.9979 - val_loss: 2.3247e-04 - val_accuracy: 0.9991\n",
      "Epoch 256/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5498e-04 - accuracy: 0.9974 - val_loss: 2.6615e-04 - val_accuracy: 0.9983\n",
      "Epoch 257/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5509e-04 - accuracy: 0.9988 - val_loss: 2.2069e-04 - val_accuracy: 0.9991\n",
      "Epoch 258/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5141e-04 - accuracy: 0.9979 - val_loss: 2.0449e-04 - val_accuracy: 0.9991\n",
      "Epoch 259/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5094e-04 - accuracy: 0.9977 - val_loss: 2.2817e-04 - val_accuracy: 0.9991\n",
      "Epoch 260/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4907e-04 - accuracy: 0.9984 - val_loss: 3.1745e-04 - val_accuracy: 0.9983\n",
      "Epoch 261/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5002e-04 - accuracy: 0.9969 - val_loss: 2.7974e-04 - val_accuracy: 0.9991\n",
      "Epoch 262/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4609e-04 - accuracy: 0.9974 - val_loss: 2.4504e-04 - val_accuracy: 0.9974\n",
      "Epoch 263/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.5793e-04 - accuracy: 0.9974 - val_loss: 2.8944e-04 - val_accuracy: 0.9983\n",
      "Epoch 264/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4414e-04 - accuracy: 0.9981 - val_loss: 2.5343e-04 - val_accuracy: 0.9957\n",
      "Epoch 265/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4384e-04 - accuracy: 0.9981 - val_loss: 3.0355e-04 - val_accuracy: 0.9974\n",
      "Epoch 266/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4762e-04 - accuracy: 0.9977 - val_loss: 2.2799e-04 - val_accuracy: 0.9991\n",
      "Epoch 267/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.3650e-04 - accuracy: 0.9988 - val_loss: 1.9015e-04 - val_accuracy: 0.9991\n",
      "Epoch 268/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.5914e-04 - accuracy: 0.9983 - val_loss: 2.3817e-04 - val_accuracy: 0.9931\n",
      "Epoch 269/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4834e-04 - accuracy: 0.9976 - val_loss: 2.1141e-04 - val_accuracy: 0.9991\n",
      "Epoch 270/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4582e-04 - accuracy: 0.9977 - val_loss: 2.0091e-04 - val_accuracy: 0.9983\n",
      "Epoch 271/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4396e-04 - accuracy: 0.9981 - val_loss: 2.0604e-04 - val_accuracy: 0.9991\n",
      "Epoch 272/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3526e-04 - accuracy: 0.9986 - val_loss: 1.7742e-04 - val_accuracy: 0.9991\n",
      "Epoch 273/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4476e-04 - accuracy: 0.9984 - val_loss: 2.1135e-04 - val_accuracy: 1.0000\n",
      "Epoch 274/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4030e-04 - accuracy: 0.9976 - val_loss: 2.0053e-04 - val_accuracy: 1.0000\n",
      "Epoch 275/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4162e-04 - accuracy: 0.9983 - val_loss: 3.2675e-04 - val_accuracy: 0.9983\n",
      "Epoch 276/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4222e-04 - accuracy: 0.9990 - val_loss: 2.3799e-04 - val_accuracy: 0.9991\n",
      "Epoch 277/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4574e-04 - accuracy: 0.9977 - val_loss: 2.4460e-04 - val_accuracy: 0.9974\n",
      "Epoch 278/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3743e-04 - accuracy: 0.9981 - val_loss: 2.7159e-04 - val_accuracy: 0.9983\n",
      "Epoch 279/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4063e-04 - accuracy: 0.9976 - val_loss: 2.2158e-04 - val_accuracy: 0.9991\n",
      "Epoch 280/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3620e-04 - accuracy: 0.9979 - val_loss: 1.8327e-04 - val_accuracy: 0.9965\n",
      "Epoch 281/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4088e-04 - accuracy: 0.9976 - val_loss: 2.0441e-04 - val_accuracy: 0.9983\n",
      "Epoch 282/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4776e-04 - accuracy: 0.9984 - val_loss: 2.2102e-04 - val_accuracy: 0.9991\n",
