MobileNet_7 classes_100_100_1st Experiment.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "resized_images_100_100_train_dir='D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Train'\n",
    "resized_images_100_100_test_dir='D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test'\n",
    "resized_images_100_100_validation_dir='D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Validation'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.python.keras.applications import MobileNet\n",
    "\n",
    "conv_base=MobileNet(weights='imagenet',include_top=False,input_shape=(100,100,3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "input_4 (InputLayer)         (None, 100, 100, 3)       0         \n",
      "_________________________________________________________________\n",
      "conv1_pad (ZeroPadding2D)    (None, 101, 101, 3)       0         \n",
      "_________________________________________________________________\n",
      "conv1 (Conv2D)               (None, 50, 50, 32)        864       \n",
      "_________________________________________________________________\n",
      "conv1_bn (BatchNormalization (None, 50, 50, 32)        128       \n",
      "_________________________________________________________________\n",
      "conv1_relu (ReLU)            (None, 50, 50, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv_dw_1 (DepthwiseConv2D)  (None, 50, 50, 32)        288       \n",
      "_________________________________________________________________\n",
      "conv_dw_1_bn (BatchNormaliza (None, 50, 50, 32)        128       \n",
      "_________________________________________________________________\n",
      "conv_dw_1_relu (ReLU)        (None, 50, 50, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv_pw_1 (Conv2D)           (None, 50, 50, 64)        2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_1_bn (BatchNormaliza (None, 50, 50, 64)        256       \n",
      "_________________________________________________________________\n",
      "conv_pw_1_relu (ReLU)        (None, 50, 50, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv_pad_2 (ZeroPadding2D)   (None, 51, 51, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv_dw_2 (DepthwiseConv2D)  (None, 25, 25, 64)        576       \n",
      "_________________________________________________________________\n",
      "conv_dw_2_bn (BatchNormaliza (None, 25, 25, 64)        256       \n",
      "_________________________________________________________________\n",
      "conv_dw_2_relu (ReLU)        (None, 25, 25, 64)        0         \n",
      "_________________________________________________________________\n",
      "conv_pw_2 (Conv2D)           (None, 25, 25, 128)       8192      \n",
      "_________________________________________________________________\n",
      "conv_pw_2_bn (BatchNormaliza (None, 25, 25, 128)       512       \n",
      "_________________________________________________________________\n",
      "conv_pw_2_relu (ReLU)        (None, 25, 25, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv_dw_3 (DepthwiseConv2D)  (None, 25, 25, 128)       1152      \n",
      "_________________________________________________________________\n",
      "conv_dw_3_bn (BatchNormaliza (None, 25, 25, 128)       512       \n",
      "_________________________________________________________________\n",
      "conv_dw_3_relu (ReLU)        (None, 25, 25, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv_pw_3 (Conv2D)           (None, 25, 25, 128)       16384     \n",
      "_________________________________________________________________\n",
      "conv_pw_3_bn (BatchNormaliza (None, 25, 25, 128)       512       \n",
      "_________________________________________________________________\n",
      "conv_pw_3_relu (ReLU)        (None, 25, 25, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv_pad_4 (ZeroPadding2D)   (None, 26, 26, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv_dw_4 (DepthwiseConv2D)  (None, 12, 12, 128)       1152      \n",
      "_________________________________________________________________\n",
      "conv_dw_4_bn (BatchNormaliza (None, 12, 12, 128)       512       \n",
      "_________________________________________________________________\n",
      "conv_dw_4_relu (ReLU)        (None, 12, 12, 128)       0         \n",
      "_________________________________________________________________\n",
      "conv_pw_4 (Conv2D)           (None, 12, 12, 256)       32768     \n",
      "_________________________________________________________________\n",
      "conv_pw_4_bn (BatchNormaliza (None, 12, 12, 256)       1024      \n",
      "_________________________________________________________________\n",
      "conv_pw_4_relu (ReLU)        (None, 12, 12, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv_dw_5 (DepthwiseConv2D)  (None, 12, 12, 256)       2304      \n",
      "_________________________________________________________________\n",
      "conv_dw_5_bn (BatchNormaliza (None, 12, 12, 256)       1024      \n",
      "_________________________________________________________________\n",
      "conv_dw_5_relu (ReLU)        (None, 12, 12, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv_pw_5 (Conv2D)           (None, 12, 12, 256)       65536     \n",
      "_________________________________________________________________\n",
      "conv_pw_5_bn (BatchNormaliza (None, 12, 12, 256)       1024      \n",
      "_________________________________________________________________\n",
      "conv_pw_5_relu (ReLU)        (None, 12, 12, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv_pad_6 (ZeroPadding2D)   (None, 13, 13, 256)       0         \n",
      "_________________________________________________________________\n",
      "conv_dw_6 (DepthwiseConv2D)  (None, 6, 6, 256)         2304      \n",
      "_________________________________________________________________\n",
      "conv_dw_6_bn (BatchNormaliza (None, 6, 6, 256)         1024      \n",
      "_________________________________________________________________\n",
      "conv_dw_6_relu (ReLU)        (None, 6, 6, 256)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_6 (Conv2D)           (None, 6, 6, 512)         131072    \n",
      "_________________________________________________________________\n",
      "conv_pw_6_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_6_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_7 (DepthwiseConv2D)  (None, 6, 6, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_7_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_7_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_7 (Conv2D)           (None, 6, 6, 512)         262144    \n",
      "_________________________________________________________________\n",
      "conv_pw_7_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_7_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_8 (DepthwiseConv2D)  (None, 6, 6, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_8_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_8_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_8 (Conv2D)           (None, 6, 6, 512)         262144    \n",
      "_________________________________________________________________\n",
      "conv_pw_8_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_8_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_9 (DepthwiseConv2D)  (None, 6, 6, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_9_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_9_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_9 (Conv2D)           (None, 6, 6, 512)         262144    \n",
      "_________________________________________________________________\n",
