Train_CNN_Digital-Readout-Small-v3.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# CNN Training\n",
    "\n",
    "Target of this code is to train a CNN network to classify images of a digital readout to the digits 0 to 9. Additionally a category \"NaN\" is introduced, to mark images that are not amibiguous.\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 = \"dig1410s3\"   # Used for tflite Filename\n",
    "Training_Percentage = 0.0              # 0.0 = Use all Images for Training\n",
    "Epoch_Anz = 500\n",
    "\n",
    "##########################################################################\n",
    "\n",
    "\n",
    "import tensorflow as tf\n",
    "import matplotlib.pyplot as plt\n",
    "import glob\n",
    "import os\n",
    "import numpy as np\n",
    "from sklearn.utils import shuffle\n",
    "from tensorflow.python import keras\n",
    "from tensorflow.python.keras import Sequential\n",
    "from tensorflow.keras.layers import Dense, InputLayer, Conv2D, MaxPool2D, Flatten, BatchNormalization\n",
    "from tensorflow.keras.preprocessing.image import ImageDataGenerator\n",
    "from sklearn.model_selection import train_test_split\n",
    "from tensorflow.keras.callbacks import History \n",
    "from tensorflow.keras.utils import to_categorical\n",
    "from PIL import Image \n",
    "from pathlib import Path\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",
    "* Inside subdirectories are expected from -1, 0, 1, ... 9 in which the pictures are sorted according to their values (=category)\n",
    "* Picture size must be 20x32 with 3 color channels (RGB)\n",
    "* The filename can be arbitrary\n",
    "\n",
    "* The images are stored in the x_data[]\n",
    "* The expected category for each image in the corresponding y_data[]\n",
    "\n",
    "* The last step is a shuffle (from sklearn.utils) and split the data into training and validation data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(1325, 32, 20, 3)\n",
      "(1325, 11)\n"
     ]
    }
   ],
   "source": [
    "Input_dir='ziffer_sortiert_resize'\n",
    "\n",
    "files = glob.glob(Input_dir + '/*.jpg')\n",
    "x_data = []\n",
    "y_data = []\n",
    "\n",
    "for aktfile in files:\n",
    "    base = os.path.basename(aktfile)\n",
    "    target = base[0:1]\n",
    "    if target == \"N\":\n",
    "        category = 10                # NaN does not work --> convert to 10\n",
    "    else:\n",
    "        category = int(target)\n",
    "    test_image = Image.open(aktfile)\n",
    "    test_image = np.array(test_image, dtype=\"float32\")\n",
    "    x_data.append(test_image)\n",
    "    y_data.append(np.array([category]))\n",
    "\n",
    "x_data = np.array(x_data)\n",
    "y_data = np.array(y_data)\n",
    "y_data = to_categorical(y_data, 11)\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, 20, 3)\n",
    "* Number of output layers: 11\n",
    "* As loss function \"categorical_crossentropy\" is choosen, as it is a categories task"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " input_1 (InputLayer)        [(None, 32, 20, 3)]       0         \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 32, 20, 3)        12        \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " conv2d (Conv2D)             (None, 32, 20, 32)        896       \n",
      "                                                                 \n",
      " max_pooling2d (MaxPooling2D  (None, 16, 10, 32)       0         \n",
      " )                                                               \n",
      "                                                                 \n",
      " conv2d_1 (Conv2D)           (None, 16, 10, 32)        9248      \n",
      "                                                                 \n",
      " max_pooling2d_1 (MaxPooling  (None, 8, 5, 32)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " conv2d_2 (Conv2D)           (None, 8, 5, 32)          9248      \n",
      "                                                                 \n",
      " max_pooling2d_2 (MaxPooling  (None, 4, 2, 32)         0         \n",
      " 2D)                                                             \n",
      "                                                                 \n",
      " flatten (Flatten)           (None, 256)               0         \n",
      "                                                                 \n",
      " dense (Dense)               (None, 128)               32896     \n",
      "                                                                 \n",
      " dense_1 (Dense)             (None, 11)                1419      \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 53,719\n",
      "Trainable params: 53,713\n",
      "Non-trainable params: 6\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "inputs = tf.keras.Input(shape=(32, 20, 3))\n",
    "inputs2 = tf.keras.layers.BatchNormalization()(inputs)\n",
    "inputs3 = tf.keras.layers.Conv2D(32, (3, 3), padding='same', activation=\"relu\")(inputs2)\n",
    "inputs4 = tf.keras.layers.MaxPool2D(pool_size=(2,2))(inputs3)\n",
    "inputs5 = tf.keras.layers.Conv2D(32, (3, 3), padding='same', activation=\"relu\")(inputs4)\n",
    "inputs6 = tf.keras.layers.MaxPool2D(pool_size=(2,2))(inputs5)\n",
    "inputs7 = tf.keras.layers.Conv2D(32, (3, 3), padding='same', activation=\"relu\")(inputs6)\n",
    "inputs8 = tf.keras.layers.MaxPool2D(pool_size=(2,2))(inputs7)\n",
    "inputs9 = tf.keras.layers.Flatten()(inputs8)\n",
    "inputs10 = tf.keras.layers.Dense(128,activation=\"relu\")(inputs9)\n",
    "output = tf.keras.layers.Dense(11, activation='softmax')(inputs10)\n",
    "\n",
    "model = tf.keras.Model(inputs=inputs, outputs=output)\n",
    "\n",
    "model.compile(loss=keras.losses.categorical_crossentropy, optimizer=tf.keras.optimizers.Adadelta(learning_rate=1.0, rho=0.95), metrics = [\"accuracy\"])\n",
    "\n",
    "model.summary()\n",
    "\n",
    "#model = Sequential()\n",
    "#model.add(BatchNormalization(input_shape=(32,20,3)))\n",
    "#model.add(Conv2D(32, (3, 3), padding='same', activation=\"relu\"))\n",
    "#model.add(MaxPool2D(pool_size=(2,2)))\n",
    "#model.add(Conv2D(32, (3, 3), padding='same', activation=\"relu\"))\n",
    "#model.add(MaxPool2D(pool_size=(2,2)))\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(11, activation = \"softmax\"))\n",
    "#model.summary()\n",
    "\n",
    "#model.compile(loss=keras.losses.categorical_crossentropy, 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, pixel shift variations and rotation angle. This is implemented with a ImageDataGenerator.\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/500\n",
      "332/332 [==============================] - 4s 9ms/step - loss: 2.1510 - accuracy: 0.3608\n",
      "Epoch 2/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 1.7540 - accuracy: 0.4483\n",
      "Epoch 3/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 1.3917 - accuracy: 0.5449\n",
      "Epoch 4/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 1.0348 - accuracy: 0.6589\n",
      "Epoch 5/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.9221 - accuracy: 0.7223\n",
      "Epoch 6/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.8009 - accuracy: 0.7577\n",
      "Epoch 7/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.7159 - accuracy: 0.7713\n",
      "Epoch 8/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.6742 - accuracy: 0.7758\n",
      "Epoch 9/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.6520 - accuracy: 0.7955\n",
      "Epoch 10/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.5178 - accuracy: 0.8400\n",
      "Epoch 11/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.5563 - accuracy: 0.8332\n",
      "Epoch 12/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.4928 - accuracy: 0.8385\n",
      "Epoch 13/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.4947 - accuracy: 0.8423\n",
      "Epoch 14/500\n",
      "332/332 [==============================] - 5s 13ms/step - loss: 0.4857 - accuracy: 0.8513\n",
      "Epoch 15/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.4602 - accuracy: 0.8626\n",
      "Epoch 16/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.4367 - accuracy: 0.8687\n",
      "Epoch 17/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.4362 - accuracy: 0.8785\n",
      "Epoch 18/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.4339 - accuracy: 0.8717\n",
      "Epoch 19/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.4535 - accuracy: 0.8740\n",
      "Epoch 20/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.4089 - accuracy: 0.8725\n",
      "Epoch 21/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.3593 - accuracy: 0.8823\n",
      "Epoch 22/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.3652 - accuracy: 0.8898\n",
      "Epoch 23/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.3607 - accuracy: 0.8913\n",
      "Epoch 24/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.3887 - accuracy: 0.8785\n",
      "Epoch 25/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.3116 - accuracy: 0.9049\n",
      "Epoch 26/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.3259 - accuracy: 0.8989\n",
      "Epoch 27/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.3344 - accuracy: 0.9064\n",
      "Epoch 28/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2680 - accuracy: 0.9162\n",
      "Epoch 29/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.3185 - accuracy: 0.9132\n",
      "Epoch 30/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2786 - accuracy: 0.9125\n",
      "Epoch 31/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2969 - accuracy: 0.9042\n",
      "Epoch 32/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2922 - accuracy: 0.9117\n",
      "Epoch 33/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2902 - accuracy: 0.9215\n",
      "Epoch 34/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2942 - accuracy: 0.9125\n",
      "Epoch 35/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.2887 - accuracy: 0.9034\n",
      "Epoch 36/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2650 - accuracy: 0.9185\n",
      "Epoch 37/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2333 - accuracy: 0.9238\n",
      "Epoch 38/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2630 - accuracy: 0.9268\n",
      "Epoch 39/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.2486 - accuracy: 0.9306\n",
      "Epoch 40/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2254 - accuracy: 0.9283\n",
      "Epoch 41/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2230 - accuracy: 0.9374\n",
      "Epoch 42/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2213 - accuracy: 0.9336\n",
      "Epoch 43/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2158 - accuracy: 0.9291\n",
      "Epoch 44/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2440 - accuracy: 0.9245\n",
      "Epoch 45/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2329 - accuracy: 0.9260\n",
      "Epoch 46/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2385 - accuracy: 0.9230\n",
      "Epoch 47/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2114 - accuracy: 0.9321\n",
      "Epoch 48/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2427 - accuracy: 0.9366\n",
      "Epoch 49/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2213 - accuracy: 0.9306\n",
      "Epoch 50/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2251 - accuracy: 0.9381\n",
      "Epoch 51/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2069 - accuracy: 0.9396\n",
      "Epoch 52/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1998 - accuracy: 0.9374\n",
      "Epoch 53/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2477 - accuracy: 0.9283\n",
      "Epoch 54/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1699 - accuracy: 0.9472\n",
      "Epoch 55/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2268 - accuracy: 0.9381\n",
      "Epoch 56/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1769 - accuracy: 0.9502\n",
      "Epoch 57/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2262 - accuracy: 0.9374\n",
      "Epoch 58/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1572 - accuracy: 0.9487\n",
      "Epoch 59/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1852 - accuracy: 0.9396\n",
      "Epoch 60/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2148 - accuracy: 0.9268\n",