      "Epoch 283/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4266e-04 - accuracy: 0.9983 - val_loss: 2.0585e-04 - val_accuracy: 1.0000\n",
      "Epoch 284/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4536e-04 - accuracy: 0.9972 - val_loss: 2.2696e-04 - val_accuracy: 1.0000\n",
      "Epoch 285/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4212e-04 - accuracy: 0.9986 - val_loss: 3.3452e-04 - val_accuracy: 0.9991\n",
      "Epoch 286/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4489e-04 - accuracy: 0.9976 - val_loss: 2.0441e-04 - val_accuracy: 0.9974\n",
      "Epoch 287/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4533e-04 - accuracy: 0.9976 - val_loss: 2.9387e-04 - val_accuracy: 1.0000\n",
      "Epoch 288/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4660e-04 - accuracy: 0.9981 - val_loss: 2.0446e-04 - val_accuracy: 0.9983\n",
      "Epoch 289/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4029e-04 - accuracy: 0.9981 - val_loss: 1.8921e-04 - val_accuracy: 1.0000\n",
      "Epoch 290/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3905e-04 - accuracy: 0.9983 - val_loss: 2.1802e-04 - val_accuracy: 0.9974\n",
      "Epoch 291/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3403e-04 - accuracy: 0.9983 - val_loss: 2.3663e-04 - val_accuracy: 1.0000\n",
      "Epoch 292/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4202e-04 - accuracy: 0.9983 - val_loss: 2.2302e-04 - val_accuracy: 0.9991\n",
      "Epoch 293/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3950e-04 - accuracy: 0.9990 - val_loss: 2.0055e-04 - val_accuracy: 0.9991\n",
      "Epoch 294/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3702e-04 - accuracy: 0.9981 - val_loss: 2.2317e-04 - val_accuracy: 0.9957\n",
      "Epoch 295/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3194e-04 - accuracy: 0.9986 - val_loss: 2.3910e-04 - val_accuracy: 0.9983\n",
      "Epoch 296/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.4005e-04 - accuracy: 0.9976 - val_loss: 2.3803e-04 - val_accuracy: 1.0000\n",
      "Epoch 297/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3658e-04 - accuracy: 0.9981 - val_loss: 2.3527e-04 - val_accuracy: 0.9983\n",
      "Epoch 298/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3788e-04 - accuracy: 0.9970 - val_loss: 1.8400e-04 - val_accuracy: 1.0000\n",
      "Epoch 299/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3560e-04 - accuracy: 0.9984 - val_loss: 2.1076e-04 - val_accuracy: 0.9991\n",
      "Epoch 300/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3423e-04 - accuracy: 0.9986 - val_loss: 2.5979e-04 - val_accuracy: 0.9991\n",
      "Epoch 301/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3172e-04 - accuracy: 0.9981 - val_loss: 2.1238e-04 - val_accuracy: 1.0000\n",
      "Epoch 302/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3193e-04 - accuracy: 0.9983 - val_loss: 1.9624e-04 - val_accuracy: 0.9991\n",
      "Epoch 303/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3682e-04 - accuracy: 0.9974 - val_loss: 2.1924e-04 - val_accuracy: 0.9983\n",
      "Epoch 304/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.3308e-04 - accuracy: 0.9988 - val_loss: 1.8860e-04 - val_accuracy: 1.0000\n",
      "Epoch 305/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.4729e-04 - accuracy: 0.9972 - val_loss: 2.3006e-04 - val_accuracy: 0.9991\n",
      "Epoch 306/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3636e-04 - accuracy: 0.9983 - val_loss: 2.4195e-04 - val_accuracy: 0.9983\n",
      "Epoch 307/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3451e-04 - accuracy: 0.9984 - val_loss: 2.9592e-04 - val_accuracy: 0.9974\n",
      "Epoch 308/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3282e-04 - accuracy: 0.9983 - val_loss: 2.3961e-04 - val_accuracy: 0.9957\n",
      "Epoch 309/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3495e-04 - accuracy: 0.9983 - val_loss: 1.8162e-04 - val_accuracy: 1.0000\n",
      "Epoch 310/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3332e-04 - accuracy: 0.9974 - val_loss: 1.8423e-04 - val_accuracy: 1.0000\n",