      "conv_pw_9_bn (BatchNormaliza (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_9_relu (ReLU)        (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_10 (DepthwiseConv2D) (None, 6, 6, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_10_bn (BatchNormaliz (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_10_relu (ReLU)       (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_10 (Conv2D)          (None, 6, 6, 512)         262144    \n",
      "_________________________________________________________________\n",
      "conv_pw_10_bn (BatchNormaliz (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_10_relu (ReLU)       (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_11 (DepthwiseConv2D) (None, 6, 6, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_11_bn (BatchNormaliz (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_11_relu (ReLU)       (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_11 (Conv2D)          (None, 6, 6, 512)         262144    \n",
      "_________________________________________________________________\n",
      "conv_pw_11_bn (BatchNormaliz (None, 6, 6, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_pw_11_relu (ReLU)       (None, 6, 6, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pad_12 (ZeroPadding2D)  (None, 7, 7, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_dw_12 (DepthwiseConv2D) (None, 3, 3, 512)         4608      \n",
      "_________________________________________________________________\n",
      "conv_dw_12_bn (BatchNormaliz (None, 3, 3, 512)         2048      \n",
      "_________________________________________________________________\n",
      "conv_dw_12_relu (ReLU)       (None, 3, 3, 512)         0         \n",
      "_________________________________________________________________\n",
      "conv_pw_12 (Conv2D)          (None, 3, 3, 1024)        524288    \n",
      "_________________________________________________________________\n",
      "conv_pw_12_bn (BatchNormaliz (None, 3, 3, 1024)        4096      \n",
      "_________________________________________________________________\n",
      "conv_pw_12_relu (ReLU)       (None, 3, 3, 1024)        0         \n",
      "_________________________________________________________________\n",
      "conv_dw_13 (DepthwiseConv2D) (None, 3, 3, 1024)        9216      \n",
      "_________________________________________________________________\n",
      "conv_dw_13_bn (BatchNormaliz (None, 3, 3, 1024)        4096      \n",
      "_________________________________________________________________\n",
      "conv_dw_13_relu (ReLU)       (None, 3, 3, 1024)        0         \n",
      "_________________________________________________________________\n",
      "conv_pw_13 (Conv2D)          (None, 3, 3, 1024)        1048576   \n",
      "_________________________________________________________________\n",
      "conv_pw_13_bn (BatchNormaliz (None, 3, 3, 1024)        4096      \n",
      "_________________________________________________________________\n",
      "conv_pw_13_relu (ReLU)       (None, 3, 3, 1024)        0         \n",
      "=================================================================\n",
      "Total params: 3,228,864\n",
      "Trainable params: 3,206,976\n",
      "Non-trainable params: 21,888\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "conv_base.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "mobilenet_1.00_224 (Model)   (None, 3, 3, 1024)        3228864   \n",
      "_________________________________________________________________\n",
      "flatten_2 (Flatten)          (None, 9216)              0         \n",
      "_________________________________________________________________\n",
      "dense_4 (Dense)              (None, 32)                294944    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 32)                0         \n",
      "_________________________________________________________________\n",
      "dense_5 (Dense)              (None, 7)                 231       \n",
      "=================================================================\n",
      "Total params: 3,524,039\n",
      "Trainable params: 3,502,151\n",
      "Non-trainable params: 21,888\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "import numpy as np\n",
    "import keras\n",
    "from tensorflow.python.keras import models\n",
    "from tensorflow.python.keras import layers\n",
    "from tensorflow.python.keras.preprocessing.image import ImageDataGenerator\n",
    "from tensorflow.python.keras import regularizers\n",
    "from tensorflow.python.keras import optimizers\n",
    "from tensorflow.python.keras.callbacks import ModelCheckpoint\n",
    "\n",
    "model = models.Sequential()\n",
    "model.add(conv_base)\n",
    "model.add(layers.Flatten())\n",
    "model.add(layers.Dense(32, kernel_regularizer=regularizers.l2(0.001), activation='relu'))\n",
    "model.add(layers.Dropout(0.2))\n",
    "model.add(layers.Dense(7, activation='softmax'))\n",
    "\n",
    "model.compile(loss='categorical_crossentropy',optimizer=optimizers.Adam(lr=1e-4),metrics=['categorical_accuracy'])\n",
    "model.summary()\n",
    "\n",
    "filepath=\"MobileNet_Indus_Fruits_7_classes_100_100_1st Experiment.h5\"\n",
    "checkpoint = ModelCheckpoint(filepath, save_best_only=True)\n",
    "callbacks_list = [checkpoint]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 2800 images belonging to 7 classes.\n",
      "Found 280 images belonging to 7 classes.\n",
      "Found 560 images belonging to 7 classes.\n"
     ]
    }
   ],
   "source": [
    "batch_size=16\n",
    "\n",
    "train_datagen = ImageDataGenerator(\n",
    "    rescale=1./255,\n",
    "    rotation_range=40,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2,\n",
    "    shear_range=0.2,\n",
    "    zoom_range=0.2,\n",
    "    horizontal_flip=True,)\n",
    "\n",
    "# Note that the validation  and test data should not be augmented!\n",
    "validation_datagen=ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "test_datagen = ImageDataGenerator(rescale=1./255)\n",
    "\n",
    "train_generator = train_datagen.flow_from_directory(\n",
    "        # This is the target directory\n",
    "        resized_images_100_100_train_dir,\n",
    "        target_size=(100, 100),\n",
    "        batch_size=batch_size)\n",
    "\n",
    "validation_generator = validation_datagen.flow_from_directory(\n",
    "        resized_images_100_100_validation_dir,\n",
    "        target_size=(100, 100),\n",
    "        batch_size=batch_size, shuffle=False)\n",
    "\n",
    "test_generator = test_datagen.flow_from_directory(\n",
    "        resized_images_100_100_test_dir,\n",
    "        target_size=(100, 100),\n",
    "        batch_size=batch_size, shuffle=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "number_of_training_samples=2800\n",
    "number_of_validation_samples=280\n",
    "number_of_test_samples=560"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From C:\\Users\\Raheel\\Anaconda3\\envs\\EnvName\\lib\\site-packages\\tensorflow\\python\\ops\\math_ops.py:3066: to_int32 (from tensorflow.python.ops.math_ops) is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Use tf.cast instead.\n",
      "Epoch 1/200\n",
      "18/18 [==============================] - 2s 129ms/step - loss: 0.7742 - categorical_accuracy: 0.7536\n",
      "175/175 [==============================] - 46s 261ms/step - loss: 1.5113 - categorical_accuracy: 0.4932 - val_loss: 0.7742 - val_categorical_accuracy: 0.7536\n",
      "Epoch 2/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.5626 - categorical_accuracy: 0.8464\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.9624 - categorical_accuracy: 0.6954 - val_loss: 0.5626 - val_categorical_accuracy: 0.8464\n",