      "Epoch 61/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1815 - accuracy: 0.9494\n",
      "Epoch 62/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1765 - accuracy: 0.9494\n",
      "Epoch 63/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2054 - accuracy: 0.9479\n",
      "Epoch 64/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2154 - accuracy: 0.9366\n",
      "Epoch 65/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1593 - accuracy: 0.9472\n",
      "Epoch 66/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1404 - accuracy: 0.9517\n",
      "Epoch 67/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1929 - accuracy: 0.9442\n",
      "Epoch 68/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1880 - accuracy: 0.9449\n",
      "Epoch 69/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2505 - accuracy: 0.9366\n",
      "Epoch 70/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.2316 - accuracy: 0.9306\n",
      "Epoch 71/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1466 - accuracy: 0.9540\n",
      "Epoch 72/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2220 - accuracy: 0.9434\n",
      "Epoch 73/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1638 - accuracy: 0.9547\n",
      "Epoch 74/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2008 - accuracy: 0.9479\n",
      "Epoch 75/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1706 - accuracy: 0.9532\n",
      "Epoch 76/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1481 - accuracy: 0.9540\n",
      "Epoch 77/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1511 - accuracy: 0.9494\n",
      "Epoch 78/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1370 - accuracy: 0.9577\n",
      "Epoch 79/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1611 - accuracy: 0.9540\n",
      "Epoch 80/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1724 - accuracy: 0.9442\n",
      "Epoch 81/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1621 - accuracy: 0.9532\n",
      "Epoch 82/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1731 - accuracy: 0.9494\n",
      "Epoch 83/500\n",
      "332/332 [==============================] - 4s 10ms/step - loss: 0.1818 - accuracy: 0.9494\n",
      "Epoch 84/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1654 - accuracy: 0.9540\n",
      "Epoch 85/500\n",
      "332/332 [==============================] - 4s 10ms/step - loss: 0.1569 - accuracy: 0.9555\n",
      "Epoch 86/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1363 - accuracy: 0.9570\n",
      "Epoch 87/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1671 - accuracy: 0.9540\n",
      "Epoch 88/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1778 - accuracy: 0.9502\n",
      "Epoch 89/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1464 - accuracy: 0.9592\n",
      "Epoch 90/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1673 - accuracy: 0.9592\n",
      "Epoch 91/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1535 - accuracy: 0.9585\n",
      "Epoch 92/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1370 - accuracy: 0.9577\n",
      "Epoch 93/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1785 - accuracy: 0.9479\n",
      "Epoch 94/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1330 - accuracy: 0.9547\n",
      "Epoch 95/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1324 - accuracy: 0.9555\n",
      "Epoch 96/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1440 - accuracy: 0.9600\n",
      "Epoch 97/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1289 - accuracy: 0.9668\n",
      "Epoch 98/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1522 - accuracy: 0.9517\n",
      "Epoch 99/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1206 - accuracy: 0.9615\n",
      "Epoch 100/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1546 - accuracy: 0.9555\n",
      "Epoch 101/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.2216 - accuracy: 0.9411\n",
      "Epoch 102/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1660 - accuracy: 0.9540\n",
      "Epoch 103/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1495 - accuracy: 0.9577\n",
      "Epoch 104/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1636 - accuracy: 0.9547\n",
      "Epoch 105/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1355 - accuracy: 0.9577\n",
      "Epoch 106/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1528 - accuracy: 0.9668\n",
      "Epoch 107/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1475 - accuracy: 0.9608\n",
      "Epoch 108/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1368 - accuracy: 0.9623\n",
      "Epoch 109/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1874 - accuracy: 0.9547\n",
      "Epoch 110/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1643 - accuracy: 0.9517\n",
      "Epoch 111/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1608 - accuracy: 0.9532\n",
      "Epoch 112/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1783 - accuracy: 0.9487\n",
      "Epoch 113/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1569 - accuracy: 0.9494\n",
      "Epoch 114/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1262 - accuracy: 0.9600\n",
      "Epoch 115/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1434 - accuracy: 0.9623\n",
      "Epoch 116/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1186 - accuracy: 0.9623\n",
      "Epoch 117/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1686 - accuracy: 0.9585\n",
      "Epoch 118/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1695 - accuracy: 0.9502\n",
      "Epoch 119/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1197 - accuracy: 0.9653\n",
      "Epoch 120/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1502 - accuracy: 0.9608\n",
      "Epoch 121/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1545 - accuracy: 0.9532\n",
      "Epoch 122/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1081 - accuracy: 0.9638\n",
      "Epoch 123/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1210 - accuracy: 0.9675\n",
      "Epoch 124/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1314 - accuracy: 0.9600\n",
      "Epoch 125/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1411 - accuracy: 0.9577\n",
      "Epoch 126/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1465 - accuracy: 0.9592\n",
      "Epoch 127/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1327 - accuracy: 0.9615\n",
      "Epoch 128/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1183 - accuracy: 0.9585\n",
      "Epoch 129/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1483 - accuracy: 0.9668\n",
      "Epoch 130/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1147 - accuracy: 0.9728\n",
      "Epoch 131/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1329 - accuracy: 0.9638\n",
      "Epoch 132/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1182 - accuracy: 0.9638\n",
      "Epoch 133/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1192 - accuracy: 0.9630\n",
      "Epoch 134/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1193 - accuracy: 0.9638\n",
      "Epoch 135/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1504 - accuracy: 0.9630\n",
      "Epoch 136/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1267 - accuracy: 0.9645\n",
      "Epoch 137/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1101 - accuracy: 0.9706\n",
      "Epoch 138/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0998 - accuracy: 0.9653\n",
      "Epoch 139/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1447 - accuracy: 0.9600\n",
      "Epoch 140/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1355 - accuracy: 0.9608\n",
      "Epoch 141/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1442 - accuracy: 0.9623\n",
      "Epoch 142/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1024 - accuracy: 0.9721\n",
      "Epoch 143/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1360 - accuracy: 0.9577\n",
      "Epoch 144/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1072 - accuracy: 0.9668\n",
      "Epoch 145/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1346 - accuracy: 0.9630\n",
      "Epoch 146/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1276 - accuracy: 0.9570\n",
      "Epoch 147/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0857 - accuracy: 0.9713\n",
      "Epoch 148/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1508 - accuracy: 0.9540\n",
      "Epoch 149/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1129 - accuracy: 0.9675\n",
      "Epoch 150/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1570 - accuracy: 0.9592\n",
      "Epoch 151/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0925 - accuracy: 0.9691\n",
      "Epoch 152/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1237 - accuracy: 0.9653\n",
      "Epoch 153/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1677 - accuracy: 0.9592\n",
      "Epoch 154/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1193 - accuracy: 0.9608\n",
      "Epoch 155/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1341 - accuracy: 0.9608\n",
      "Epoch 156/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1171 - accuracy: 0.9675\n",
      "Epoch 157/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1209 - accuracy: 0.9615\n",
      "Epoch 158/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1345 - accuracy: 0.9585\n",
      "Epoch 159/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1328 - accuracy: 0.9592\n",
      "Epoch 160/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1069 - accuracy: 0.9698\n",
      "Epoch 161/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0892 - accuracy: 0.9691\n",
      "Epoch 162/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1032 - accuracy: 0.9683\n",
      "Epoch 163/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1463 - accuracy: 0.9615\n",
      "Epoch 164/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1468 - accuracy: 0.9585\n",
      "Epoch 165/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1221 - accuracy: 0.9653\n",
      "Epoch 166/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1052 - accuracy: 0.9713\n",
      "Epoch 167/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1127 - accuracy: 0.9660\n",
      "Epoch 168/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1138 - accuracy: 0.9683\n",
      "Epoch 169/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1255 - accuracy: 0.9615\n",
      "Epoch 170/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1611 - accuracy: 0.9570\n",
      "Epoch 171/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1191 - accuracy: 0.9668\n",
      "Epoch 172/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1281 - accuracy: 0.9645\n",
      "Epoch 173/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1176 - accuracy: 0.9608\n",
      "Epoch 174/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1167 - accuracy: 0.9683\n",
      "Epoch 175/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1210 - accuracy: 0.9653\n",
      "Epoch 176/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0885 - accuracy: 0.9698\n",
      "Epoch 177/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1257 - accuracy: 0.9638\n",
      "Epoch 178/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1280 - accuracy: 0.9653\n",
      "Epoch 179/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0971 - accuracy: 0.9668\n",
      "Epoch 180/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1800 - accuracy: 0.9562\n",
      "Epoch 181/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1072 - accuracy: 0.9683\n",
      "Epoch 182/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1208 - accuracy: 0.9668\n",
      "Epoch 183/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1071 - accuracy: 0.9660\n",
      "Epoch 184/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1272 - accuracy: 0.9698\n",
      "Epoch 185/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0850 - accuracy: 0.9698\n",
      "Epoch 186/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1183 - accuracy: 0.9653\n",
      "Epoch 187/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1105 - accuracy: 0.9683\n",
      "Epoch 188/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1132 - accuracy: 0.9698\n",
      "Epoch 189/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1517 - accuracy: 0.9623\n",