      "Epoch 311/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.2840e-04 - accuracy: 0.9976 - val_loss: 2.8470e-04 - val_accuracy: 0.9948\n",
      "Epoch 312/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3076e-04 - accuracy: 0.9977 - val_loss: 2.1277e-04 - val_accuracy: 1.0000\n",
      "Epoch 313/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.2498e-04 - accuracy: 0.9977 - val_loss: 2.2994e-04 - val_accuracy: 0.9983\n",
      "Epoch 314/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.3372e-04 - accuracy: 0.9986 - val_loss: 1.8433e-04 - val_accuracy: 0.9983\n",
      "Epoch 315/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2810e-04 - accuracy: 0.9983 - val_loss: 2.0464e-04 - val_accuracy: 1.0000\n",
      "Epoch 316/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2877e-04 - accuracy: 0.9986 - val_loss: 2.2923e-04 - val_accuracy: 0.9939\n",
      "Epoch 317/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2613e-04 - accuracy: 0.9977 - val_loss: 2.0860e-04 - val_accuracy: 0.9965\n",
      "Epoch 318/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3090e-04 - accuracy: 0.9979 - val_loss: 2.1811e-04 - val_accuracy: 0.9991\n",
      "Epoch 319/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3267e-04 - accuracy: 0.9984 - val_loss: 2.3175e-04 - val_accuracy: 1.0000\n",
      "Epoch 320/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2380e-04 - accuracy: 0.9976 - val_loss: 2.5259e-04 - val_accuracy: 0.9991\n",
      "Epoch 321/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2413e-04 - accuracy: 0.9986 - val_loss: 2.1671e-04 - val_accuracy: 0.9983\n",
      "Epoch 322/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2754e-04 - accuracy: 0.9981 - val_loss: 2.6700e-04 - val_accuracy: 0.9974\n",
      "Epoch 323/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2285e-04 - accuracy: 0.9986 - val_loss: 2.6810e-04 - val_accuracy: 0.9965\n",
      "Epoch 324/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3009e-04 - accuracy: 0.9983 - val_loss: 2.2538e-04 - val_accuracy: 0.9983\n",
      "Epoch 325/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2518e-04 - accuracy: 0.9983 - val_loss: 2.7109e-04 - val_accuracy: 0.9983\n",
      "Epoch 326/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2388e-04 - accuracy: 0.9988 - val_loss: 1.9563e-04 - val_accuracy: 0.9991\n",
      "Epoch 327/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2831e-04 - accuracy: 0.9983 - val_loss: 1.8896e-04 - val_accuracy: 1.0000\n",
      "Epoch 328/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2422e-04 - accuracy: 0.9977 - val_loss: 1.8889e-04 - val_accuracy: 0.9983\n",
      "Epoch 329/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.3096e-04 - accuracy: 0.9972 - val_loss: 2.1360e-04 - val_accuracy: 0.9991\n",
      "Epoch 330/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2963e-04 - accuracy: 0.9983 - val_loss: 2.4173e-04 - val_accuracy: 0.9965\n",
      "Epoch 331/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2801e-04 - accuracy: 0.9983 - val_loss: 1.7852e-04 - val_accuracy: 0.9983\n",
      "Epoch 332/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2350e-04 - accuracy: 0.9988 - val_loss: 1.9930e-04 - val_accuracy: 0.9991\n",
      "Epoch 333/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2900e-04 - accuracy: 0.9986 - val_loss: 3.2722e-04 - val_accuracy: 0.9991\n",
      "Epoch 334/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2564e-04 - accuracy: 0.9976 - val_loss: 2.3516e-04 - val_accuracy: 0.9974\n",
      "Epoch 335/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2058e-04 - accuracy: 0.9979 - val_loss: 2.0714e-04 - val_accuracy: 0.9974\n",
      "Epoch 336/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1823e-04 - accuracy: 0.9983 - val_loss: 2.0984e-04 - val_accuracy: 0.9974\n",
      "Epoch 337/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1636e-04 - accuracy: 0.9983 - val_loss: 3.0126e-04 - val_accuracy: 0.9991\n",