      "Epoch 3/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.4651 - categorical_accuracy: 0.8607A: 0s - loss: 0.7485 - categori\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.7469 - categorical_accuracy: 0.7729 - val_loss: 0.4651 - val_categorical_accuracy: 0.8607\n",
      "Epoch 4/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4574 - categorical_accuracy: 0.8500\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.6123 - categorical_accuracy: 0.8271 - val_loss: 0.4574 - val_categorical_accuracy: 0.8500\n",
      "Epoch 5/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4772 - categorical_accuracy: 0.8607\n",
      "175/175 [==============================] - 12s 69ms/step - loss: 0.5683 - categorical_accuracy: 0.8396 - val_loss: 0.4772 - val_categorical_accuracy: 0.8607\n",
      "Epoch 6/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4337 - categorical_accuracy: 0.8679\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.5341 - categorical_accuracy: 0.8504 - val_loss: 0.4337 - val_categorical_accuracy: 0.8679\n",
      "Epoch 7/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3717 - categorical_accuracy: 0.9036\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.4925 - categorical_accuracy: 0.8668 - val_loss: 0.3717 - val_categorical_accuracy: 0.9036\n",
      "Epoch 8/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3607 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 13s 77ms/step - loss: 0.4560 - categorical_accuracy: 0.8757 - val_loss: 0.3607 - val_categorical_accuracy: 0.9286\n",
      "Epoch 9/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4128 - categorical_accuracy: 0.9000\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.4175 - categorical_accuracy: 0.8857 - val_loss: 0.4128 - val_categorical_accuracy: 0.9000\n",
      "Epoch 10/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2797 - categorical_accuracy: 0.9214A: 0s - loss: 0.3921 - categorical_accu\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.3918 - categorical_accuracy: 0.8961 - val_loss: 0.2797 - val_categorical_accuracy: 0.9214\n",
      "Epoch 11/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3420 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.3807 - categorical_accuracy: 0.8961 - val_loss: 0.3420 - val_categorical_accuracy: 0.9214\n",
      "Epoch 12/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3337 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.3721 - categorical_accuracy: 0.9004 - val_loss: 0.3337 - val_categorical_accuracy: 0.9179\n",
      "Epoch 13/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3504 - categorical_accuracy: 0.9107\n",
      "175/175 [==============================] - 12s 69ms/step - loss: 0.3380 - categorical_accuracy: 0.9154 - val_loss: 0.3504 - val_categorical_accuracy: 0.9107\n",
      "Epoch 14/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3218 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.3258 - categorical_accuracy: 0.9204 - val_loss: 0.3218 - val_categorical_accuracy: 0.9393\n",
      "Epoch 15/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3152 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 69ms/step - loss: 0.3319 - categorical_accuracy: 0.9118 - val_loss: 0.3152 - val_categorical_accuracy: 0.9214\n",
      "Epoch 16/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3009 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.3041 - categorical_accuracy: 0.9214 - val_loss: 0.3009 - val_categorical_accuracy: 0.9286\n",
      "Epoch 17/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2214 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.2749 - categorical_accuracy: 0.9296 - val_loss: 0.2214 - val_categorical_accuracy: 0.9607\n",
      "Epoch 18/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2964 - categorical_accuracy: 0.9357 0s - loss: 0.4009 - categorical_accuracy: \n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2872 - categorical_accuracy: 0.9282 - val_loss: 0.2964 - val_categorical_accuracy: 0.9357\n",
      "Epoch 19/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2598 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2555 - categorical_accuracy: 0.9364 - val_loss: 0.2598 - val_categorical_accuracy: 0.9286\n",
      "Epoch 20/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2956 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2820 - categorical_accuracy: 0.9332 - val_loss: 0.2956 - val_categorical_accuracy: 0.9214\n",
      "Epoch 21/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3715 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2848 - categorical_accuracy: 0.9293 - val_loss: 0.3715 - val_categorical_accuracy: 0.9214\n",
      "Epoch 22/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2580 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.2548 - categorical_accuracy: 0.9421 - val_loss: 0.2580 - val_categorical_accuracy: 0.9429\n",
      "Epoch 23/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2756 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2524 - categorical_accuracy: 0.9418 - val_loss: 0.2756 - val_categorical_accuracy: 0.9214\n",
      "Epoch 24/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2978 - categorical_accuracy: 0.9179A: 0s - loss: 0.2509 - categori\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.2496 - categorical_accuracy: 0.9446 - val_loss: 0.2978 - val_categorical_accuracy: 0.9179\n",
      "Epoch 25/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3289 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2590 - categorical_accuracy: 0.9400 - val_loss: 0.3289 - val_categorical_accuracy: 0.9357\n",
      "Epoch 26/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3599 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2600 - categorical_accuracy: 0.9400 - val_loss: 0.3599 - val_categorical_accuracy: 0.9500\n",
      "Epoch 27/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3059 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2198 - categorical_accuracy: 0.9486 - val_loss: 0.3059 - val_categorical_accuracy: 0.9250\n",
      "Epoch 28/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3286 - categorical_accuracy: 0.9429A: \n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1954 - categorical_accuracy: 0.9546 - val_loss: 0.3286 - val_categorical_accuracy: 0.9429\n",
      "Epoch 29/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2426 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2354 - categorical_accuracy: 0.9500 - val_loss: 0.2426 - val_categorical_accuracy: 0.9321\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 30/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2444 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.2382 - categorical_accuracy: 0.9518 - val_loss: 0.2444 - val_categorical_accuracy: 0.9464\n",
      "Epoch 31/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3658 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2475 - categorical_accuracy: 0.9439 - val_loss: 0.3658 - val_categorical_accuracy: 0.9250\n",
      "Epoch 32/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3604 - categorical_accuracy: 0.9143A: 0s - loss: 0\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2096 - categorical_accuracy: 0.9525 - val_loss: 0.3604 - val_categorical_accuracy: 0.9143\n",
      "Epoch 33/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3581 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2172 - categorical_accuracy: 0.9550 - val_loss: 0.3581 - val_categorical_accuracy: 0.9179\n",
      "Epoch 34/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3977 - categorical_accuracy: 0.9357 0s - loss: 0.4356 - categorical_accuracy: 0.\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1934 - categorical_accuracy: 0.9571 - val_loss: 0.3977 - val_categorical_accuracy: 0.9357\n",