      "Epoch 190/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0955 - accuracy: 0.9728\n",
      "Epoch 191/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1374 - accuracy: 0.9623\n",
      "Epoch 192/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0955 - accuracy: 0.9736\n",
      "Epoch 193/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1048 - accuracy: 0.9743\n",
      "Epoch 194/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1381 - accuracy: 0.9615\n",
      "Epoch 195/500\n",
      "332/332 [==============================] - 4s 10ms/step - loss: 0.0987 - accuracy: 0.9698\n",
      "Epoch 196/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0725 - accuracy: 0.9789\n",
      "Epoch 197/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0990 - accuracy: 0.9683\n",
      "Epoch 198/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1457 - accuracy: 0.9675\n",
      "Epoch 199/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0898 - accuracy: 0.9713\n",
      "Epoch 200/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1085 - accuracy: 0.9691\n",
      "Epoch 201/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1176 - accuracy: 0.9713\n",
      "Epoch 202/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1097 - accuracy: 0.9675\n",
      "Epoch 203/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0982 - accuracy: 0.9721\n",
      "Epoch 204/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1174 - accuracy: 0.9653\n",
      "Epoch 205/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1077 - accuracy: 0.9706\n",
      "Epoch 206/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0933 - accuracy: 0.9713\n",
      "Epoch 207/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0858 - accuracy: 0.9758\n",
      "Epoch 208/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1187 - accuracy: 0.9675\n",
      "Epoch 209/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1090 - accuracy: 0.9668\n",
      "Epoch 210/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1016 - accuracy: 0.9706\n",
      "Epoch 211/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1234 - accuracy: 0.9615\n",
      "Epoch 212/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0816 - accuracy: 0.9804\n",
      "Epoch 213/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1002 - accuracy: 0.9608\n",
      "Epoch 214/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1108 - accuracy: 0.9706\n",
      "Epoch 215/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0782 - accuracy: 0.9736\n",
      "Epoch 216/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1138 - accuracy: 0.9675\n",
      "Epoch 217/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1007 - accuracy: 0.9758\n",
      "Epoch 218/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1263 - accuracy: 0.9728\n",
      "Epoch 219/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1285 - accuracy: 0.9668\n",
      "Epoch 220/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0813 - accuracy: 0.9706\n",
      "Epoch 221/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1038 - accuracy: 0.9630\n",
      "Epoch 222/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1046 - accuracy: 0.9728\n",
      "Epoch 223/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0991 - accuracy: 0.9660\n",
      "Epoch 224/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1302 - accuracy: 0.9653\n",
      "Epoch 225/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0842 - accuracy: 0.9751\n",
      "Epoch 226/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1080 - accuracy: 0.9691\n",
      "Epoch 227/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0741 - accuracy: 0.9766\n",
      "Epoch 228/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1097 - accuracy: 0.9668\n",
      "Epoch 229/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0839 - accuracy: 0.9758\n",
      "Epoch 230/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0873 - accuracy: 0.9706\n",
      "Epoch 231/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1002 - accuracy: 0.9736\n",
      "Epoch 232/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1299 - accuracy: 0.9653\n",
      "Epoch 233/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0991 - accuracy: 0.9713\n",
      "Epoch 234/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0699 - accuracy: 0.9789\n",
      "Epoch 235/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1275 - accuracy: 0.9623\n",
      "Epoch 236/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1009 - accuracy: 0.9706\n",
      "Epoch 237/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1028 - accuracy: 0.9713\n",
      "Epoch 238/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0736 - accuracy: 0.9758\n",
      "Epoch 239/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1051 - accuracy: 0.9774\n",
      "Epoch 240/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0788 - accuracy: 0.9811\n",
      "Epoch 241/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0918 - accuracy: 0.9796\n",
      "Epoch 242/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0827 - accuracy: 0.9743\n",
      "Epoch 243/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.1438 - accuracy: 0.9630\n",
      "Epoch 244/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0965 - accuracy: 0.9728\n",
      "Epoch 245/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0872 - accuracy: 0.9781\n",
      "Epoch 246/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1011 - accuracy: 0.9691\n",
      "Epoch 247/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0900 - accuracy: 0.9706\n",
      "Epoch 248/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1138 - accuracy: 0.9706\n",
      "Epoch 249/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0787 - accuracy: 0.9766\n",
      "Epoch 250/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1284 - accuracy: 0.9668\n",
      "Epoch 251/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1017 - accuracy: 0.9691\n",
      "Epoch 252/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1033 - accuracy: 0.9706\n",
      "Epoch 253/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1190 - accuracy: 0.9660\n",
      "Epoch 254/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1099 - accuracy: 0.9653\n",
      "Epoch 255/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1000 - accuracy: 0.9706\n",
      "Epoch 256/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1368 - accuracy: 0.9706\n",
      "Epoch 257/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0692 - accuracy: 0.9804\n",
      "Epoch 258/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0935 - accuracy: 0.9675\n",
      "Epoch 259/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0898 - accuracy: 0.9706\n",
      "Epoch 260/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1209 - accuracy: 0.9668\n",
      "Epoch 261/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1204 - accuracy: 0.9630\n",
      "Epoch 262/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0893 - accuracy: 0.9698\n",
      "Epoch 263/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0962 - accuracy: 0.9698\n",
      "Epoch 264/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0788 - accuracy: 0.9736\n",
      "Epoch 265/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0936 - accuracy: 0.9706\n",
      "Epoch 266/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0766 - accuracy: 0.9774\n",
      "Epoch 267/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0755 - accuracy: 0.9758\n",
      "Epoch 268/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1131 - accuracy: 0.9713\n",
      "Epoch 269/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0884 - accuracy: 0.9766\n",
      "Epoch 270/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1072 - accuracy: 0.9721\n",
      "Epoch 271/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0975 - accuracy: 0.9804\n",
      "Epoch 272/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0871 - accuracy: 0.9736\n",
      "Epoch 273/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0743 - accuracy: 0.9751\n",
      "Epoch 274/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0765 - accuracy: 0.9826\n",
      "Epoch 275/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0635 - accuracy: 0.9796\n",
      "Epoch 276/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0815 - accuracy: 0.9774\n",
      "Epoch 277/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0873 - accuracy: 0.9766\n",
      "Epoch 278/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1037 - accuracy: 0.9758\n",
      "Epoch 279/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0682 - accuracy: 0.9804\n",
      "Epoch 280/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.1000 - accuracy: 0.9728\n",
      "Epoch 281/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1291 - accuracy: 0.9706\n",
      "Epoch 282/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1347 - accuracy: 0.9675\n",
      "Epoch 283/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0911 - accuracy: 0.9774\n",
      "Epoch 284/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0886 - accuracy: 0.9743\n",
      "Epoch 285/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1139 - accuracy: 0.9675\n",
      "Epoch 286/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0926 - accuracy: 0.9743\n",
      "Epoch 287/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1060 - accuracy: 0.9743\n",
      "Epoch 288/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0784 - accuracy: 0.9743\n",
      "Epoch 289/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1027 - accuracy: 0.9728\n",
      "Epoch 290/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0938 - accuracy: 0.9736\n",
      "Epoch 291/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0658 - accuracy: 0.9736\n",
      "Epoch 292/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0980 - accuracy: 0.9743\n",
      "Epoch 293/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1024 - accuracy: 0.9713\n",
      "Epoch 294/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0975 - accuracy: 0.9736\n",
      "Epoch 295/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1101 - accuracy: 0.9683\n",
      "Epoch 296/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0934 - accuracy: 0.9668\n",
      "Epoch 297/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0939 - accuracy: 0.9713\n",
      "Epoch 298/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0985 - accuracy: 0.9789\n",
      "Epoch 299/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0661 - accuracy: 0.9796\n",
      "Epoch 300/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1131 - accuracy: 0.9706\n",
      "Epoch 301/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0974 - accuracy: 0.9736\n",
      "Epoch 302/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1025 - accuracy: 0.9758\n",
      "Epoch 303/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0723 - accuracy: 0.9758\n",
      "Epoch 304/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0836 - accuracy: 0.9698\n",
      "Epoch 305/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1037 - accuracy: 0.9781\n",
      "Epoch 306/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0869 - accuracy: 0.9751\n",
      "Epoch 307/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0868 - accuracy: 0.9751\n",
      "Epoch 308/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1176 - accuracy: 0.9721\n",
      "Epoch 309/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1012 - accuracy: 0.9743\n",
      "Epoch 310/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1284 - accuracy: 0.9706\n",
      "Epoch 311/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0993 - accuracy: 0.9743\n",
      "Epoch 312/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1041 - accuracy: 0.9758\n",
      "Epoch 313/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0883 - accuracy: 0.9789\n",
      "Epoch 314/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0640 - accuracy: 0.9834\n",
      "Epoch 315/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1022 - accuracy: 0.9758\n",
      "Epoch 316/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1155 - accuracy: 0.9615\n",
      "Epoch 317/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1046 - accuracy: 0.9660\n",