      "Epoch 338/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1474e-04 - accuracy: 0.9983 - val_loss: 2.3299e-04 - val_accuracy: 0.9974\n",
      "Epoch 339/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2708e-04 - accuracy: 0.9981 - val_loss: 2.2213e-04 - val_accuracy: 0.9991\n",
      "Epoch 340/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2345e-04 - accuracy: 0.9976 - val_loss: 1.5906e-04 - val_accuracy: 1.0000\n",
      "Epoch 341/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2117e-04 - accuracy: 0.9981 - val_loss: 2.8025e-04 - val_accuracy: 0.9957\n",
      "Epoch 342/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2026e-04 - accuracy: 0.9981 - val_loss: 1.9388e-04 - val_accuracy: 1.0000\n",
      "Epoch 343/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1972e-04 - accuracy: 0.9991 - val_loss: 2.1586e-04 - val_accuracy: 0.9983\n",
      "Epoch 344/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1526e-04 - accuracy: 0.9977 - val_loss: 1.9019e-04 - val_accuracy: 0.9991\n",
      "Epoch 345/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1964e-04 - accuracy: 0.9976 - val_loss: 2.1105e-04 - val_accuracy: 1.0000\n",
      "Epoch 346/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1458e-04 - accuracy: 0.9981 - val_loss: 2.1525e-04 - val_accuracy: 0.9983\n",
      "Epoch 347/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.1450e-04 - accuracy: 0.9983 - val_loss: 2.3953e-04 - val_accuracy: 1.0000\n",
      "Epoch 348/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1391e-04 - accuracy: 0.9983 - val_loss: 1.8714e-04 - val_accuracy: 1.0000\n",
      "Epoch 349/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1727e-04 - accuracy: 0.9974 - val_loss: 1.7017e-04 - val_accuracy: 0.9991\n",
      "Epoch 350/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2005e-04 - accuracy: 0.9983 - val_loss: 2.0916e-04 - val_accuracy: 1.0000\n",
      "Epoch 351/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2468e-04 - accuracy: 0.9979 - val_loss: 1.5992e-04 - val_accuracy: 1.0000\n",
      "Epoch 352/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.2835e-04 - accuracy: 0.9979 - val_loss: 1.8731e-04 - val_accuracy: 0.9991\n",
      "Epoch 353/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.2176e-04 - accuracy: 0.9974 - val_loss: 2.0791e-04 - val_accuracy: 0.9991\n",
      "Epoch 354/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.1074e-04 - accuracy: 0.9983 - val_loss: 1.8231e-04 - val_accuracy: 1.0000\n",
      "Epoch 355/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1993e-04 - accuracy: 0.9983 - val_loss: 2.1212e-04 - val_accuracy: 0.9965\n",
      "Epoch 356/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1086e-04 - accuracy: 0.9983 - val_loss: 1.8590e-04 - val_accuracy: 0.9974\n",
      "Epoch 357/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1707e-04 - accuracy: 0.9972 - val_loss: 2.3438e-04 - val_accuracy: 1.0000\n",
      "Epoch 358/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.1838e-04 - accuracy: 0.9979 - val_loss: 1.6484e-04 - val_accuracy: 0.9991\n",
      "Epoch 359/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2158e-04 - accuracy: 0.9981 - val_loss: 1.9212e-04 - val_accuracy: 0.9983\n",
      "Epoch 360/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1826e-04 - accuracy: 0.9984 - val_loss: 2.1813e-04 - val_accuracy: 0.9974\n",
      "Epoch 361/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1068e-04 - accuracy: 0.9979 - val_loss: 2.2877e-04 - val_accuracy: 0.9983\n",
      "Epoch 362/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1516e-04 - accuracy: 0.9990 - val_loss: 2.5157e-04 - val_accuracy: 0.9974\n",
      "Epoch 363/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0318e-04 - accuracy: 0.9983 - val_loss: 2.6513e-04 - val_accuracy: 1.0000\n",
      "Epoch 364/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1784e-04 - accuracy: 0.9990 - val_loss: 2.1626e-04 - val_accuracy: 0.9991\n",
      "Epoch 365/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1583e-04 - accuracy: 0.9984 - val_loss: 2.1997e-04 - val_accuracy: 0.9957\n",