      "Epoch 35/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3372 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1947 - categorical_accuracy: 0.9525 - val_loss: 0.3372 - val_categorical_accuracy: 0.9429\n",
      "Epoch 36/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3126 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2305 - categorical_accuracy: 0.9450 - val_loss: 0.3126 - val_categorical_accuracy: 0.9357\n",
      "Epoch 37/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2327 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2005 - categorical_accuracy: 0.9514 - val_loss: 0.2327 - val_categorical_accuracy: 0.9429\n",
      "Epoch 38/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3598 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.2177 - categorical_accuracy: 0.9571 - val_loss: 0.3598 - val_categorical_accuracy: 0.9179\n",
      "Epoch 39/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2953 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1965 - categorical_accuracy: 0.9575 - val_loss: 0.2953 - val_categorical_accuracy: 0.9464\n",
      "Epoch 40/200\n",
      "18/18 [==============================] - 0s 26ms/step - loss: 0.2732 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 77ms/step - loss: 0.1779 - categorical_accuracy: 0.9586 - val_loss: 0.2732 - val_categorical_accuracy: 0.9321\n",
      "Epoch 41/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2755 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.1772 - categorical_accuracy: 0.9596 - val_loss: 0.2755 - val_categorical_accuracy: 0.9357\n",
      "Epoch 42/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3639 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1736 - categorical_accuracy: 0.9593 - val_loss: 0.3639 - val_categorical_accuracy: 0.9321\n",
      "Epoch 43/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3856 - categorical_accuracy: 0.9143\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1721 - categorical_accuracy: 0.9679 - val_loss: 0.3856 - val_categorical_accuracy: 0.9143\n",
      "Epoch 44/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3147 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.1821 - categorical_accuracy: 0.9632 - val_loss: 0.3147 - val_categorical_accuracy: 0.9250\n",
      "Epoch 45/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.4396 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1811 - categorical_accuracy: 0.9614 - val_loss: 0.4396 - val_categorical_accuracy: 0.9179\n",
      "Epoch 46/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3059 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1962 - categorical_accuracy: 0.9596 - val_loss: 0.3059 - val_categorical_accuracy: 0.9321\n",
      "Epoch 47/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2896 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1665 - categorical_accuracy: 0.9643 - val_loss: 0.2896 - val_categorical_accuracy: 0.9393\n",
      "Epoch 48/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2890 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1749 - categorical_accuracy: 0.9618 - val_loss: 0.2890 - val_categorical_accuracy: 0.9393\n",
      "Epoch 49/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2889 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1770 - categorical_accuracy: 0.9646 - val_loss: 0.2889 - val_categorical_accuracy: 0.9357\n",
      "Epoch 50/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3929 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1665 - categorical_accuracy: 0.9686 - val_loss: 0.3929 - val_categorical_accuracy: 0.9250\n",
      "Epoch 51/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3487 - categorical_accuracy: 0.9464A: 1s - l\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1830 - categorical_accuracy: 0.9589 - val_loss: 0.3487 - val_categorical_accuracy: 0.9464\n",
      "Epoch 52/200\n",
      "18/18 [==============================] - 0s 27ms/step - loss: 0.3254 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.1518 - categorical_accuracy: 0.9679 - val_loss: 0.3254 - val_categorical_accuracy: 0.9357\n",
      "Epoch 53/200\n",
      "18/18 [==============================] - ETA: 0s - loss: 0.3114 - categorical_accuracy: 0.94 - 0s 23ms/step - loss: 0.2972 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1545 - categorical_accuracy: 0.9714 - val_loss: 0.2972 - val_categorical_accuracy: 0.9500\n",
      "Epoch 54/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2531 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1405 - categorical_accuracy: 0.9725 - val_loss: 0.2531 - val_categorical_accuracy: 0.9500\n",
      "Epoch 55/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3326 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1536 - categorical_accuracy: 0.9689 - val_loss: 0.3326 - val_categorical_accuracy: 0.9429\n",
      "Epoch 56/200\n",
      "18/18 [==============================] - 0s 27ms/step - loss: 0.2392 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.1614 - categorical_accuracy: 0.9654 - val_loss: 0.2392 - val_categorical_accuracy: 0.9429\n",
      "Epoch 57/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2681 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.1732 - categorical_accuracy: 0.9696 - val_loss: 0.2681 - val_categorical_accuracy: 0.9607\n",
      "Epoch 58/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2306 - categorical_accuracy: 0.9571\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1507 - categorical_accuracy: 0.9711 - val_loss: 0.2306 - val_categorical_accuracy: 0.9571\n",
      "Epoch 59/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2487 - categorical_accuracy: 0.9571\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1490 - categorical_accuracy: 0.9675 - val_loss: 0.2487 - val_categorical_accuracy: 0.9571\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 60/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2324 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1315 - categorical_accuracy: 0.9750 - val_loss: 0.2324 - val_categorical_accuracy: 0.9536\n",
      "Epoch 61/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2587 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1340 - categorical_accuracy: 0.9739 - val_loss: 0.2587 - val_categorical_accuracy: 0.9321\n",
      "Epoch 62/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2547 - categorical_accuracy: 0.9536A: 1s - l\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1366 - categorical_accuracy: 0.9729 - val_loss: 0.2547 - val_categorical_accuracy: 0.9536\n",
      "Epoch 63/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2193 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.1712 - categorical_accuracy: 0.9664 - val_loss: 0.2193 - val_categorical_accuracy: 0.9500\n",
      "Epoch 64/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.1962 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.1443 - categorical_accuracy: 0.9711 - val_loss: 0.1962 - val_categorical_accuracy: 0.9607\n",
      "Epoch 65/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2985 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1249 - categorical_accuracy: 0.9771 - val_loss: 0.2985 - val_categorical_accuracy: 0.9250\n",
      "Epoch 66/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2911 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1600 - categorical_accuracy: 0.9679 - val_loss: 0.2911 - val_categorical_accuracy: 0.9429\n",
      "Epoch 67/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3293 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1296 - categorical_accuracy: 0.9779 - val_loss: 0.3293 - val_categorical_accuracy: 0.9429\n",