      "Epoch 318/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0745 - accuracy: 0.9774\n",
      "Epoch 319/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1124 - accuracy: 0.9660\n",
      "Epoch 320/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0718 - accuracy: 0.9796\n",
      "Epoch 321/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1243 - accuracy: 0.9645\n",
      "Epoch 322/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1005 - accuracy: 0.9706\n",
      "Epoch 323/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0594 - accuracy: 0.9781\n",
      "Epoch 324/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1634 - accuracy: 0.9645\n",
      "Epoch 325/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1029 - accuracy: 0.9736\n",
      "Epoch 326/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0845 - accuracy: 0.9781\n",
      "Epoch 327/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1095 - accuracy: 0.9728\n",
      "Epoch 328/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0933 - accuracy: 0.9721\n",
      "Epoch 329/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1020 - accuracy: 0.9728\n",
      "Epoch 330/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0741 - accuracy: 0.9774\n",
      "Epoch 331/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0820 - accuracy: 0.9766\n",
      "Epoch 332/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1009 - accuracy: 0.9728\n",
      "Epoch 333/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1201 - accuracy: 0.9713\n",
      "Epoch 334/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0676 - accuracy: 0.9834\n",
      "Epoch 335/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0997 - accuracy: 0.9751\n",
      "Epoch 336/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.0860 - accuracy: 0.9751\n",
      "Epoch 337/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1129 - accuracy: 0.9766\n",
      "Epoch 338/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0715 - accuracy: 0.9819\n",
      "Epoch 339/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1008 - accuracy: 0.9736\n",
      "Epoch 340/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0716 - accuracy: 0.9758\n",
      "Epoch 341/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1078 - accuracy: 0.9781\n",
      "Epoch 342/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1018 - accuracy: 0.9721\n",
      "Epoch 343/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0973 - accuracy: 0.9698\n",
      "Epoch 344/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1091 - accuracy: 0.9675\n",
      "Epoch 345/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0932 - accuracy: 0.9736\n",
      "Epoch 346/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0838 - accuracy: 0.9781\n",
      "Epoch 347/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0640 - accuracy: 0.9819\n",
      "Epoch 348/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0676 - accuracy: 0.9766\n",
      "Epoch 349/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0832 - accuracy: 0.9758\n",
      "Epoch 350/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1030 - accuracy: 0.9691\n",
      "Epoch 351/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0902 - accuracy: 0.9743\n",
      "Epoch 352/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1118 - accuracy: 0.9713\n",
      "Epoch 353/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0870 - accuracy: 0.9774\n",
      "Epoch 354/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0824 - accuracy: 0.9766\n",
      "Epoch 355/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0872 - accuracy: 0.9781\n",
      "Epoch 356/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0913 - accuracy: 0.9721\n",
      "Epoch 357/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0800 - accuracy: 0.9751\n",
      "Epoch 358/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0691 - accuracy: 0.9789\n",
      "Epoch 359/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0728 - accuracy: 0.9796\n",
      "Epoch 360/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1049 - accuracy: 0.9683\n",
      "Epoch 361/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1001 - accuracy: 0.9811\n",
      "Epoch 362/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0885 - accuracy: 0.9751\n",
      "Epoch 363/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1351 - accuracy: 0.9630\n",
      "Epoch 364/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.0654 - accuracy: 0.9789\n",
      "Epoch 365/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0982 - accuracy: 0.9766\n",
      "Epoch 366/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0652 - accuracy: 0.9796\n",
      "Epoch 367/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1157 - accuracy: 0.9675\n",
      "Epoch 368/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0755 - accuracy: 0.9789\n",
      "Epoch 369/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0956 - accuracy: 0.9743\n",
      "Epoch 370/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1059 - accuracy: 0.9675\n",
      "Epoch 371/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1199 - accuracy: 0.9668\n",
      "Epoch 372/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1022 - accuracy: 0.9743\n",
      "Epoch 373/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.1021 - accuracy: 0.9721\n",
      "Epoch 374/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0520 - accuracy: 0.9872\n",
      "Epoch 375/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1510 - accuracy: 0.9638\n",
      "Epoch 376/500\n",
      "332/332 [==============================] - 4s 10ms/step - loss: 0.0939 - accuracy: 0.9751\n",
      "Epoch 377/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1100 - accuracy: 0.9698\n",
      "Epoch 378/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0684 - accuracy: 0.9781\n",
      "Epoch 379/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0858 - accuracy: 0.9706\n",
      "Epoch 380/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0891 - accuracy: 0.9736\n",
      "Epoch 381/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0725 - accuracy: 0.9811\n",
      "Epoch 382/500\n",
      "332/332 [==============================] - 3s 10ms/step - loss: 0.1081 - accuracy: 0.9698\n",
      "Epoch 383/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0801 - accuracy: 0.9728\n",
      "Epoch 384/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0828 - accuracy: 0.9751\n",
      "Epoch 385/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0903 - accuracy: 0.9721\n",
      "Epoch 386/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0930 - accuracy: 0.9743\n",
      "Epoch 387/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0780 - accuracy: 0.9826\n",
      "Epoch 388/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0930 - accuracy: 0.9713\n",
      "Epoch 389/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0896 - accuracy: 0.9728\n",
      "Epoch 390/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1125 - accuracy: 0.9766\n",
      "Epoch 391/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0805 - accuracy: 0.9758\n",
      "Epoch 392/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0972 - accuracy: 0.9766\n",
      "Epoch 393/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1074 - accuracy: 0.9743\n",
      "Epoch 394/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0814 - accuracy: 0.9811\n",
      "Epoch 395/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0899 - accuracy: 0.9826\n",
      "Epoch 396/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0667 - accuracy: 0.9796\n",
      "Epoch 397/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1065 - accuracy: 0.9698\n",
      "Epoch 398/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0906 - accuracy: 0.9743\n",
      "Epoch 399/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0931 - accuracy: 0.9758\n",
      "Epoch 400/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1060 - accuracy: 0.9721\n",
      "Epoch 401/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1004 - accuracy: 0.9751\n",
      "Epoch 402/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0830 - accuracy: 0.9736\n",
      "Epoch 403/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1039 - accuracy: 0.9743\n",
      "Epoch 404/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1332 - accuracy: 0.9706\n",
      "Epoch 405/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0905 - accuracy: 0.9706\n",
      "Epoch 406/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1035 - accuracy: 0.9706\n",
      "Epoch 407/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0705 - accuracy: 0.9811\n",
      "Epoch 408/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0948 - accuracy: 0.9774\n",
      "Epoch 409/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1041 - accuracy: 0.9789\n",
      "Epoch 410/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0858 - accuracy: 0.9774\n",
      "Epoch 411/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0601 - accuracy: 0.9834\n",
      "Epoch 412/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1062 - accuracy: 0.9736\n",
      "Epoch 413/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0667 - accuracy: 0.9811\n",
      "Epoch 414/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0969 - accuracy: 0.9721\n",
      "Epoch 415/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1207 - accuracy: 0.9660\n",
      "Epoch 416/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0669 - accuracy: 0.9819\n",
      "Epoch 417/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0724 - accuracy: 0.9789\n",
      "Epoch 418/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0862 - accuracy: 0.9758\n",
      "Epoch 419/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0814 - accuracy: 0.9758\n",
      "Epoch 420/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1043 - accuracy: 0.9751\n",
      "Epoch 421/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0878 - accuracy: 0.9766\n",
      "Epoch 422/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0827 - accuracy: 0.9789\n",
      "Epoch 423/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0666 - accuracy: 0.9819\n",
      "Epoch 424/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0986 - accuracy: 0.9736\n",
      "Epoch 425/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1257 - accuracy: 0.9728\n",
      "Epoch 426/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0608 - accuracy: 0.9796\n",
      "Epoch 427/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1230 - accuracy: 0.9630\n",
      "Epoch 428/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0579 - accuracy: 0.9864\n",
      "Epoch 429/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0938 - accuracy: 0.9774\n",
      "Epoch 430/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0813 - accuracy: 0.9758\n",
      "Epoch 431/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0772 - accuracy: 0.9758\n",
      "Epoch 432/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0682 - accuracy: 0.9796\n",
      "Epoch 433/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0806 - accuracy: 0.9804\n",
      "Epoch 434/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0775 - accuracy: 0.9819\n",
      "Epoch 435/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0693 - accuracy: 0.9864\n",
      "Epoch 436/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1432 - accuracy: 0.9691\n",
      "Epoch 437/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.1026 - accuracy: 0.9713\n",
      "Epoch 438/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0911 - accuracy: 0.9751\n",
      "Epoch 439/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0854 - accuracy: 0.9736\n",
      "Epoch 440/500\n",
      "332/332 [==============================] - 5s 13ms/step - loss: 0.1006 - accuracy: 0.9736\n",
      "Epoch 441/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1043 - accuracy: 0.9766\n",
      "Epoch 442/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0537 - accuracy: 0.9857\n",
      "Epoch 443/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1018 - accuracy: 0.9751\n",
      "Epoch 444/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0682 - accuracy: 0.9819\n",
      "Epoch 445/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1072 - accuracy: 0.9721\n",