      "Epoch 366/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0791e-04 - accuracy: 0.9984 - val_loss: 2.1902e-04 - val_accuracy: 0.9991\n",
      "Epoch 367/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1583e-04 - accuracy: 0.9986 - val_loss: 1.8512e-04 - val_accuracy: 1.0000\n",
      "Epoch 368/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0860e-04 - accuracy: 0.9986 - val_loss: 1.8042e-04 - val_accuracy: 0.9991\n",
      "Epoch 369/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1123e-04 - accuracy: 0.9990 - val_loss: 2.1319e-04 - val_accuracy: 1.0000\n",
      "Epoch 370/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.2065e-04 - accuracy: 0.9983 - val_loss: 1.8312e-04 - val_accuracy: 0.9983\n",
      "Epoch 371/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1473e-04 - accuracy: 0.9979 - val_loss: 1.7268e-04 - val_accuracy: 1.0000\n",
      "Epoch 372/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1194e-04 - accuracy: 0.9986 - val_loss: 1.7784e-04 - val_accuracy: 0.9991\n",
      "Epoch 373/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1327e-04 - accuracy: 0.9990 - val_loss: 1.8293e-04 - val_accuracy: 1.0000\n",
      "Epoch 374/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1261e-04 - accuracy: 0.9988 - val_loss: 1.8237e-04 - val_accuracy: 1.0000\n",
      "Epoch 375/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.0880e-04 - accuracy: 0.9981 - val_loss: 1.9300e-04 - val_accuracy: 0.9983\n",
      "Epoch 376/400\n",
      "721/721 [==============================] - 4s 6ms/step - loss: 2.1249e-04 - accuracy: 0.9981 - val_loss: 1.8610e-04 - val_accuracy: 0.9983\n",
      "Epoch 377/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.2138e-04 - accuracy: 0.9986 - val_loss: 1.8076e-04 - val_accuracy: 0.9991\n",
      "Epoch 378/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1702e-04 - accuracy: 0.9977 - val_loss: 2.0025e-04 - val_accuracy: 0.9974\n",
      "Epoch 379/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0209e-04 - accuracy: 0.9984 - val_loss: 1.8470e-04 - val_accuracy: 0.9983\n",
      "Epoch 380/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1174e-04 - accuracy: 0.9977 - val_loss: 1.6402e-04 - val_accuracy: 1.0000\n",
      "Epoch 381/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.0880e-04 - accuracy: 0.9988 - val_loss: 1.8866e-04 - val_accuracy: 0.9991\n",
      "Epoch 382/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1057e-04 - accuracy: 0.9974 - val_loss: 2.0746e-04 - val_accuracy: 0.9991\n",
      "Epoch 383/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0920e-04 - accuracy: 0.9979 - val_loss: 1.6194e-04 - val_accuracy: 1.0000\n",
      "Epoch 384/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0476e-04 - accuracy: 0.9988 - val_loss: 1.8297e-04 - val_accuracy: 0.9991\n",
      "Epoch 385/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.0006e-04 - accuracy: 0.9988 - val_loss: 1.6510e-04 - val_accuracy: 0.9991\n",
      "Epoch 386/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1369e-04 - accuracy: 0.9984 - val_loss: 1.5704e-04 - val_accuracy: 0.9991\n",
      "Epoch 387/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1022e-04 - accuracy: 0.9972 - val_loss: 1.9529e-04 - val_accuracy: 0.9983\n",
      "Epoch 388/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1083e-04 - accuracy: 0.9977 - val_loss: 2.0217e-04 - val_accuracy: 0.9991\n",
      "Epoch 389/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0559e-04 - accuracy: 0.9983 - val_loss: 2.0562e-04 - val_accuracy: 0.9974\n",
      "Epoch 390/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0726e-04 - accuracy: 0.9983 - val_loss: 1.8532e-04 - val_accuracy: 1.0000\n",
      "Epoch 391/400\n",
      "721/721 [==============================] - 5s 6ms/step - loss: 2.0917e-04 - accuracy: 0.9981 - val_loss: 2.3513e-04 - val_accuracy: 0.9974\n",
      "Epoch 392/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0537e-04 - accuracy: 0.9986 - val_loss: 2.0796e-04 - val_accuracy: 0.9983\n",