      "Epoch 68/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2701 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1274 - categorical_accuracy: 0.9779 - val_loss: 0.2701 - val_categorical_accuracy: 0.9536\n",
      "Epoch 69/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2862 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1348 - categorical_accuracy: 0.9696 - val_loss: 0.2862 - val_categorical_accuracy: 0.9500\n",
      "Epoch 70/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3393 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1135 - categorical_accuracy: 0.9829 - val_loss: 0.3393 - val_categorical_accuracy: 0.9393\n",
      "Epoch 71/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3014 - categorical_accuracy: 0.9607 0s - loss: 0.3652 - categorical_accuracy: 0.\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1247 - categorical_accuracy: 0.9768 - val_loss: 0.3014 - val_categorical_accuracy: 0.9607\n",
      "Epoch 72/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.4008 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1549 - categorical_accuracy: 0.9714 - val_loss: 0.4008 - val_categorical_accuracy: 0.9179\n",
      "Epoch 73/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3654 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.1282 - categorical_accuracy: 0.9764 - val_loss: 0.3654 - val_categorical_accuracy: 0.9393\n",
      "Epoch 74/200\n",
      "18/18 [==============================] - 1s 31ms/step - loss: 0.4145 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1339 - categorical_accuracy: 0.9754 - val_loss: 0.4145 - val_categorical_accuracy: 0.9179\n",
      "Epoch 75/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.3585 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.1295 - categorical_accuracy: 0.9786 - val_loss: 0.3585 - val_categorical_accuracy: 0.9536\n",
      "Epoch 76/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4512 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 13s 77ms/step - loss: 0.1643 - categorical_accuracy: 0.9686 - val_loss: 0.4512 - val_categorical_accuracy: 0.9214\n",
      "Epoch 77/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3887 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1105 - categorical_accuracy: 0.9796 - val_loss: 0.3887 - val_categorical_accuracy: 0.9214\n",
      "Epoch 78/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3645 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1015 - categorical_accuracy: 0.9843 - val_loss: 0.3645 - val_categorical_accuracy: 0.9321\n",
      "Epoch 79/200\n",
      "18/18 [==============================] - 1s 32ms/step - loss: 0.3360 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 14s 78ms/step - loss: 0.1147 - categorical_accuracy: 0.9811 - val_loss: 0.3360 - val_categorical_accuracy: 0.9464\n",
      "Epoch 80/200\n",
      "18/18 [==============================] - 1s 35ms/step - loss: 0.3797 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 15s 88ms/step - loss: 0.1128 - categorical_accuracy: 0.9811 - val_loss: 0.3797 - val_categorical_accuracy: 0.9250\n",
      "Epoch 81/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.4369 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1172 - categorical_accuracy: 0.9807 - val_loss: 0.4369 - val_categorical_accuracy: 0.9286\n",
      "Epoch 82/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4558 - categorical_accuracy: 0.9286 0s - loss: 0.5780 - categorical_accuracy: 0.\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.1140 - categorical_accuracy: 0.9804 - val_loss: 0.4558 - val_categorical_accuracy: 0.9286\n",
      "Epoch 83/200\n",
      "18/18 [==============================] - 0s 28ms/step - loss: 0.3628 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1181 - categorical_accuracy: 0.9771 - val_loss: 0.3628 - val_categorical_accuracy: 0.9357\n",
      "Epoch 84/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4843 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.1319 - categorical_accuracy: 0.9750 - val_loss: 0.4843 - val_categorical_accuracy: 0.9250\n",
      "Epoch 85/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3779 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1143 - categorical_accuracy: 0.9779 - val_loss: 0.3779 - val_categorical_accuracy: 0.9214\n",
      "Epoch 86/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3315 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1172 - categorical_accuracy: 0.9786 - val_loss: 0.3315 - val_categorical_accuracy: 0.9357\n",
      "Epoch 87/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.4007 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1212 - categorical_accuracy: 0.9771 - val_loss: 0.4007 - val_categorical_accuracy: 0.9357\n",
      "Epoch 88/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2384 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1248 - categorical_accuracy: 0.9793 - val_loss: 0.2384 - val_categorical_accuracy: 0.9429\n",
      "Epoch 89/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3822 - categorical_accuracy: 0.9286ETA: 0s - loss: 1.2477 - categorical_accu\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0964 - categorical_accuracy: 0.9818 - val_loss: 0.3822 - val_categorical_accuracy: 0.9286\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 90/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2853 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1052 - categorical_accuracy: 0.9804 - val_loss: 0.2853 - val_categorical_accuracy: 0.9357\n",
      "Epoch 91/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.3381 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.1264 - categorical_accuracy: 0.9775 - val_loss: 0.3381 - val_categorical_accuracy: 0.9464\n",
      "Epoch 92/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2871 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1048 - categorical_accuracy: 0.9807 - val_loss: 0.2871 - val_categorical_accuracy: 0.9500\n",
      "Epoch 93/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2977 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1097 - categorical_accuracy: 0.9796 - val_loss: 0.2977 - val_categorical_accuracy: 0.9321\n",
      "Epoch 94/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2933 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1295 - categorical_accuracy: 0.9771 - val_loss: 0.2933 - val_categorical_accuracy: 0.9357\n",
      "Epoch 95/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3180 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1165 - categorical_accuracy: 0.9786 - val_loss: 0.3180 - val_categorical_accuracy: 0.9393\n",
      "Epoch 96/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3857 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0844 - categorical_accuracy: 0.9832 - val_loss: 0.3857 - val_categorical_accuracy: 0.9393\n",
      "Epoch 97/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3465 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0917 - categorical_accuracy: 0.9814 - val_loss: 0.3465 - val_categorical_accuracy: 0.9429\n",
      "Epoch 98/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2572 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0982 - categorical_accuracy: 0.9836 - val_loss: 0.2572 - val_categorical_accuracy: 0.9536\n",
      "Epoch 99/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3894 - categorical_accuracy: 0.9179\n",
      "175/175 [==============================] - 12s 69ms/step - loss: 0.1009 - categorical_accuracy: 0.9836 - val_loss: 0.3894 - val_categorical_accuracy: 0.9179\n",
      "Epoch 100/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3146 - categorical_accuracy: 0.9357 0s - loss: 0.4761 - categorical_accuracy: \n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0862 - categorical_accuracy: 0.9857 - val_loss: 0.3146 - val_categorical_accuracy: 0.9357\n",