      "Epoch 446/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.0868 - accuracy: 0.9721\n",
      "Epoch 447/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0961 - accuracy: 0.9736\n",
      "Epoch 448/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.1084 - accuracy: 0.9819\n",
      "Epoch 449/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0840 - accuracy: 0.9751\n",
      "Epoch 450/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1174 - accuracy: 0.9675\n",
      "Epoch 451/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0936 - accuracy: 0.9721\n",
      "Epoch 452/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0758 - accuracy: 0.9736\n",
      "Epoch 453/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0695 - accuracy: 0.9804\n",
      "Epoch 454/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0457 - accuracy: 0.9834\n",
      "Epoch 455/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0957 - accuracy: 0.9789\n",
      "Epoch 456/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0848 - accuracy: 0.9796\n",
      "Epoch 457/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0817 - accuracy: 0.9728\n",
      "Epoch 458/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0573 - accuracy: 0.9834\n",
      "Epoch 459/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0856 - accuracy: 0.9743\n",
      "Epoch 460/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0849 - accuracy: 0.9758\n",
      "Epoch 461/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0856 - accuracy: 0.9796\n",
      "Epoch 462/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0748 - accuracy: 0.9804\n",
      "Epoch 463/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.0852 - accuracy: 0.9751\n",
      "Epoch 464/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0624 - accuracy: 0.9774\n",
      "Epoch 465/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0805 - accuracy: 0.9804\n",
      "Epoch 466/500\n",
      "332/332 [==============================] - 5s 13ms/step - loss: 0.1025 - accuracy: 0.9743\n",
      "Epoch 467/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0816 - accuracy: 0.9811\n",
      "Epoch 468/500\n",
      "332/332 [==============================] - 5s 14ms/step - loss: 0.0847 - accuracy: 0.9811\n",
      "Epoch 469/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0812 - accuracy: 0.9774\n",
      "Epoch 470/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1019 - accuracy: 0.9713\n",
      "Epoch 471/500\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0920 - accuracy: 0.9721\n",
      "Epoch 472/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0720 - accuracy: 0.9826\n",
      "Epoch 473/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1006 - accuracy: 0.9728\n",
      "Epoch 474/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0853 - accuracy: 0.9751\n",
      "Epoch 475/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0766 - accuracy: 0.9804\n",
      "Epoch 476/500\n",
      "332/332 [==============================] - 5s 13ms/step - loss: 0.0728 - accuracy: 0.9774\n",
      "Epoch 477/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0742 - accuracy: 0.9774\n",
      "Epoch 478/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0832 - accuracy: 0.9811\n",
      "Epoch 479/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0856 - accuracy: 0.9774\n",
      "Epoch 480/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0559 - accuracy: 0.9842\n",
      "Epoch 481/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0743 - accuracy: 0.9804\n",
      "Epoch 482/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0745 - accuracy: 0.9774\n",
      "Epoch 483/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0676 - accuracy: 0.9819\n",
      "Epoch 484/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.0748 - accuracy: 0.9796\n",
      "Epoch 485/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0796 - accuracy: 0.9804\n",
      "Epoch 486/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1238 - accuracy: 0.9766\n",
      "Epoch 487/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0894 - accuracy: 0.9751\n",
      "Epoch 488/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0874 - accuracy: 0.9766\n",
      "Epoch 489/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0755 - accuracy: 0.9766\n",
      "Epoch 490/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.1423 - accuracy: 0.9743\n",
      "Epoch 491/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0989 - accuracy: 0.9675\n",
      "Epoch 492/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0583 - accuracy: 0.9834\n",
      "Epoch 493/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0904 - accuracy: 0.9706\n",
      "Epoch 494/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0744 - accuracy: 0.9751\n",
      "Epoch 495/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0942 - accuracy: 0.9706\n",
      "Epoch 496/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1175 - accuracy: 0.9691\n",
      "Epoch 497/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0778 - accuracy: 0.9766\n",
      "Epoch 498/500\n",
      "332/332 [==============================] - 4s 13ms/step - loss: 0.0844 - accuracy: 0.9774\n",
      "Epoch 499/500\n",
      "332/332 [==============================] - 4s 12ms/step - loss: 0.0904 - accuracy: 0.9728\n",
      "Epoch 500/500\n",
      "332/332 [==============================] - 4s 11ms/step - loss: 0.1068 - accuracy: 0.9728\n"
     ]
    }
   ],
   "source": [
    "Batch_Size = 4\n",
    "Shift_Range = 1\n",
    "Brightness_Range = 0.3\n",
    "Rotation_Angle = 10\n",
    "ZoomRange = 0.4\n",
    "\n",
    "datagen = ImageDataGenerator(width_shift_range=[-Shift_Range,Shift_Range], \n",
    "                             height_shift_range=[-Shift_Range,Shift_Range],\n",
    "                             brightness_range=[1-Brightness_Range,1+Brightness_Range],\n",
    "                             zoom_range=[1-ZoomRange, 1+ZoomRange],\n",
    "                             rotation_range=Rotation_Angle)\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(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(train_iterator, epochs = Epoch_Anz)\n",
    "\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Learing result\n",
    " \n",
    "* Visualization of the training and validation results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYoAAAEWCAYAAAB42tAoAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8qNh9FAAAACXBIWXMAAAsTAAALEwEAmpwYAABDZ0lEQVR4nO2dd5wU5f3HP9/t1487DjzqUZUiIE0QC3YEW36WaNREYyTG5KdGUzCaRGNM/CXGJJbYoiYxRjQqRizBAqggSBGQ3tvRDo7rbdvz+2PmmZ2ZnZ3du9u9vdv7vn3x8nZmduaZvb3nM9/6kBACDMMwDBMLR7oHwDAMw3RuWCgYhmEYW1goGIZhGFtYKBiGYRhbWCgYhmEYW1goGIZhGFtYKBgmiRDR34jo1wkeu4eIzmvveRgm1bBQMAzDMLawUDAMwzC2sFAw3Q7V5fNjIvqKiBqI6Hki6k1E7xNRHRF9REQ9dMdfSkQbiaiaiBYT0QjdvlOI6Ev1fa8C8JmudTERrVXf+zkRjWnjmG8hoh1EdJyI3iaiPup2IqI/ElEFEdUS0XoiGq3um0lEm9SxHSCiH7XpA2O6PSwUTHflCgDnAxgO4BIA7wP4GYASKH8XtwMAEQ0H8AqAO9V97wGYT0QeIvIAeAvASwCKAPxbPS/U954C4AUA3wVQDOAZAG8Tkbc1AyWicwD8FsDVAEoB7AUwV919AYAz1fsoUI+pVPc9D+C7Qog8AKMBLGzNdRlGwkLBdFceF0IcEUIcAPAZgC+EEGuEEM0A5gE4RT3u6wDeFUJ8KIQIAHgEQBaA0wBMAeAG8CchREAI8TqAlbprzAbwjBDiCyFESAjxdwAt6vtaw3UAXhBCfCmEaAFwD4CpRFQGIAAgD8BJAEgIsVkIcUh9XwDASCLKF0JUCSG+bOV1GQYACwXTfTmi+7nJ4nWu+nMfKE/wAAAhRBjAfgB91X0HhLGz5l7dzwMB3K26naqJqBpAf/V9rcE8hnooVkNfIcRCAE8AeBJABRE9S0T56qFXAJgJYC8RfUJEU1t5XYYBwELBMPE4CGXCB6DEBKBM9gcAHALQV90mGaD7eT+Ah4QQhbp/2UKIV9o5hhworqwDACCEeEwIMQHASCguqB+r21cKIS4D0AuKi+y1Vl6XYQCwUDBMPF4DMIuIziUiN4C7obiPPgewDEAQwO1E5Cai/wEwWffe5wDcSkSnqkHnHCKaRUR5rRzDKwBuIqJxanzjN1BcZXuIaJJ6fjeABgDNAMJqDOU6IipQXWa1AMLt+ByYbgwLBcPYIITYCuB6AI8DOAYl8H2JEMIvhPAD+B8ANwI4DiWe8abuvasA3ALFNVQFYId6bGvH8BGAnwN4A4oVMwTANerufCiCVAXFPVUJ4PfqvhsA7CGiWgC3Qol1MEyrIV64iGEYhrGDLQqGYRjGFhYKhmEYxhYWCoZhGMYWFgqGYRjGFle6B5AKevbsKcrKytI9DIZhmC7D6tWrjwkhSqz2ZaRQlJWVYdWqVekeBsMwTJeBiPbG2seuJ4ZhGMYWFgqGYRjGFhYKhmEYxpaMjFFYEQgEUF5ejubm5nQPJaX4fD7069cPbrc73UNhGCZD6DZCUV5ejry8PJSVlcHY7DNzEEKgsrIS5eXlGDRoULqHwzBMhtBtXE/Nzc0oLi7OWJEAACJCcXFxxltNDMN0LN1GKABktEhIusM9MgzTsXQroYjHsfoWVDf60z0MhmGYTgULhY7jDX7UNAVScu7q6mr85S9/afX7Zs6cierq6uQPiGEYJkFYKHQ4iBAKp2Z9jlhCEQwGbd/33nvvobCwMCVjYhiGSYRuk/WUCE5H6oRizpw52LlzJ8aNGwe32w2fz4cePXpgy5Yt2LZtGy6//HLs378fzc3NuOOOOzB79mwAkXYk9fX1uOiii3D66afj888/R9++ffGf//wHWVlZKRkvwzCMpFsKxQPzN2LTwdqo7S3BEMJhIMvjbPU5R/bJxy8vGRVz/8MPP4wNGzZg7dq1WLx4MWbNmoUNGzZoaawvvPACioqK0NTUhEmTJuGKK65AcXGx4Rzbt2/HK6+8gueeew5XX3013njjDVx//fWtHivDMExryCjXExFdQkTP1tTUtPUM6KiFYSdPnmyodXjssccwduxYTJkyBfv378f27duj3jNo0CCMGzcOADBhwgTs2bOng0bLMEx3JqMsCiHEfADzJ06ceIvdcbGe/A9WN6GqwY9RfQtSMTwDOTk52s+LFy/GRx99hGXLliE7OxvTp0+3rIXwer3az06nE01NTSkfJ8MwTEZZFO3F6SCEhIAQybcr8vLyUFdXZ7mvpqYGPXr0QHZ2NrZs2YLly5cn/foMwzBtJaMsivbiUIvVwkLAmeTCteLiYkybNg2jR49GVlYWevfure2bMWMGnn76aYwYMQInnngipkyZktRrMwzDtAdKxdNzupk4caIwL1y0efNmjBgxwvZ9lQ0tOFDVhJNOyIfH1XWNrUTulWEYRg8RrRZCTLTa13VnwxTg1FkUDMMwjAILhQ6HQxGKVNVSMAzDdEW6lVDEc7NlgkWRia5EhmHSS7cRCp/Ph8rKStuJVFoU4S5qUcj1KHw+X7qHwjBMBtFtsp769euH8vJyHD16NOYxwXAYR2pa4D/mRo63a340coU7hmGYZNE1Z8M24Ha74676VtMYwCW/+gA/v3gkbh7HK8QxDMMA3cj1lAg5XqXHU32zfUdXhmGY7gQLhQ6X04EstxP1LalZk4JhGKYrwkJhItfnQn0LWxQMwzASFgoTeV4X6tj1xDAMo8FCYYItCoZhGCMsFCZyvS4OZjMMw+hgoTCR62WLgmEYRg8LhYlcjlEwDMMYYKEwkeVxojkQSvcwGIZhOg0sFCayPU40+lkoGIZhJCwUJrLcTjQFQtyFlWEYRoWFwoTPo7TxaAmG0zwShmGYzgELhYlstyIUTex+YhiGAcBCEUWWalE0ckCbYRgGAAtFFD62KBiGYQywUJjIUoWCU2QZhmEUWChMZHuUtZyaWCgYhmEAsFBEkeVRPhKupWAYhlFgoTDBMQqGYRgjLBQmpOuJYxQMwzAKLBQmZDCbYxQMwzAKrnQPIB5ElAPgLwD8ABYLIV5O5fWkUHCMgmEYRiEtFgURvUBEFUS0wbR9BhFtJaIdRDRH3fw/AF4XQtwC4NJUj00W3DX5udU4wzAMkD7X098AzNBvICIngCcBXARgJIBriWgkgH4A9quHpfwx3+NyINfrQmWDP9WXYhiG6RKkRSiEEJ8COG7aPBnADiHELiGEH8BcAJcBKIciFkAHjbdXvhdHaps74lIMwzCdns4UzO6LiOUAKALRF8CbAK4goqcAzI/1ZiKaTUSriGjV0aNH2zWQE/J9OFLb0q5zMAzDZAqdPpgthGgAcFMCxz0L4FkAmDhxYrsWk+id78OK3WaDh2EYpnvSmSyKAwD66173U7d1OL3yvaioa+bFixiGYdC5hGIlgGFENIiIPACuAfB2OgZyQr4PgZDAcQ5oMwzDpC099hUAywCcSETlRHSzECII4AcAFgDYDOA1IcTGdIyvZ64XADjziWEYBmmKUQghro2x/T0A77X1vER0CYBLhg4d2tZTAACKczwAgMp6P9C7XadiGIbp8nQm11O7EULMF0LMLigoaNd5ijWLgjOfGIZhMkookkVxrs6iYBiG6eawUFjQI9sDIo5RMAzDACwUljgdhB7ZHlTWs+uJYRiGhSIGxTkedj0xDMMgw4SCiC4homdramrafa6SPC+OskXBMAyTWUKRrKwnQCm6O1zDjQEZhmEySiiSSe8CH47UNiMc5jYeDMN0b1goYnBCvg/BsODMJ4Zhuj0sFDHone8DAF6XgmGYbg8LRQxOKFCE4mB1U5pHwjAMk15YKGIwpCQHALDlcF2aR8IwDJNeMkookpkem+dzY3DPHKw/0P5zMQzDdGUySiiSmR4LAKP7FmDd/mpewIhhmG5NRglFsjlzeAkq6lrw5b7qdA+FYRgmbbBQ2HDhqN5wOQgLtxxJ91AYhmHSBguFDXk+Nwqy3KhuDKR7KAzDMGmDhSIOOV4X6luC6R4GwzBM2mChiEOu14UGFgqGYboxLBRxyGWLgmGYbk5GCUUy6ygkOV4nGlpCSTsfwzBMVyOjhCLZdRSAEqNg1xPDMN2ZjBKKVJDrdaGOhYJhmG4MC0Uccr0uHK1rwcw/f5buoTAMw6QFFoo45HhdAIBNh2rTPBKGYZj0wEIRB+7zxDBMd4eFIg6HdOtmN/k5+4lhmO4HC0Uczh3RS/u5oo5Xu2MYpvvBQhGHGaNL8eKNkwAYrQuGYZjuQkYJRSoK7gCgtFBZFvWaZ5ejsr4lqedmGIbp7GSUUKSi4A4AhvXKw9UT+wEAnlq8M6nnZhiG6ey40j2AroDTQfjdlWNRXtWE5bsr0z0chmGYDiWjLIpUM35AD2w+VIdGP1dqMwzTfWCLohWMH1iIUFjgq/IaVNb70bdHFsb1L0z3sBiGYVIKC0UrOKV/DwDAsp2V+PPH2wEAex6elc4hMQzDpBx2PbWCHjkeDC7J0USCYRimO8BC0Uq+MXkA3E4CAHhd/PExDJP58EzXSr5zxmBs+tUM3Hz6ILQEw2gOcFsPhmEyGxaKNuB2OjCiNB8AcJirtRmGyXBYKNpInwKlWvtgTVOaR8IwDJNaWCjaSK98LwDgaB239GAYJrPJKKFIVa8nK3rmKkJxrN6PN1aXY9GWipRfk2EYJh1klFCkqteTFQVZbridhKN1Lbj73+tw099WpvyaDMMw6SCjhKIjISL0zPVib2VDuofCMAyTUlgo2kHPXC/e33A43cNgGIZJKSwU7eAQZzwxDNMNYKFoB3edfyKG9coFAOT7uG0WwzCZSUJCQUR3EFE+KTxPRF8S0QWpHlxn5xunDsCHd52F26YPQaM/BCFEuofEMAyTdBK1KL4thKgFcAGAHgBuAPBwykbVxcj1uRAMC7QEw+keCsMwTNJJVChI/f9MAC8JITbqtnV78ryK22nyQx9h/rqDaR4NwzBMcklUKFYT0QdQhGIBEeUB4MdnlVw1PlHbHMRzn+1K82gYhmGSS6IR2JsBjAOwSwjRSERFAG5K2ai6GLlet/YzAdh2pA7De+elb0AMwzBJJFGLYiqArUKIaiK6HsB9AFLfJ6OLINenAIB15TW44I+fpnE0DMMwySVRoXgKQCMRjQVwN4CdAP6RslF1MfoUZqV7CAzDMCkjUaEICiX38zIATwghngTAvhWV4b3zsO6XF2Da0OJ0D4VhGCbpJCoUdUR0D5S02HeJyAHAHec9HU5Hdo81U5DlRrYnEvK56cUV2H6kDjsq6jt8LAzDMMkkUaH4OoAWKPUUhwH0A/D7lI2qjXRk91gr9PnCi7Yexfl//BTnPfoJKut5zQqGYbouCQmFKg4vAyggoosBNAshOEZhImyqzJZrVrzz1aF0DIdhGCYpJNrC42oAKwBcBeBqAF8Q0ZWpHFhXJBiOCMX4AYVYee+5yPY4sfsYtyJnGKbrkmgdxb0AJgkhKgCAiEoAfATg9VQNrCsS0gnFvbNGgogwsDgH+443pnFUDMMw7SPRGIVDioRKZSve220IhhSh+OfNp2LCwB4AgIFF2djDixsxDNOFSXSy/y8RLSCiG4noRgDvAngvdcPqmvzowuHo1yMLY/tHgukDi7Ox62gDfr9gC5oDoTSOjmEYpm0k5HoSQvyYiK4AME3d9KwQYl7qhtU1mTCwCEt+eo5hW2mBDwDw5KKdyPG6cNv0oekYGsMwTJtJeLUdIcQbAN5I4VgykpI8n/ZzMmoqhBAIhgXcTvb8MQzTMdjONkRUR0S1Fv/qiKi2owbZlSnJ82o/lx+PvXSqEAJzV+xDbXPA9nxvrzuIYfe+j32VHCBnGKZjsBUKIUSeECLf4l+eECK/owbZldELxW41qL1g42E8smArhBD496r9aA6EsGZ/Nea8uR4/f2uD7fk+3qzkFHyxuzJ1g2YYhtHBCz2nmJ65Hu3no3UtaA6E8N2XVgMAFm6pwKZDtdh8qA5nnVgCAKhqtLcoSgsVV9ahmuYUjZhhGMYIO7pTTK7XqMX6CX7TIcV7t+tYPfzqMqoep/3CgTlqPykWCoZhOgoWihRDROhbmIXhvXMBAAeqmtA732s4ptEfQiCkCEW8