      "Epoch 393/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0563e-04 - accuracy: 0.9990 - val_loss: 1.8280e-04 - val_accuracy: 0.9983\n",
      "Epoch 394/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1053e-04 - accuracy: 0.9986 - val_loss: 1.7343e-04 - val_accuracy: 0.9991\n",
      "Epoch 395/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0539e-04 - accuracy: 0.9977 - val_loss: 2.2689e-04 - val_accuracy: 0.9983\n",
      "Epoch 396/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0661e-04 - accuracy: 0.9988 - val_loss: 1.9710e-04 - val_accuracy: 1.0000\n",
      "Epoch 397/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0872e-04 - accuracy: 0.9981 - val_loss: 1.6625e-04 - val_accuracy: 0.9983\n",
      "Epoch 398/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1358e-04 - accuracy: 0.9993 - val_loss: 1.9879e-04 - val_accuracy: 1.0000\n",
      "Epoch 399/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.1319e-04 - accuracy: 0.9979 - val_loss: 3.0463e-04 - val_accuracy: 1.0000\n",
      "Epoch 400/400\n",
      "721/721 [==============================] - 5s 7ms/step - loss: 2.0512e-04 - accuracy: 0.9984 - val_loss: 1.6234e-04 - val_accuracy: 0.9983\n"
     ]
    }
   ],
   "source": [
    "Batch_Size = 8\n",
    "Epoch_Anz = 400\n",
    "Shift_Range = 3\n",
    "Brightness_Range = 0.3\n",
    "\n",
    "datagen = ImageDataGenerator(width_shift_range=[-Shift_Range,Shift_Range], height_shift_range=[-Shift_Range,Shift_Range],brightness_range=[1-Brightness_Range,1+Brightness_Range])\n",
    "\n",
    "if (Training_Percentage > 0):\n",
    "    train_iterator = datagen.flow(x_data, y_data, batch_size=Batch_Size)\n",
    "    validation_iterator = datagen.flow(X_test, y_test, batch_size=Batch_Size)\n",
    "    history = model.fit_generator(train_iterator, validation_data = validation_iterator, epochs = Epoch_Anz)\n",
    "else:\n",
    "    train_iterator = datagen.flow(x_data, y_data, batch_size=Batch_Size)\n",
    "    history = model.fit_generator(train_iterator, epochs = Epoch_Anz)\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Overall Learing results (Step 1 & Step 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "loss_ges = np.append(loss_ges, history.history['loss'])\n",
    "\n",
    "\n",
    "if (Training_Percentage > 0):\n",
    "    val_loss_ges = np.append(val_loss_ges, history.history['val_loss'])\n",
    "    plt.semilogy(val_loss_ges)\n",
    "\n",
    "\n",
    "plt.semilogy(loss_ges)\n",
    "plt.title('model loss')\n",
    "plt.ylabel('loss')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train','eval'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check the model by hand\n",
    "\n",
    "* The following code uses the trained model to check the deviation for each picture.\n",
    "* The evaluation takes the periodic character of the results into account (dev1 ... dev2).\n",
    "* Images, that have a bigger deviation as the parameter \"deviation_max_list\" are printed in a list to check the picture and labeling itself"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "data_resize_all/0.1_PRODUCED_ANALOG_2.jpg 0.01 0.19911532531326115 -0.18911532531326114\n"
     ]
    }
   ],
   "source": [
    "Input_dir='data_resize_all'\n",
    "#Input_dir='test_result'\n",
    "files = sorted(glob.glob(Input_dir + '/*.*'))\n",
    "res = []\n",
    "stat_Anz = []\n",
    "stat_Abweichung = []\n",
    "i = 0\n",
    "deviation_max_list = 0.05\n",
    "\n",
    "for i in range(100):\n",
    "    stat_Anz.append(0)\n",
    "    stat_Abweichung.append(0)\n",
    "\n",
    "for aktfile in sorted(files):\n",
    "    base = os.path.basename(aktfile)\n",
    "    target = (float(base[0:3])) / 10\n",
    "    \n",
    "    target_sin = math.sin(target * math.pi * 2)\n",
    "    target_cos = math.cos(target * math.pi * 2)\n",
    "\n",
    "    test_image = Image.open(aktfile)\n",
    "    test_image = np.array(test_image, dtype=\"float32\")\n",
    "    img = np.reshape(test_image,[1,32,32,3])\n",
    "    classes = model.predict(img)\n",
    "    \n",