      "Epoch 101/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2958 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1255 - categorical_accuracy: 0.9761 - val_loss: 0.2958 - val_categorical_accuracy: 0.9500\n",
      "Epoch 102/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3146 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1063 - categorical_accuracy: 0.9786 - val_loss: 0.3146 - val_categorical_accuracy: 0.9429\n",
      "Epoch 103/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3718 - categorical_accuracy: 0.9143\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0851 - categorical_accuracy: 0.9861 - val_loss: 0.3718 - val_categorical_accuracy: 0.9143\n",
      "Epoch 104/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2604 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1075 - categorical_accuracy: 0.9796 - val_loss: 0.2604 - val_categorical_accuracy: 0.9536\n",
      "Epoch 105/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3536 - categorical_accuracy: 0.9214\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1105 - categorical_accuracy: 0.9771 - val_loss: 0.3536 - val_categorical_accuracy: 0.9214\n",
      "Epoch 106/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2750 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1057 - categorical_accuracy: 0.9804 - val_loss: 0.2750 - val_categorical_accuracy: 0.9464\n",
      "Epoch 107/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3307 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0930 - categorical_accuracy: 0.9836 - val_loss: 0.3307 - val_categorical_accuracy: 0.9500\n",
      "Epoch 108/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2875 - categorical_accuracy: 0.9464A: \n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0678 - categorical_accuracy: 0.9904 - val_loss: 0.2875 - val_categorical_accuracy: 0.9464\n",
      "Epoch 109/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2510 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1133 - categorical_accuracy: 0.9771 - val_loss: 0.2510 - val_categorical_accuracy: 0.9464\n",
      "Epoch 110/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2718 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0930 - categorical_accuracy: 0.9836 - val_loss: 0.2718 - val_categorical_accuracy: 0.9500\n",
      "Epoch 111/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2655 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.1027 - categorical_accuracy: 0.9829 - val_loss: 0.2655 - val_categorical_accuracy: 0.9429\n",
      "Epoch 112/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.1848 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.0925 - categorical_accuracy: 0.9836 - val_loss: 0.1848 - val_categorical_accuracy: 0.9607\n",
      "Epoch 113/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2464 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.0899 - categorical_accuracy: 0.9832 - val_loss: 0.2464 - val_categorical_accuracy: 0.9357\n",
      "Epoch 114/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2782 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.1053 - categorical_accuracy: 0.9782 - val_loss: 0.2782 - val_categorical_accuracy: 0.9393\n",
      "Epoch 115/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3443 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0986 - categorical_accuracy: 0.9846 - val_loss: 0.3443 - val_categorical_accuracy: 0.9429\n",
      "Epoch 116/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3235 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0656 - categorical_accuracy: 0.9904 - val_loss: 0.3235 - val_categorical_accuracy: 0.9393\n",
      "Epoch 117/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3538 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.1084 - categorical_accuracy: 0.9818 - val_loss: 0.3538 - val_categorical_accuracy: 0.9500\n",
      "Epoch 118/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2626 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0835 - categorical_accuracy: 0.9846 - val_loss: 0.2626 - val_categorical_accuracy: 0.9429\n",
      "Epoch 119/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2441 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0857 - categorical_accuracy: 0.9861 - val_loss: 0.2441 - val_categorical_accuracy: 0.9536\n",
      "Epoch 120/200\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2748 - categorical_accuracy: 0.9536 0s - loss: 0.2529 - categorical_accuracy: 0.95\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0763 - categorical_accuracy: 0.9886 - val_loss: 0.2748 - val_categorical_accuracy: 0.9536\n",
      "Epoch 121/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2943 - categorical_accuracy: 0.9536 0s - loss: 0.4041 - categorical_accuracy: \n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.1026 - categorical_accuracy: 0.9850 - val_loss: 0.2943 - val_categorical_accuracy: 0.9536\n",
      "Epoch 122/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3801 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 14s 77ms/step - loss: 0.0767 - categorical_accuracy: 0.9850 - val_loss: 0.3801 - val_categorical_accuracy: 0.9357\n",
      "Epoch 123/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3567 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0793 - categorical_accuracy: 0.9879 - val_loss: 0.3567 - val_categorical_accuracy: 0.9464\n",
      "Epoch 124/200\n",
      "18/18 [==============================] - 1s 34ms/step - loss: 0.2959 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 15s 84ms/step - loss: 0.0888 - categorical_accuracy: 0.9843 - val_loss: 0.2959 - val_categorical_accuracy: 0.9429\n",
      "Epoch 125/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2771 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.0964 - categorical_accuracy: 0.9807 - val_loss: 0.2771 - val_categorical_accuracy: 0.9536\n",
      "Epoch 126/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2593 - categorical_accuracy: 0.9571\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0893 - categorical_accuracy: 0.9825 - val_loss: 0.2593 - val_categorical_accuracy: 0.9571\n",
      "Epoch 127/200\n",
      "18/18 [==============================] - 1s 28ms/step - loss: 0.2548 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1124 - categorical_accuracy: 0.9768 - val_loss: 0.2548 - val_categorical_accuracy: 0.9393\n",
      "Epoch 128/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3389 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0817 - categorical_accuracy: 0.9857 - val_loss: 0.3389 - val_categorical_accuracy: 0.9321\n",
      "Epoch 129/200\n",
      "18/18 [==============================] - 0s 28ms/step - loss: 0.3504 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.0738 - categorical_accuracy: 0.9893 - val_loss: 0.3504 - val_categorical_accuracy: 0.9321\n",
      "Epoch 130/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2258 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 77ms/step - loss: 0.0811 - categorical_accuracy: 0.9861 - val_loss: 0.2258 - val_categorical_accuracy: 0.9429\n",
      "Epoch 131/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3026 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0806 - categorical_accuracy: 0.9879 - val_loss: 0.3026 - val_categorical_accuracy: 0.9393\n",
      "Epoch 132/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.4496 - categorical_accuracy: 0.9000\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.1023 - categorical_accuracy: 0.9818 - val_loss: 0.4496 - val_categorical_accuracy: 0.9000\n",