IHWTmmJbXsUxCoZhOgYWig5g6Zxz8Py3JgEAthyuRU1TAJMHFWn7V+w+jlV7qgAA7284jLI57+Inr6+zPFeTXxGKRj/XZDAM0zGwUHQQfQqzMKpPPn797mY0B8KYVNbDsP+FpbsNr19bVW55HikUq/dW4bxHP2n3uB58ZxNe/mJvu8/DMEzmwkLRQTgdhFdmT9E6ypYWxF8Vr6rBj4aWIFqCIazeexwA0Kir7t5RUQ9h6ljbWp5fshv3zrPPtGIYpnvDQtGB5PvcePr68QBgcD1dMb6f5fGnPPghZj72Gf73X2twxVPLcLzBr1kUkhY1CM4wDJMqOD22g5lYVoTdv50JIsKAomwU53pQVpwd8/i9lY3YqxbXNbQE0RQIGvY3B0LwuZ0pHTPDMN0bFoo0QKSkwH76k7MBAO+tT2xho6ZAKCqI3RQIoTCpo2MYhjHCrqdOwOnDeiZ0XJM/FOV6+u5Lq9HoD8Z4hz3tjW8wDNM9yCihIKJLiOjZmpqadA+lVeT73Jgx6oS4xzUFQlodheSr8po2L7XK8Q2GYRIho4RCCDFfCDG7oKAg/sGdjKdvmACvy/7XsWhLhRav0COL9VpLS4CFgmGY+GSUUHR14nmCnvl0l+V2ubJea2kJWhftVda3xNzHMEz3g4WiExFOMGZg7h/V1iptvesprFvve8KvP8L3/vllm87JMEzmwULRiUhEKAaX5OD2c42r5FU3+g2vX19djk+2HY17Lv3SrDL2IQVj4ZYKy/d0NoKhsNZQkWGY1MBC0YkIJ2BQ5Pnc8JgaBx5viAhFKCzwo3+vw7deWAEAWLnnONbtrwYAvPPVQbyxOtIaRG9RSKukWedyemTBVlTUde4utZc+sRTD73s/3cNgmIyGhaITI9fb1pPldsDjMhbYVeksis2HjAsPXvX0Mlz25FIAwA/+tQZ3/zvSbFBvUcif9em3TyzagQfmb2rHHaSeTYd4oUWGSTUsFJ2Qh742GgBQmO3B6vvOw8K7z4JbXafC53bC47K2KB79cBsufnyJtr0uzrKq1haF0Y1jrttgGKb7wULRiZg2tBgAMKqPkt7bI9uN4lwvBpfkam06stxOTTQkclW8xz7ebtj+6bZj2s9r9lVFXc8qRmEWBreTEAiFcf6jn2DawwsRSsQ/liBCCLy6cl9cQWMYJr1wC49OxHPfnIjDNc3wq3URPXIiy6hmuZ2oaw7C53ZG1Vscb/BbTuD6gPTX/vJ51H69RbHraD1GluYbxAMAQmHgWH0LtlfUA1AExZx11Va+3FeNn76xHst2VuJP15ySlHMyyaWhJQing7ifWDeHLYpORLbHhcElufCqMYge2W5tX5ZH2WbleqppCmD3sXrttc+t7F+28xjs0IvCXa+tw73z1kcJRWVDi6Ewz7y/PTS0KK1HKupa4h674UANNh3keERHM+qXC3DuH9q/7gnTtWGh6IRIi6Eo22hRAIoIONSmguMHFOK+WSMAAOv2K21LxvUvxLzbpqFPgQ8HY6yrLXs8mVt4LN52NKpFyNG6FkMmVLKEIhAKa+nA8n7suPjxJZj52GdJuTbTOg5UN6V7CEyaYddTJyTP54LH6UDfHpHFjQqyFOsiy+2EdDLl+tzona9kRu08qlgUj149FoNLcnHKgB44GKMrbUswDJ/biRbTpF+Q5dZiFG98byrmrzuEV1bsQ7POojhS2wwhgP5FsVujJ8Kwe9/XFnFKQCfiIoTQuvIyDJNcWCg6IXk+N96/8wwM0E3GRWq8IsvthFQK0m1fvVcJVvdQrZB+PWKvoNfkD2FHRT0+NhXV7aiox79W7AMAFGR50DPXg5ZgGPXNke60Vzy1DACw5+FZbb4/adEcq1dcTsmY4INhYQjyf77zGCaXFcHlZKO5qxIMhSEAuPl3mHb4N9BJGVKSa/gDyfMpmq4PKhJFhOGL3cpSqfmq5VGcG3FbmWkMhHDx40vw2fZjcJjm6MVblYruLI9TC1pXNtjHEOpbglq8oTkQwmVPLrXMsgKAPccaMOie9wzbkmEH6BsjLt9ViW889wWeXLQzCWdm0sWZv1uEUb9YkO5hMGCh6DLkehUBEOp/kl75XsNxTnXmL8oxbtejT4HN8bjw4GWjoo7xuRzI9SnXPBoj2LzhQA3mrzuI0b9cgFMe/BAAcKimGev2V2sWjpn1B6JbwFsZFC8t24MH39lk6EFlRyAYOe5QjeJT31PZYDimqsGPI7Wdu9KciXBQlwHIpBcWii6CtCjqm4OYWFaEsuJs3HX+cPTM9eJHFwyPOl5vUbhMZsNTiyNP2tleJ26YWobzRvQ2HKO3KGJ1rb348SX431fWAIDWb0m6qY43+PGHD7Zi/IMf4t5567UJ31wDAkQHs5sDIfz8Pxvx/JLdOFLXjGACk4V+QpHddJ2m+57w6w9x6m8+jnuuZNASDOHRD7clNUuMST9VDX6s3ns8rWMQQiT8AJUsWCi6CFIoapuDyPe5sfjHZ2NMv0IAwMVj+kQd31O1KHI8Tuz4zUyMLM3X9r3xZaTfU7ZHOa+5rbjP5dSuaWVRVDX4o7YBihsKUIoAH1+4A8cb/Hj5i32oaVKK6lyO6K+cWTr0Fk9tUxCNusk21qp8wXBEKGRNidMkQB35t/Xy8n147OPtePoTdn9lEtf99Qtc8dSytK4OecfctRj8s/fiH5hEWCi6CNOGKsulTj+xJGpfaWF0T6gi1aKQT9py0jcj51Lzk6/DQbaFdatiuJZkrMIsJDIV12VhUZjRp+PWNAXQ2BJ5HYix9obe9RRUFeFgTRMaWoJ4ZUVEqDoK+blzC5TMQvYWS+fqkG+vO9jh12Sh6CKMKM3Hrt/MxPQTe0XtsyrQK1azofLVOMPkQUXWJ1bn12aL1e5yY4gLgJjFbw1+aVH4tdoPAPjOP1Ziw4Eay0WW5B/doZomhMPCUOBX0xTQzgkgqs5Donc9SYvis+3HMOqXC3DPm+vxs3nrY95LKpBer1gtTz7bfhRlc97V0ppbSzAUxu5jDfEP7OR8sasStV2ohYv8vXa31SFZKLoQDnOKko6P7joTH/zwLO21z+3EA5eOwqvfnQoAuPO84fj9lWOi3iensXtnjcDgkhws+tF0vH6r8p48G4ti06HooPS8NeU615MfA4sj6b0bDtTi4seX4Lv/XB31viU7juH2V9Zg6m8X4ppnlxssilqTRRHrCV2f9RS0mJyPJVD9nUxk3CWWu2velwcAAGv2Vbfp/LfPXYOzH1ncpS2W2uYAvv7scnz/5dQtkiWEwO8XbMHWw3VJOZ9Mt27uBCtAJhK7SxYsFBnC0F55KMkzZjp967QyDO2VC0AJ7F48pg/6FhrrK6SvdcrgYiy8ezoG9czBxDLF+rC1KA7V4uS+BYaMpR++ug51zZEYBYCotTNiPWFLc3rFnuOorI+4rWqaApr4ALEtCr1QWK0hnkj1dzw+33EMG0xZW1+VV6OyPlqESBOKGK4ym+B+Iry3/jAAGKytrkazKnKbDyVnEreiriWIJxftxAcbDyflfG71Ya0zJCl0pPuLhaIbkeVxYslPz8YPz4vOkrI83qYR3P7jTTjphDytulqy73gjACVG0RwIaXUdrUEfPK9tDqBRNxn+8cNt2s8vLNmt/RwIhbF673EIISyXhm2LToTCAre9vBqvrlSKEL/x1y8MbdwBZeGky55cikVbKwxP99L40wc975y7Rpuw5NOgVXA/HvpJqj2r+209XIfr/ro8bVaJtPxSWU8nY2bJSrPVLIpO4HrqSLFioehmEBFuO3sIbppWBgCwy92IVzE9pFcu+pgWV9qpdpkNhgUq6/0oyGp98b9eKJQYReQP4u11B9EcCOHTbUfxq3ciiyrNX3cIVzy1DK+t2m/5B6S3KBLNWHlrzQG8t/4wHl+4w3K/TFEsr2rCTS+uxL1vReIgVq6nt9YexOyXFNebtHoSCe6bqdUF5tvzVPnLtzdg6Y5Ky5qXQChs+BxTkY6ppTHH+Z6159pSKFrzOTW0BDHnja9Q0xgdO5EWIFsUTMbjdjpw/ZSBAIB4c+ZT142PuW9wzxyUFhhdWbt0Ada6lqDWo6o1HNO5cmqbgjhucu3sP96IVXuMuewbD9ao+5oMFohEPxfFcn/964t9eEVtYQJE+med3LfA8nizn3qLzoUi40nS9WSe7GT2VltcT3rhNKc1twb5u7cKfV32xFKc9PP/aq9TUfjmDyljt4u9AUAg3PZr16vxLXNfMzteWr4Xc1fux/RHFmGPKWFAWoAdLRThsMDJ9y/AXN33k4WCSTmyCE/Y2hTARSeXxtw3pFcuTjBZFOaai7YIhWw77nE5UNPkj+pe+tv3t+Avi431CSv3KE/FWR4nmvzRf0B66yjWpPezeetxz5sRq0A+jVrFPJT9xslCP2lLq0Xqg/masu5DCOD5Jbvx0LubEp58GnQxm2Rk31h9A8xLzBqSBUJhLN9V2e7ryonOXBi59XCdQVhjpUQngp1FUdMYwCMLtkYFhaUrrqoxgCufNq7jIsdqXgky1bQEw6hrDuIX/9mobWPXE5NySguycGLvPPz68pPbfI5+PbLQx6KGQ/+U3BqhMBf49SnwoaoxECUUC7dUWGY2AcDrq8s160KPfirad7wRNY0BHG/wIxgKY9HWCvzff7dEvadON8lYWSFmy0U/GcnJTQqGfqI943cLsWynMtEGwwIPvrMJz322G1/G6I8VfV29RdG2CauuOfK5JjLh6GMhH20+gmueXY59lY1turb5nHrX01fl1bjwT5/ir0t2RR3XFupthOLX727CE4t24KPNxuaY+mOPm+qBEnE9Ha1raZelZ4X8DPTJEfq0cn3CRyrg7rHdFI/LgQU/PDOhYx+/9hRUNwVw4cje2Hu8EVc9rXSQ9bqcWiGg5JKxfXDeiF64Y+5aAPGF4rQhxeid78O8NQfw7WmD8OePt2tCUVqQhcO1SguPM4b1xLkn9cL98zfZni9WbYF+op/xp89QnONBZYMf10zqDwB4bdX+qPdogdBg2NKdZQ6aG4UibLiu/ql4//GI8OnrShINKuszndo6iV719DKUVzWp5wup42qEPxTGkJLcqOP14z+qZqXVtbSv/kF+XkTKmheHa5pRXqWIz7ryiNjHsugSQf7erD6nqkblPsyeLzsRiASzYx9z0Z8/w3fOGIRbzxrS2uHGpEV104V0QiHHcPUzy3DxmD746YyTknY9M2xRMHG5ZGwf3DBlIHrl+7RCPsmoPgWYO3sKAOCbUwfi8WtPwVnDI9Xj8YRiYHE2fn/lGLx40yR8b7ryhyVjFKWFPhyuacb2inr065GFG6cNwkd3JSZuZswTfaX6pDh3pRL8NhsMy3ZWYsHGIwCUicpqEo+yKHSTh3RnhIXy8+8sLBbA2HpEn/q7o6Le4GIyXFfn8vpsx1FDXMWK5kAoSkC36OoKGtXrnPG7RTFXs9NPtDKYHgwJ1DYH2vz0LM/pcjhw5u8W4YqnPtfEw62bvdtnUagxCosxyswl8zKvesE3J3S44whFKCxwrL4FB5O82JP8DPQxRTnOIzUtlinayYSFgmkVVmsnnzqoCH++Zhx+NlNZbU9WgwOImx7rczvhcjpw9om94HU54HaSNon3KchCUyCERn8IZwxTxGdorzx8c+rAVo/broWHuTajoq4Z1z63XHvtD4UtU26jYxSRCcYfkjEKgXfXH8LcldEWC2CcBKUYCSFw3qOf4Ka/rbR8j96ieOaTXYa4ihX3vLkeZz+yOKZ7osF0bzP+9Gn0OHVP9bKSOhgOY8z9H+Abz31he/331h/CexaLaMl7dzhIs7xk6xd9i327QPpv39+Mk++P3YrcLkYhJ3vzehd2ge+I68l6TPKc+jVckoHeopPa1RIIoSUYgj8UbpeYJgILBdMqvK7orwwR4bJxfTUR0Wex5MSo7p6lBsmzPfr1NUgTGaeDtBbqfQuzMFMXVLer74iFnVCY/+gnP2TsMOsPhg2Tc3lVI+5/e6NWXCjRT0bSovAHw1GCokc/ectJRsZfVuy27lKaiIvqrTUHcOVTnxvOE6uRY5PJMtpiUcX85KId2HZE2V7bJN05yjj16bX7Khuj0m1ve/lL3GZRfS0FQD9Py0QGvThcrbo6rXjmk11Rvwc9mlBYTOzy9xU0ZVXZxX2ccQru5O+zLskxA70QuHQBdfndSnU7dhYKplVIMRjWK9qPrUdO5lbCAgD9irIMx0m0BZpcDm1RJnNmlZVVo+eGKdEWh51QVDdaT6CSQEgYJufT/28R/vb5HizZcczmPcofbkswFDUR6dFPctKy0U9UVjUOVtXY5ifKO19di1V7q9AcCGmfabVFXYByvlDcJ9J5aw7gsieWAjBaFJFxHsfsf6zCmb9XXEiJYBXMlvGpKt1YK9Wkg7kr9qFszruWk/QHGw9jwcbDEEIYkh8iwWwr15MqzKasKv2x5sRdt8O64O6tNQewas9xrY19LLdhW9H/fmSNTksgZBDCTQdr8eaX5SmxLlgomFaR43Xhz9eMw8u3nGp7XM88ZZLXT+r69iGytYd50s9TLQqv26lleJSYqr/jCcUZw3pGbbNLsZTV5LHwB8NR6aKAsfBNsulgLZbtrNSu1xIM215bb1FI95b+D91q0m20sFCG3/c+6iya61U3BrQuwMd1gqhvrdLYEjTUrsRCCpm8b/1keN9bG/HBpiNxzyFZt78au9Q6Fb0FerimWR23UbwDIYFH1ap8K8Gb/dJqfPel1Xhr7QFMe3ghHpi/Eaf99mOtaE4vvuVVjSib8y62q8WhUtSlANpZFDKdvDkYghBKjKY5EMKdr67FlTrLJ9lZSLLmBDBaFPW6yvOPNh/BXa+tS8oa9GY6vVAQ0WAiep6IXk/3WBiFy8b1Ra+86LRYPbK1h96imPf907SfZeFSlie2RTF1SDFK8rz4wTlDDcdkuY1f20euGmtYwzuekJipivGkLTlQ3WTIX5dYdT2d+dhnuPa55TqLIoxQqy2K2K6l219Zg+c+s15I6v0NhxEOC0OQvbrJr32m6/ZX46vyatz44gpD7cLfl+3FW2sPxLymmVp1zHorbajJwpQW2MIt1uJx2ZNL8Zha8a63KOSqhAeqjMFgfzCsjdkuC+orNVvqxaV7cLCmWWuHv+VwHd5edxBlc97F7H8YG1MGwwKLtlZgzP0fYMXu47YZTdIt2BwI4Wfz1mPM/R9Yxq8SiVHsPtaQcJcAvXjJz6ElEIpYTIEwqhr9yPO6UrLGeErTY4noBQAXA6gQQozWbZ8B4M8AnAD+KoR4ONY5hBC7ANzMQtG1kEIhv+B5Xhe8zsgE7nYpX3az60nGKLK9LvTK82HlvedFndssLuaJw7w/VVhZFBLpzjhS24yXlu817JtcVoQVamW53gpotrAoAGVSkuInmycW5Xiicvx/8vpXOFrXgn/qrlfVENCstEc/3KY9lZv5x+d7Lbeb+eGra1Gn3reMVQDRrpbjjX70cfvw7b+tintOvUUhYxSVpnvzh8IRl4uNkOZ4jFOa3lq8XV2N0aqY8KtdisCs3HPclPVkPL8Muje2hPCqmlI9Xl0GWE+8GMXSHcdw3V+/wKNXj8Wn247C43LgjvOGo0e2W1tMTI/fSih0FkVLKIyaxgAKsltf4JoIqbYo/gZghn4DETkBPAngIgAjAVxLRCOJ6GQiesf0L3rxBaZL0K+H4mZq9Iew4mfnYslPz4FXZwlI10d2DIuiyJSGq8dsMZiFwudqvVCcbbEgFGDd3kJy3Ca2Ice0t7LRUDcxeVARXrxpkvY6XowCgOZz19eCFGS5LV0MH2w8jEM1kXXBD9c2xexgqyfWwlZm5q05oFlSD723Wdu+cIuxaK2qwR+VTRardYq519Nt06PrDwKhMGT/RKvKe0m2t/W/+2BIaNlMgVDYttpduhG3V9h3vI1nUci25+v2V+OttQfx2qpyTHt4Ia6NkUEWL0bhD4ZR3RRAYVcUCiHEpwDMqRuTAewQQuwSQvgBzAVwmRBivRDiYtO/iqiTxoCIZhPRKiJadfTo0STeBdMW7r7gRPzg7KG4ZGwpeuX7UJDtNvjF3XFiFOZ6DT3m95ifwH0m15T5tRUDi3PwwKWjDNuumdQf37UpmjK7R/QrAgZiTIr3zRqBHK8Lz9wwAYDRomgKRDKl9OxXn4r1x/rcTstEAX2hGqC0fn/nq+jUVAD49rRBuF116yUSo5DoLYlYVDcGorKRYrVEX2ZqB3LKgB5Rv39/MGJRxGo1D0RbFIkQCIW172MwJGwtFulGjNcavSkQsl0vQmqj+Vuybn+15fH6rCa9W1MKUkswhOpGPwqzYv/dtId0xCj6AtAnlZer2ywhomIiehrAKUR0T6zjhBDPCiEmCiEmlpRYPx0yHUeu14UfXXiitvoeYHQxyK6pUa4ntdus3ZOR+T3m6nCzkAwoyoaZz35yNt76/rRIdpbbAY9p4i3M9miT8YWjeuPBy0cb9pu1QP9UHisXX05IF4zsDcAY9Jy/7iDmrtgXNVFJC0EfE/C6HLZdV63u2UyWx4G7LjgRfQuz4sZp9CSSilnV6I8SChmEj2VZSPoU+qK66gZCYe1+pVBY+ffjnduKQFhE4h/hsKE9DJnynuQ+O7GSNLSE8MTC7fjcIjtOnjXRpbf1Dw9BXbV/vcmi6Kqup3YjhKgUQtwqhBgihPhtusfDJAc5YZrjCfKp3GvjPpLvGde/EHsenoURpfmG/V63A49cNRYDirIxrn8hLhtnfA7xuhzor+6Ty8dmuZ1Riyx5XRHxcDsd8FoECfXWil4oYuX2y/smIrgchG1HjEuhznlzPZbvMhrhkUwgo1DE6ncFAK+pKxtKrNY/lyKZLHfFoJ45WHHvuQCkUBjFp8GmnYaeklxv1DodLcGw9qDR5A9h//FG7LXoNdWWbKOgrqV6SyBscGX6Q2HDoketEaK6lgAe+WAbvvHX2AWJibgFAXNTxkj/sIaWiLuyujGAwjY04UyEdAjFAQD9da/7qduYboR86jcX5Mk/RLv225HOt9ZkuZ24ckI/fKpaDb1MK//pxUm6unxupyGGAqhWhjqxe5zRFgcA9NGl/OonY7M7RaIXo1gT/e8XbDW8PlKrCoXeonA7bSctr8uB04YUW45N4kuyUHicSu0LEVBZ74+atBtagviqvDpuE8Icr8vSopAG6b7jDTjjd4sw/ZHFUe+1q5eJRTAktAm3tikQFSOSa4jIYxPFTrTkNRKtedAfJy26QCiMerXfVnNAcT3J2qNkkw6hWAlgGBENIiIPgGsAvJ2GcTBp5PwRvfGHq8ZicM8cw/bIgj6xv5qy/06s1EKz6ynqtc5akdXf1haF02BRWHl69LUhub74E67M9moNh2utXU8Sq7oRr9uBf90yBTeeVqaOzcKi8EihSGxyuWxcH9v9HpcDbqcDPXO9OFzTHBXQfWN1OS59Yik2WHT31ZPtcWrxiMmDlGV59TEKcxaZnld1rVK+dkpMj7aBQDjS9LG6KWAbzA6Gw1GrOsbCLqAtg9BWKdZWWNV2zF25H899pqzyWNccRFgkT/TNpFQoiOgVAMsAnEhE5UR0sxAiCOAHABYA2AzgNSFEdJJ62653CRE9W1Nj/0Vk0k+Wx4krJvSLarrWX/WtW3UwlcRZ5yYqj9wc0/j+2ZEA9Qn5Sj2Iw0FRFoPereR2keXTpMzuApQU4HiYxUgSK3if7XFqrqcaXZaV1+XQmi9eMOqEmNeRgmKV1SQ/lx4JTi6JTpClBT4crGmKcr/Jdt6xOvxKiEgTZSny1z//hdZaRJ9FZkY+xT99/Xgtcy5eb7BgSGj9rmqa7JschsICI0rzbM9nHosV8npHaqOTCKxW9Eu0RYe5i0GySHXW07VCiFIhhFsI0U8I8by6/T0hxHA17vBQEq83Xwgxu6DAekUypvNz6dg+eOWWKbhifOynwRGl+bh0bB88ctXYhM6ptyj2PDwLN0wt017LP6yjdS1RcRGP06GZ/G6nw7IVRx/dCn85MVIzR+piKG4L99WLN07CBz88E/mmyXxISQ5mnlyquZ6qDDEKJ3535RhcPq4PLlQD43pcJqGQDRf1SKHI0/XXsiM/jsUk/e2lBUrXX1lL8Mb3lHiJTPOV2WIXjooet0RaDz51/LGq22eebC2SM0aXavfTpzAL3zl9UMxrVTX6sVLth3WouikqSQGIuESDYYFBJis4Fnqh9AfDBgtYWjAVtc3R77MQGK2BYpyHJHPhY7Lo9MFspntBRJg6pNh2vW6304HHrj0Fw3sn9mSX5Yn9NZfLnBZlu+ExuYWcDtItWeqwnKym6uIAVjGMJ78xHi/cGKmbsLIosjxOFOd60beHMVPpX7dMQR81IykYCmstLwDFtVSc68WfrjkFvfJjP0XKMRFIE0KXg3DGsJ4Y279Qub4qGPHWro6XZiznwdKCLByqadaC2ebfkxSMqyf217K/zDhMFkUsyoqjJ20p6FJsQmERFX/S8+LSPZp772BN9MQNRNauCIVEzMrn4b2Nk7TeVfj8kt0YdM97GPmL/+LWl1bjy73VMa9X3ejHp9uOYtA97+JQjfJZycr0kX3yo47XY/V5JAMWCqZD+eiuM/HJj6en5NyPXj0W1506IGq7XQbVBaNOwMvfORU3TC2DlXUvYyZuJxniEZITT7AXq1ljSg3uAKtJRhYdmmMuSnBYeYqvaQpgqy5DKlazRTPy3gWEJhrfP3soXrr5VC0QLwUgFCcDx6riXR8wl+8uLfChviWIwzXNyPY4NYtFItdq8LqcMbN+ZFpqvE7BvS1EUloE8qMOhwU8TvvzxKNSXawpGBZwOQj/uuXUqO/aU9dPMBSQ6oXizS/LASgFqP/deBhbj8Suwzha14Lb566BEMB29XfuD4bhcTpsrbp8n6vV7WsShYWC6VCG9srDwBQ99fzP+H546GvRS7vGa+kxbWhPOB0U5ZsWQi8UDpx9Ui889LVILcXrt041/GGac+6tXBRW7h05GZqzmLxuh+ZnP1bvx84KvVAkNiFIcQiHI7Ud5sWkYl1f8vdvT8aWB2dYVrz311lBUuhk5s2B6iYtNjKkRPksCrPdmkXhc8dO8Y3EKOynqN75seMm545QrJUzhpdoFoUrnu8mBpUNSiwhGFae7E8b0hOzxhjXk/c4Hfj47rPw+q2Kq03f2HB7hTEN2o5DNc1aKrS0ivyhMDwuh+1CYP0TqJ1pKywUTMaT6FPW6D6KG0qm0wpEXEuyqG/akEiG0cSyIsPEY/bcvPO/p7dqfOYneo/ToWUkbTlcC38orGW1JG5RqEIhIgHb4lxj4Nwb5/PxOB3wuZ3wWQiuPotLDl8K85HaZi0t95/fORUv3jQJ/XtkawFcryt2iq90Pcb63Uk3T1FObKEYP6AH9jw8C+P6F2qfg91Dg11W1yfbjuKKpz5HICS037nZOvS4HCgtyMLEsiLkel1RRYyJ/s70wX7ZNdYfVITCzqJIpMiyrWSUUHDWE2NFogsd9cjxYM/DszRREELgtCE9sfXXMzCpTEnTNPu69bEUvU4MKMqOuWhT1Pg81jECl871JNe3lhOFVTzECjle/XRs7qMV7/ORVpAMLBuywXSTpXQjSfdLRV2LljJcWpCFs0/sZbi21+2IKRR2MYrbzx0W6T7sduL+S0bi3dtPx990PbTMSAss1r1OHVyMP319XMz3P/PJLm1tEKd6bbN1qI8/5XpdUe3Qzas9zjq5FLt/OxOr7jM2vlyra+Mhg9j1LUFke5xa5wIr2KJIEM56YqxIpNeTnhmjlUwaGezVu3kScfl4XA5DABsAnrlhAr4VI01TTl5/uW48yoqNf+zSjSMnDzkJm62XKYOLLBspSt+8fkI2HxfP4pLznxS0Il3dhd6ikleQ3U+rGwNRmVx6a8brii0U5vRYyfgBhbjr/OE4Y3hP7Xw3ThuEUX0KbIVZCmusY4pyPbYJFHpkMaDbEW1RSHJ9LtQ0GZtGmhMZsj1OEJEh7bggy20QClk/sfVwHYb1yo2K9+gZp35fU0FGCQXDWNHabrIXjjoBOx66yDKrKhH3wT0XnRSVpnjhqBPwwGWjLY/XVg3snYe/fmuiYZ90NX2oLgokJ2tzPGTu7Kl43vRe/Xj1gfJooYjck5ULX/rJ5URXpJvs9fO82aIAoivC9ZOizx07mK2lx5pEXtYT/PiCE7H4R9MNlfGJWI6xjrHz/Zv1Q4qj2aJwmywK6XqSx5utUSvRGlKSY2gf7w+G0eQPYefReowszY8SXgA4fWhP/P3bk3HR6OhU4WTBQsFkPI42BDBjVYbbCUWkYrx119JPOEN75eHWs4Zoa4qbJ1pzW3bj2KL3tdb1ZFVUJ8cnC8j0cQHDRC+ix2gu9NMXFlpZFP93hZKMECtGIV0xLqcDZaZkAbv4g5WIAZF7s+uRpC+q1L/HXJeif90z14NDatBeXtNsUVjV3UwZXGx4/ePXv8Kpv/kIYQGM7JOPC0efAJeDDIkS2R4nzhpekrBF1BZYKBimFbicDpx0Qh7++PXoYr+JZT0AIG6uezzmXHQSnrxuPABlwtS36LBa1EZi5WKTk5N+QjYLin4y/vetU5Fjmkzl0/2kQUUYUZqPey46SdtnoROGCTvXa5yAi3VCZA5mf2/6EHx90gD1mso2c8