    "    out_sin = classes[0][0]  \n",
    "    out_cos = classes[0][1]\n",
    "    out_target = (np.arctan2(out_sin, out_cos)/(2*math.pi)) % 1\n",
    "\n",
    "    dev_sin = target_sin - out_sin\n",
    "    dev_cos = target_cos - out_cos\n",
    "    dev_target = target - out_target\n",
    "    \n",
    "    if abs(dev_target + 1) < abs(dev_target):\n",
    "        out_target = out_target - 1\n",
    "        dev_target = target - out_target\n",
    "    else:\n",
    "        if abs(dev_target - 1) < abs(dev_target):\n",
    "            out_target = out_target + 1\n",
    "            dev_target = target - out_target\n",
    "            \n",
    "    target_int = int ((float(base[0:3])) * 10)\n",
    "    stat_Abweichung[target_int] = stat_Abweichung[target_int] + dev_target  \n",
    "    stat_Anz[target_int] = stat_Anz[target_int] + 1\n",
    "               \n",
    "    res.append(np.array([target, out_target, dev_target, out_sin, out_cos, i]))\n",
    "    if abs(dev_target) > deviation_max_list:\n",
    "        print(aktfile + \" \" + str(target) + \" \" + str(out_target) +  \" \" + str(dev_target))\n",
    "\n",
    "    \n",
    "for i in range(100):\n",
    "    stat_Abweichung[i] = stat_Abweichung[i] / stat_Anz[i]\n",
    "\n",
    "res = np.asarray(res)\n",
    "res_step_1 = res"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(res[:,3])\n",
    "plt.plot(res[:,4])\n",
    "plt.title('Result')\n",
    "plt.ylabel('value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['sin', 'cos'], loc='lower left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(res[:,0])\n",
    "plt.plot(res[:,1])\n",
    "plt.title('Result')\n",
    "plt.ylabel('Counter Value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['Orginal', 'Prediction'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(stat_Abweichung)\n",
    "plt.title('Result')\n",
    "plt.ylabel('Counter Value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['Orginal', 'Prediction'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(stat_Anz)\n",
    "plt.title('Result')\n",
    "plt.ylabel('Counter Value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['Orginal', 'Prediction'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Deviation from Expected Value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.0012  0.0089 -0.1891  0.0495]\n"
     ]
    }
   ],
   "source": [
    "plt.plot(res[:,2])\n",
    "plt.title('Deviation')\n",
    "plt.ylabel('Deviation from expected value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['Deviation'], loc='upper left')\n",
    "#plt.ylim(-0.3, 0.3)\n",
    "plt.show()\n",
    "\n",
    "statistic = np.array([np.mean(res[:,2]), np.std(res[:,2]), np.min(res[:,2]), np.max(res[:,2])])\n",
    "print(statistic)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save the model\n",
    "\n",
    "* Save the model to the file with the \"h5\" file format"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-07-15 21:05:44.598346: W tensorflow/python/util/util.cc:368] Sets are not currently considered sequences, but this may change in the future, so consider avoiding using them.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmp7prxuda2/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "2022-07-15 21:05:48.329050: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:357] Ignored output_format.\n",
      "2022-07-15 21:05:48.329063: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:360] Ignored drop_control_dependency.\n",
      "WARNING:absl:Buffer deduplication procedure will be skipped when flatbuffer library is not properly loaded\n",
      "2022-07-15 21:05:48.330136: I tensorflow/cc/saved_model/reader.cc:43] Reading SavedModel from: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmp7prxuda2\n",
      "2022-07-15 21:05:48.332577: I tensorflow/cc/saved_model/reader.cc:78] Reading meta graph with tags { serve }\n",
      "2022-07-15 21:05:48.332593: I tensorflow/cc/saved_model/reader.cc:119] Reading SavedModel debug info (if present) from: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmp7prxuda2\n",