      "Epoch 133/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2995 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0831 - categorical_accuracy: 0.9854 - val_loss: 0.2995 - val_categorical_accuracy: 0.9429\n",
      "Epoch 134/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2916 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 14s 78ms/step - loss: 0.0876 - categorical_accuracy: 0.9825 - val_loss: 0.2916 - val_categorical_accuracy: 0.9321\n",
      "Epoch 135/200\n",
      "18/18 [==============================] - 0s 27ms/step - loss: 0.3018 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.0540 - categorical_accuracy: 0.9929 - val_loss: 0.3018 - val_categorical_accuracy: 0.9429\n",
      "Epoch 136/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2926 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.0706 - categorical_accuracy: 0.9882 - val_loss: 0.2926 - val_categorical_accuracy: 0.9393\n",
      "Epoch 137/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2649 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.0707 - categorical_accuracy: 0.9882 - val_loss: 0.2649 - val_categorical_accuracy: 0.9321\n",
      "Epoch 138/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3014 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.0722 - categorical_accuracy: 0.9850 - val_loss: 0.3014 - val_categorical_accuracy: 0.9393\n",
      "Epoch 139/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2239 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.0781 - categorical_accuracy: 0.9843 - val_loss: 0.2239 - val_categorical_accuracy: 0.9500\n",
      "Epoch 140/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2485 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0597 - categorical_accuracy: 0.9936 - val_loss: 0.2485 - val_categorical_accuracy: 0.9321\n",
      "Epoch 141/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2765 - categorical_accuracy: 0.9393 0s - loss: 0.3775 - categorical_accuracy: 0. - ETA: 0s - loss: 0.2911 - categorical_accuracy: 0.93\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.0713 - categorical_accuracy: 0.9871 - val_loss: 0.2765 - val_categorical_accuracy: 0.9393\n",
      "Epoch 142/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2619 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 14s 77ms/step - loss: 0.0973 - categorical_accuracy: 0.9811 - val_loss: 0.2619 - val_categorical_accuracy: 0.9500\n",
      "Epoch 143/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2318 - categorical_accuracy: 0.9571\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0519 - categorical_accuracy: 0.9932 - val_loss: 0.2318 - val_categorical_accuracy: 0.9571\n",
      "Epoch 144/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2739 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0784 - categorical_accuracy: 0.9839 - val_loss: 0.2739 - val_categorical_accuracy: 0.9429\n",
      "Epoch 145/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2828 - categorical_accuracy: 0.9357\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0585 - categorical_accuracy: 0.9907 - val_loss: 0.2828 - val_categorical_accuracy: 0.9357\n",
      "Epoch 146/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2152 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0587 - categorical_accuracy: 0.9886 - val_loss: 0.2152 - val_categorical_accuracy: 0.9607\n",
      "Epoch 147/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3649 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0746 - categorical_accuracy: 0.9882 - val_loss: 0.3649 - val_categorical_accuracy: 0.9286\n",
      "Epoch 148/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2775 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0889 - categorical_accuracy: 0.9857 - val_loss: 0.2775 - val_categorical_accuracy: 0.9321\n",
      "Epoch 149/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2122 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0674 - categorical_accuracy: 0.9889 - val_loss: 0.2122 - val_categorical_accuracy: 0.9536\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 150/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2762 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0588 - categorical_accuracy: 0.9886 - val_loss: 0.2762 - val_categorical_accuracy: 0.9393\n",
      "Epoch 151/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2612 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0740 - categorical_accuracy: 0.9857 - val_loss: 0.2612 - val_categorical_accuracy: 0.9429\n",
      "Epoch 152/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2520 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0784 - categorical_accuracy: 0.9857 - val_loss: 0.2520 - val_categorical_accuracy: 0.9500\n",
      "Epoch 153/200\n",
      "18/18 [==============================] - 0s 27ms/step - loss: 0.2762 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 77ms/step - loss: 0.0842 - categorical_accuracy: 0.9846 - val_loss: 0.2762 - val_categorical_accuracy: 0.9429\n",
      "Epoch 154/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2648 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.0630 - categorical_accuracy: 0.9900 - val_loss: 0.2648 - val_categorical_accuracy: 0.9429\n",
      "Epoch 155/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3317 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0589 - categorical_accuracy: 0.9914 - val_loss: 0.3317 - val_categorical_accuracy: 0.9429\n",
      "Epoch 156/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2264 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0639 - categorical_accuracy: 0.9907 - val_loss: 0.2264 - val_categorical_accuracy: 0.9500\n",
      "Epoch 157/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3553 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0630 - categorical_accuracy: 0.9886 - val_loss: 0.3553 - val_categorical_accuracy: 0.9250\n",
      "Epoch 158/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2338 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0480 - categorical_accuracy: 0.9914 - val_loss: 0.2338 - val_categorical_accuracy: 0.9500\n",
      "Epoch 159/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2057 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0743 - categorical_accuracy: 0.9861 - val_loss: 0.2057 - val_categorical_accuracy: 0.9500\n",
      "Epoch 160/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2323 - categorical_accuracy: 0.9429A: 1s - loss: 0.0648 - categorical_accuracy - ETA: 1s -\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0621 - categorical_accuracy: 0.9886 - val_loss: 0.2323 - val_categorical_accuracy: 0.9429\n",
      "Epoch 161/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.1669 - categorical_accuracy: 0.9643\n",
      "175/175 [==============================] - 13s 77ms/step - loss: 0.0596 - categorical_accuracy: 0.9904 - val_loss: 0.1669 - val_categorical_accuracy: 0.9643\n",
      "Epoch 162/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2144 - categorical_accuracy: 0.9536 0s - loss: 0.2980 - categorical_accuracy: \n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0785 - categorical_accuracy: 0.9896 - val_loss: 0.2144 - val_categorical_accuracy: 0.9536\n",
      "Epoch 163/200\n",
      "18/18 [==============================] - 0s 26ms/step - loss: 0.2835 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0661 - categorical_accuracy: 0.9886 - val_loss: 0.2835 - val_categorical_accuracy: 0.9393\n",
      "Epoch 164/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2545 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0622 - categorical_accuracy: 0.9900 - val_loss: 0.2545 - val_categorical_accuracy: 0.9464\n",
      "Epoch 165/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2726 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0730 - categorical_accuracy: 0.9861 - val_loss: 0.2726 - val_categorical_accuracy: 0.9393\n",