DYrjeSnUUhLSqzmMjr2FkU5rTolqB1mxn9RD2wOEdLHtAaXbqdWDrnHN1aLNG/S31djkSuKFjWMwelBVnY8ZuZhu/Yry+3tlSTCQsF0y3I9jjxDYsai7bw3zvPxNdO6Re1/eIxfbD8nnOjngrby4M6l1WsGAVgnb0kt4VFbB+2XigGFufgf88dZtgvn6DzfW68f8cZUd16JfKJWr8mRJRFoXNb+dwOQ6aXXnTk7ZkNu9lnxl4fxE4oZI2M2aKQRZTSyrLu52WMG8m1O+xSbQfqYk3ZqlC4HUotzvATFLekXpClhTCprAjnj+yNf30nek16g2Cpn9V9s0bYFlwmi5QuhcownYVNv5oR/6AkkIpeO/psGU0oLI7zWbjF5NOrEAKvfXeqZRsSsyVy8+mDkOdz4b63NkAI+9Yed5w7DHXNQRRkubWeSlkJup6IyFDkqHdjSSvGoZu59euiW2HnepJCMLgkF0D0Ot5yonbr2rZI+ppcT8fU4jtzh1s9+lRVGYuQdyfvWR+jmHfbaThc2wyf24nnvjnRcl1wfSBbflaJpty2l4wSCiK6BMAlQ4cOTfdQmG7ETdPKtEV5UoF+spVP/1ZPvlbZS3KSDwulMttj4UQwP4m7nQ5cd+pAPDB/k6FrqxU9cjz4w9VGN5x+8jILhZywT1W7wuob4Ol/lpeM138q1nXNXDiqN/5y3XicO6IXnlq8M2q/FAqvhVD0M7me+qvCYV4zw/AeXSFirtdYeS/vUy8UhdkeQxffeIWBUlMTTZNuLxklFEKI+QDmT5w48ZZ0j4XpPvzyklHxD7Jg6ZxzYq6Ep0fvpz9reAleXLoHEwYWRR1nNVHK+cZuvZ1Y6bGyrqM1kzVg9NWbYxT9emTjD1eNxXlq1bTewtGPUZ7DSYRcryuhjB67YC4RYebJpTH3SwFzuxyAqaGrXDXu8nF98M3TyrTCTLvJXG/dyFiEvL1YgXXzeO0Q6tlYKBgmw7HqHRWP6Sf2wuZfzbB0s1hlapHmwol9zphC4YhM1mb+/u3JCCSw6M7A4ugisCsmROI7Vum1QGS8RIQND1wY9zp6WtNBdUhJDnYebdA+p8llRfivbkU7QJmMN/9qBjwuh0E09a6nrb82ujb1oi2D2fL25H0mWpD5/bOHaO1IJJpF0c4eVonCQsEwXQw7X3xxjgc3nxFpqT24Zw6+PW0QrpsSO5Afy2KQm608LHItjHicFKdp4pTBRVrLihLdSoSONlozS+ecY1lrYMWiH01H38Isg0D98evjcFtFHS59Yqm2zeUga2HWfTBRLep1QiHTYCMWhfJ/O4tCz+Xj+mKYqaZHnsMuTpJMWCgYJoNY/fPzDa8dDsIvLhmZ0HvNNRSOGIVlrcFupUIAeODS0fjOGYPxVXk1Lh4T6bUUEYrWXa81VppV08YsjxNj+hUatsVqv243SetrJjTLQcYoZKpugq1lrB8MlHN0jEywUDBMl+COc4dpy2emgrd/MA2lBabCMorteorH/B+cnlDrFI/LgSEluVErGspL2gXSO4pYBZt2MQqDUJhiFK21lqwERRpBqSyy08NCwTBdgB+ePzyl5zc/RQORCbKVheYAgJP7ta/fWkdNgIkQS6zsxqgXlxxTjOLZGybg5S/2GVq022FVmCd/Jx31KXHBHcMwlkhLIlbjvlQiJ8DWtkNJFvqW4+3wvAGI1KnITKVhvfNw/6WjEm4tY2WZSfdVR+lpRgkFtxlnmORx9km9ABgrrTuKSFpv8pXiL9eNx9PXT7A95s/XnILxAwqVsbRTKdxa0WPb3m9luURcT20dVevIKKHgNuMMkzx+ddkoLPnp2VodQUci3T2pMGZmnlyqtZK3Q1ZCW61z3hpka5Nkat556lrjg3omngrcHjhGwTCMJW6nw1Bh3JFEOvGmyfcE4JGrxmLemnKMameTxwFFSnbV1yf1T8awAADXnzoAl47tY9vMMJmwUDAM0+kgzfWUvjGU5HltmxAmSq88L3b9ZmZS3URE1GEiAbBQMAzTCYmEBdKoFEnC6aA2xTmW33MuGv2pS4luDSwUDMN0OuQKfum0KJJFvAZ/sUhFJ+K2wkLBMEynQ3bHSEXWUzL5751nxHUBtaeyvbOQUVlPDMNkBrNOVuoYRsZYJKmzcNIJ+VEV7Wbs2pF3FdiiYBim0zFrTCkuGj2z3TUMnQFnBzXuSyVdX+p0cMEdw2QOmSASQNtjFJ2JjBIKLrhjGKazwTEKhmEYxha2KBiGYRhbOlMn3LbCQsEwDMPYwllPDMMwKeDd20/Hit3H0z2MpMBCwTAMkwJG9SnAqD6ZkVjDrieGYRjGFhYKhmEYxhYWCoZhGMYWFgqGYRjGFhYKhmEYxpaMEgru9cQwDJN8MkoouNcTwzBM8skooWAYhmGSD4lOvoJUWyCiowD2tvHtPQEcS+JwugJ8z90DvufuQVvveaAQosRqR0YKRXsgolVCiInpHkdHwvfcPeB77h6k4p7Z9cQwDMPYwkLBMAzD2MJCEc2z6R5AGuB77h7wPXcPkn7PHKNgGIZhbGGLgmEYhrGFhYJhGIaxhYVChYhmENFWItpBRHPSPZ5kQkQvEFEFEW3QbSsiog+JaLv6/x7qdiKix9TP4SsiGp++kbcNIupPRIuIaBMRbSSiO9TtmXzPPiJaQUTr1Ht+QN0+iIi+UO/tVSLyqNu96usd6v6ytN5AOyAiJxGtIaJ31NcZfc9EtIeI1hPRWiJapW5L6XebhQLKFw3AkwAuAjASwLVENDK9o0oqfwMww7RtDoCPhRDDAHysvgaUz2CY+m82gKc6aIzJJAjgbiHESABTAHxf/X1m8j23ADhHCDEWwDgAM4hoCoD/A/BHIcRQAFUAblaPvxlAlbr9j+pxXZU7AGzWve4O93y2EGKcrl4itd9tIUS3/wdgKoAFutf3ALgn3eNK8j2WAdige70VQKn6cymArerPzwC41uq4rvoPwH8AnN9d7hlANoAvAZwKpULXpW7XvucAFgCYqv7sUo+jdI+9DffaT50YzwHwDgDqBve8B0BP07aUfrfZolDoC2C/7nW5ui2T6S2EOKT+fBhAb/XnjPosVPfCKQC+QIbfs+qCWQugAsCHAHYCqBZCBNVD9Pel3bO6vwZAcYcOODn8CcBPAITV18XI/HsWAD4gotVENFvdltLvtqutI2UyByGEIKKMy5MmolwAbwC4UwhRS0Tavky8ZyFECMA4IioEMA/ASekdUWohoosBVAghVhPR9DQPpyM5XQhxgIh6AfiQiLbod6biu80WhcIBAP11r/up2zKZI0RUCgDq/yvU7RnxWRCRG4pIvCyEeFPdnNH3LBFCVANYBMXtUkhE8oFQf1/aPav7CwBUduxI2800AJcS0R4Ac6G4n/6MzL5nCCEOqP+vgPJAMBkp/m6zUCisBDBMzZbwALgGwNtpHlOqeRvAt9SfvwXFjy+3f1PNlpgCoEZn0nYJSDEdngewWQjxqG5XJt9ziWpJgIiyoMRkNkMRjCvVw8z3LD+LKwEsFKoTu6sghLhHCNFPCFEG5W92oRDiOmTwPRNRDhHlyZ8BXABgA1L93U53YKaz/AMwE8A2KH7de9M9niTf2ysADgEIQPFR3gzFN/sxgO0APgJQpB5LUDLAdgJYD2Biusffhvs9HYof9ysAa9V/MzP8nscAWKPe8wYAv1C3DwawAsAOAP8G4FW3+9TXO9T9g9N9D+28/+kA3sn0e1bvbZ36b6Ocq1L93eYWHgzDMIwt7HpiGIZhbGGhYBiGYWxhoWAYhmFsYaFgGIZhbGGhYBiGYWxhoWCYTgQRTZddUBmms8BCwTAMw9jCQsEwbYCIrlfXf1hLRM+oDfnqieiP6noQHxNRiXrsOCJarq4HME+3VsBQIvpIXUPiSyIaop4+l4heJ6ItRPQy6ZtUMUwaYKFgmFZCRCMAfB3ANCHEOAAhANcByAGwSggxCsAnAH6pvuUfAH4qhBgDpTpWbn8ZwJNCWUPiNCjV84DS7fZOKGujDIbS04hh0gZ3j2WY1nMugAkAVqoP+1lQmrCFAbyqHvNPAG8SUQGAQiHEJ+r2vwP4t9qvp68QYh4ACCGaAUA93wohRLn6ei2UtUSWpPyuGCYGLBQM03oIwN+FEPcYNhL93HRcW/vjtOh+DoH/Tpk0w64nhmk9HwO4Ul0PQK5XPBDK35PsWvoNAEuEEDUAqojoDHX7DQA+EULUASgnosvVc3iJKLsjb4JhEoWfVBimlQghNhHRfVBWGXNA6cr7fQANACar+yqgxDEApe3z06oQ7AJwk7r9BgDPENGv1HNc1YG3wTAJw91jGSZJEFG9ECI33eNgmGTDrieGYRjGFrYoGIZhGFvYomAYhmFsYaFgGIZhbGGhYBiGYWxhoWAYhmFsYaFgGIZhbPl/Nb1Hn823uGoAAAAASUVORK5CYII=\n",
      "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",
    "plt.semilogy(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(history.history['val_loss'])\n",
    "\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",
    "* x-axis walks through each pixel, y-axis shows the result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Check each image for expected and deviation\n",
    "* setting the switch \"only_deviation = true\" will only print the images for which the classification and the CNN-result deviates\n",
    "\n",
    "The output contains the following information:\n",
    "\n",
    "| Filename      | Expected Category           | Predicted Category        |\n",
    "|------------- |:-----------------------------:|--------------|\n",
    "| ziffer_sortiert_resize_NaN/5\\Ziffer_4_0034.jpg | 4  | -1 |\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 1s 681ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 108ms/step\n",
      "1/1 [==============================] - 0s 314ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 124ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 64ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 88ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 75ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 119ms/step\n",
      "1/1 [==============================] - 0s 121ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 196ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 227ms/step\n",
      "1/1 [==============================] - 0s 195ms/step\n",
      "1/1 [==============================] - 0s 108ms/step\n",
      "1/1 [==============================] - 0s 189ms/step\n",
      "1/1 [==============================] - 0s 189ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 346ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 314ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 123ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 118ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 139ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 54ms/step\n",
      "1/1 [==============================] - 0s 142ms/step\n",
      "1/1 [==============================] - 0s 295ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 251ms/step\n",
      "1/1 [==============================] - 0s 205ms/step\n",
      "1/1 [==============================] - 0s 214ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 220ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 158ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 136ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 142ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 251ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 146ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 172ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 99ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 97ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 89ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 142ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 114ms/step\n",
      "1/1 [==============================] - 0s 163ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 98ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 88ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 172ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 138ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 81ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 189ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 75ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 100ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 172ms/step\n",
      "1/1 [==============================] - 0s 190ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 138ms/step\n",
      "1/1 [==============================] - 0s 144ms/step\n",
      "1/1 [==============================] - 0s 140ms/step\n",
      "1/1 [==============================] - 0s 107ms/step\n",
      "1/1 [==============================] - 0s 123ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 98ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 64ms/step\n",
      "1/1 [==============================] - 0s 71ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 176ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 114ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 117ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 143ms/step\n",
      "1/1 [==============================] - 0s 174ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 138ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 168ms/step\n",
      "1/1 [==============================] - 0s 87ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 108ms/step\n",
      "1/1 [==============================] - 0s 111ms/step\n",
      "1/1 [==============================] - 0s 95ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 132ms/step\n",
      "1/1 [==============================] - 0s 111ms/step\n",
      "1/1 [==============================] - 0s 128ms/step\n",
      "1/1 [==============================] - 0s 73ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 114ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 159ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 113ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 142ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 150ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 95ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 158ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 138ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 152ms/step\n",
      "1/1 [==============================] - 0s 151ms/step\n",
      "1/1 [==============================] - 0s 150ms/step\n",
      "1/1 [==============================] - 0s 122ms/step\n",
      "1/1 [==============================] - 0s 123ms/step\n",
      "1/1 [==============================] - 0s 104ms/step\n",
      "1/1 [==============================] - 0s 156ms/step\n",
      "1/1 [==============================] - 0s 133ms/step\n",
      "1/1 [==============================] - 0s 129ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 185ms/step\n",
      "1/1 [==============================] - 0s 111ms/step\n",
      "1/1 [==============================] - 0s 107ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 71ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 164ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 100ms/step\n",
      "1/1 [==============================] - 0s 172ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 251ms/step\n",
      "1/1 [==============================] - 0s 168ms/step\n",
      "1/1 [==============================] - 0s 266ms/step\n",
      "1/1 [==============================] - 0s 135ms/step\n",
      "1/1 [==============================] - 0s 121ms/step\n",
      "1/1 [==============================] - 0s 114ms/step\n",
      "1/1 [==============================] - 0s 105ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 75ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 97ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 251ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 95ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 152ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 97ms/step\n",
      "1/1 [==============================] - 0s 164ms/step\n",
      "1/1 [==============================] - 0s 158ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 85ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 119ms/step\n",
      "1/1 [==============================] - 0s 147ms/step\n",
      "1/1 [==============================] - 0s 128ms/step\n",
      "1/1 [==============================] - 0s 100ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 155ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 153ms/step\n",
      "1/1 [==============================] - 0s 130ms/step\n",
      "1/1 [==============================] - 0s 154ms/step\n",
      "1/1 [==============================] - 0s 189ms/step\n",
      "1/1 [==============================] - 0s 99ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 136ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 174ms/step\n",