      "2022-07-15 21:05:48.341646: I tensorflow/cc/saved_model/loader.cc:228] Restoring SavedModel bundle.\n",
      "2022-07-15 21:05:48.414782: I tensorflow/cc/saved_model/loader.cc:212] Running initialization op on SavedModel bundle at path: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmp7prxuda2\n",
      "2022-07-15 21:05:48.436857: I tensorflow/cc/saved_model/loader.cc:301] SavedModel load for tags { serve }; Status: success: OK. Took 106725 microseconds.\n",
      "2022-07-15 21:05:48.484134: I tensorflow/compiler/mlir/tensorflow/utils/dump_mlir_util.cc:237] disabling MLIR crash reproducer, set env var `MLIR_CRASH_REPRODUCER_DIRECTORY` to enable.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "204200"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "FileName = TFliteNamingAndVersion\n",
    "\n",
    "converter = tf.lite.TFLiteConverter.from_keras_model(model)\n",
    "tflite_model = converter.convert()\n",
    "open(FileName + \".tflite\", \"wb\").write(tflite_model)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmpxrq44piw/assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmpxrq44piw/assets\n",
      "/opt/homebrew/Caskroom/miniforge/base/envs/mlp/lib/python3.9/site-packages/tensorflow/lite/python/convert.py:746: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway.\n",
      "  warnings.warn(\"Statistics for quantized inputs were expected, but not \"\n",
      "2022-07-15 21:05:52.578500: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:357] Ignored output_format.\n",
      "2022-07-15 21:05:52.578520: W tensorflow/compiler/mlir/lite/python/tf_tfl_flatbuffer_helpers.cc:360] Ignored drop_control_dependency.\n",
      "2022-07-15 21:05:52.578604: I tensorflow/cc/saved_model/reader.cc:43] Reading SavedModel from: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmpxrq44piw\n",
      "2022-07-15 21:05:52.580020: I tensorflow/cc/saved_model/reader.cc:78] Reading meta graph with tags { serve }\n",
      "2022-07-15 21:05:52.580027: I tensorflow/cc/saved_model/reader.cc:119] Reading SavedModel debug info (if present) from: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmpxrq44piw\n",
      "2022-07-15 21:05:52.586347: I tensorflow/cc/saved_model/loader.cc:228] Restoring SavedModel bundle.\n",
      "2022-07-15 21:05:52.659650: I tensorflow/cc/saved_model/loader.cc:212] Running initialization op on SavedModel bundle at path: /var/folders/07/sj9pw71n205_354358yc8scm0000gn/T/tmpxrq44piw\n",
      "2022-07-15 21:05:52.681743: I tensorflow/cc/saved_model/loader.cc:301] SavedModel load for tags { serve }; Status: success: OK. Took 103141 microseconds.\n",
      "fully_quantize: 0, inference_type: 6, input_inference_type: 0, output_inference_type: 0\n",
      "WARNING:absl:Buffer deduplication procedure will be skipped when flatbuffer library is not properly loaded\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ana0910s1_longq.tflite\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "58744"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pathlib import Path\n",
    "import tensorflow as tf\n",
    "\n",
    "FileName = TFliteNamingAndVersion + \"q\" + \".tflite\"\n",
    "\n",
    "def representative_dataset():\n",
    "    for n in range(x_data[0].size):\n",
    "      data = np.expand_dims(x_data[n], axis=0)\n",
    "      yield [data.astype(np.float32)]\n",
    "        \n",
    "converter2 = tf.lite.TFLiteConverter.from_keras_model(model)\n",
    "converter2.representative_dataset = representative_dataset\n",
    "converter2.optimizations = [tf.lite.Optimize.DEFAULT]\n",
    "converter2.representative_dataset = representative_dataset\n",
    "tflite_quant_model = converter2.convert()\n",
    "\n",
    "open(FileName, \"wb\").write(tflite_quant_model)\n",
    "print(FileName)\n",
    "Path(FileName).stat().st_size"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.16"
  },
  "vscode": {
   "interpreter": {
    "hash": "4976e0179d97dd6d59b1329a76e601e17b789c2571b41c8b57f5fd69821c0dd3"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}