      "1/1 [==============================] - 0s 166ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 127ms/step\n",
      "1/1 [==============================] - 0s 283ms/step\n",
      "1/1 [==============================] - 0s 77ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 185ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 82ms/step\n",
      "1/1 [==============================] - 0s 128ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 192ms/step\n",
      "1/1 [==============================] - 0s 147ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 122ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 109ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 99ms/step\n",
      "1/1 [==============================] - 0s 220ms/step\n",
      "ziffer_sortiert_resize\\3_dig7_20220205-102227.jpg 3 8\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=20x32>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 121ms/step\n",
      "1/1 [==============================] - 0s 88ms/step\n",
      "1/1 [==============================] - 0s 188ms/step\n",
      "1/1 [==============================] - 0s 204ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 106ms/step\n",
      "1/1 [==============================] - 0s 160ms/step\n",
      "1/1 [==============================] - 0s 211ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 134ms/step\n",
      "1/1 [==============================] - 0s 100ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 134ms/step\n",
      "1/1 [==============================] - 0s 220ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 251ms/step\n",
      "1/1 [==============================] - 0s 77ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 158ms/step\n",
      "1/1 [==============================] - 0s 74ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 84ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 131ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 126ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 228ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 157ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 95ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 136ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 119ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 173ms/step\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 120ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 92ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 76ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 89ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 60ms/step\n",
      "1/1 [==============================] - 0s 101ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 83ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 141ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 110ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 53ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 39ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 45ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 37ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 53ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 45ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 54ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 66ms/step\n",
      "1/1 [==============================] - 0s 89ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 64ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 28ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 69ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 53ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 76ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 67ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 78ms/step\n",
      "1/1 [==============================] - 0s 69ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 76ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 61ms/step\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 56ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 40ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 60ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 71ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 65ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 39ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 71ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 97ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 39ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 64ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 65ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 40ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 64ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 65ms/step\n",
      "1/1 [==============================] - 0s 56ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 43ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 74ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 24ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 53ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 66ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 56ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 59ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 91ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "ziffer_sortiert_resize\\9_dig7_20220205-104151.jpg 9 8\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=20x32>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 37ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 94ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 67ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 55ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 54ms/step\n",
      "1/1 [==============================] - 0s 60ms/step\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 54ms/step\n",
      "1/1 [==============================] - 0s 49ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 42ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 43ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 66ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 28ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 72ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 39ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 83ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 66ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 37ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 60ms/step\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 59ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 45ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 40ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 70ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 44ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 67ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 63ms/step\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 20ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 24ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 18ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 33ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 19ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 48ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 24ms/step\n",
      "1/1 [==============================] - 0s 34ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 74ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 41ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 40ms/step\n",
      "1/1 [==============================] - 0s 45ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 28ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 37ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 28ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 12ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 36ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 38ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 14ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 17ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 15ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 23ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 22ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 26ms/step\n",
      "1/1 [==============================] - 0s 28ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 24ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 39ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 29ms/step\n",
      "1/1 [==============================] - 0s 25ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 21ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 13ms/step\n",
      "1/1 [==============================] - 0s 32ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 30ms/step\n",
      "1/1 [==============================] - 0s 16ms/step\n",
      "1/1 [==============================] - 0s 27ms/step\n",
      "1/1 [==============================] - 0s 31ms/step\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "Input_dir='ziffer_sortiert_resize'\n",
    "res = []\n",
    "only_deviation = True\n",
    "show_wrong_image = True\n",
    "\n",
    "files = glob.glob(Input_dir + '/*.jpg')\n",
    "\n",
    "for aktfile in files:\n",
    "    base = os.path.basename(aktfile)\n",
    "    target = base[0:1]\n",
    "    if target == \"N\":\n",
    "        zw1 = -1\n",
    "    else:\n",
    "        zw1 = int(target)\n",
    "    expected_class = zw1\n",
    "    image_in = Image.open(aktfile)\n",
    "    test_image = np.array(image_in, dtype=\"float32\")\n",
    "    img = np.reshape(test_image,[1,32,20,3])\n",
    "    classes = np.argmax(model.predict(img), axis=-1)\n",
    "    classes = classes[0]\n",
    "    if classes == 10: \n",
    "        classes = -1\n",
    "    zw2 = classes\n",
    "    zw3 = zw2 - zw1\n",
    "    res.append(np.array([zw1, zw2, zw3]))\n",
    "    if only_deviation == True:\n",
    "        if str(classes) != str(expected_class):\n",
    "            print(aktfile + \" \" + str(expected_class) +  \" \" + str(classes))\n",
    "            if show_wrong_image == True:\n",
    "                display(image_in)\n",
    "    else:\n",
    "        print(aktfile + \" \" + aktsubdir +  \" \" + str(classes))\n",
    "        \n",
    "\n",
    "res = np.asarray(res)\n",
    "\n",
    "\n",
    "plt.plot(res[:,0])\n",
    "plt.plot(res[:,1])\n",
    "plt.title('Result')\n",
    "plt.ylabel('Digital Value')\n",
    "plt.xlabel('#Picture')\n",
    "plt.legend(['real','model'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "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": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: C:\\Users\\Muell\\AppData\\Local\\Temp\\tmpf_q8zp6p\\assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: C:\\Users\\Muell\\AppData\\Local\\Temp\\tmpf_q8zp6p\\assets\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "219272"
      ]
     },
     "execution_count": 7,
     "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": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:absl:Found untraced functions such as _jit_compiled_convolution_op, _jit_compiled_convolution_op, _jit_compiled_convolution_op while saving (showing 3 of 3). These functions will not be directly callable after loading.\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: C:\\Users\\Muell\\AppData\\Local\\Temp\\tmpnedsxt42\\assets\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "INFO:tensorflow:Assets written to: C:\\Users\\Muell\\AppData\\Local\\Temp\\tmpnedsxt42\\assets\n",
      "C:\\Users\\Muell\\anaconda3\\envs\\py39-tf-opencv\\lib\\site-packages\\tensorflow\\lite\\python\\convert.py:766: UserWarning: Statistics for quantized inputs were expected, but not specified; continuing anyway.\n",
      "  warnings.warn(\"Statistics for quantized inputs were expected, but not \"\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "dig1410s3q.tflite\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "62296"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "FileName = TFliteNamingAndVersion + \"q.tflite\"\n",
    "\n",
    "import tensorflow as tf\n",
    "\n",
    "def representative_dataset():\n",
    "    for n in range(x_data[0].size):\n",
    "      data = np.expand_dims(x_data[5], 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": "markdown",
   "metadata": {},
   "source": [
    "Checking the images shows, that this are border line images, which can be interpreted as a good digit or a faulty one."
   ]
  }
 ],
 "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}