      "Epoch 166/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2869 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0552 - categorical_accuracy: 0.9893 - val_loss: 0.2869 - val_categorical_accuracy: 0.9429\n",
      "Epoch 167/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2239 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0599 - categorical_accuracy: 0.9882 - val_loss: 0.2239 - val_categorical_accuracy: 0.9429\n",
      "Epoch 168/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2041 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0571 - categorical_accuracy: 0.9907 - val_loss: 0.2041 - val_categorical_accuracy: 0.9500\n",
      "Epoch 169/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2155 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0709 - categorical_accuracy: 0.9868 - val_loss: 0.2155 - val_categorical_accuracy: 0.9607\n",
      "Epoch 170/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2044 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0600 - categorical_accuracy: 0.9896 - val_loss: 0.2044 - val_categorical_accuracy: 0.9429\n",
      "Epoch 171/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.1944 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0559 - categorical_accuracy: 0.9907 - val_loss: 0.1944 - val_categorical_accuracy: 0.9536\n",
      "Epoch 172/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.3296 - categorical_accuracy: 0.9321\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0658 - categorical_accuracy: 0.9857 - val_loss: 0.3296 - val_categorical_accuracy: 0.9321\n",
      "Epoch 173/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2519 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0634 - categorical_accuracy: 0.9886 - val_loss: 0.2519 - val_categorical_accuracy: 0.9500\n",
      "Epoch 174/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3215 - categorical_accuracy: 0.9393\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0529 - categorical_accuracy: 0.9921 - val_loss: 0.3215 - val_categorical_accuracy: 0.9393\n",
      "Epoch 175/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3226 - categorical_accuracy: 0.9250\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0694 - categorical_accuracy: 0.9868 - val_loss: 0.3226 - val_categorical_accuracy: 0.9250\n",
      "Epoch 176/200\n",
      "18/18 [==============================] - 0s 21ms/step - loss: 0.2981 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0634 - categorical_accuracy: 0.9893 - val_loss: 0.2981 - val_categorical_accuracy: 0.9464\n",
      "Epoch 177/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2399 - categorical_accuracy: 0.9464A: 0s - loss: 0.0635 - categorical_ac\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0648 - categorical_accuracy: 0.9864 - val_loss: 0.2399 - val_categorical_accuracy: 0.9464\n",
      "Epoch 178/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2325 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0414 - categorical_accuracy: 0.9932 - val_loss: 0.2325 - val_categorical_accuracy: 0.9500\n",
      "Epoch 179/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2312 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0523 - categorical_accuracy: 0.9943 - val_loss: 0.2312 - val_categorical_accuracy: 0.9429\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 180/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.3113 - categorical_accuracy: 0.9393A: 1s - loss: 0.0598 - cate - ETA: 0s - loss: 0.0\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0595 - categorical_accuracy: 0.9882 - val_loss: 0.3113 - val_categorical_accuracy: 0.9393\n",
      "Epoch 181/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2379 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 12s 70ms/step - loss: 0.0502 - categorical_accuracy: 0.9929 - val_loss: 0.2379 - val_categorical_accuracy: 0.9500\n",
      "Epoch 182/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2406 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0731 - categorical_accuracy: 0.9889 - val_loss: 0.2406 - val_categorical_accuracy: 0.9536\n",
      "Epoch 183/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.3204 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 14s 81ms/step - loss: 0.0683 - categorical_accuracy: 0.9893 - val_loss: 0.3204 - val_categorical_accuracy: 0.9500\n",
      "Epoch 184/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.3316 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.0576 - categorical_accuracy: 0.9907 - val_loss: 0.3316 - val_categorical_accuracy: 0.9429\n",
      "Epoch 185/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2421 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0437 - categorical_accuracy: 0.9932 - val_loss: 0.2421 - val_categorical_accuracy: 0.9536\n",
      "Epoch 186/200\n",
      "18/18 [==============================] - 0s 27ms/step - loss: 0.2931 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 13s 72ms/step - loss: 0.0627 - categorical_accuracy: 0.9900 - val_loss: 0.2931 - val_categorical_accuracy: 0.9464\n",
      "Epoch 187/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2627 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 75ms/step - loss: 0.0607 - categorical_accuracy: 0.9889 - val_loss: 0.2627 - val_categorical_accuracy: 0.9536\n",
      "Epoch 188/200\n",
      "18/18 [==============================] - 0s 22ms/step - loss: 0.2386 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 14s 80ms/step - loss: 0.0547 - categorical_accuracy: 0.9900 - val_loss: 0.2386 - val_categorical_accuracy: 0.9429\n",
      "Epoch 189/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.5199 - categorical_accuracy: 0.9036 0s - loss: 0.6101 - categorical_accuracy\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0667 - categorical_accuracy: 0.9886 - val_loss: 0.5199 - val_categorical_accuracy: 0.9036\n",
      "Epoch 190/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2837 - categorical_accuracy: 0.9429\n",
      "175/175 [==============================] - 12s 71ms/step - loss: 0.0525 - categorical_accuracy: 0.9904 - val_loss: 0.2837 - val_categorical_accuracy: 0.9429\n",
      "Epoch 191/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2526 - categorical_accuracy: 0.9464\n",
      "175/175 [==============================] - 13s 73ms/step - loss: 0.0432 - categorical_accuracy: 0.9925 - val_loss: 0.2526 - val_categorical_accuracy: 0.9464\n",
      "Epoch 192/200\n",
      "18/18 [==============================] - 0s 23ms/step - loss: 0.2155 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 14s 79ms/step - loss: 0.0400 - categorical_accuracy: 0.9932 - val_loss: 0.2155 - val_categorical_accuracy: 0.9536\n",
      "Epoch 193/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.1980 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0482 - categorical_accuracy: 0.9918 - val_loss: 0.1980 - val_categorical_accuracy: 0.9500\n",
      "Epoch 194/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.1719 - categorical_accuracy: 0.9500\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0624 - categorical_accuracy: 0.9879 - val_loss: 0.1719 - val_categorical_accuracy: 0.9500\n",
      "Epoch 195/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2344 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0462 - categorical_accuracy: 0.9918 - val_loss: 0.2344 - val_categorical_accuracy: 0.9536\n",
      "Epoch 196/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.1743 - categorical_accuracy: 0.9607\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0616 - categorical_accuracy: 0.9907 - val_loss: 0.1743 - val_categorical_accuracy: 0.9607\n",
      "Epoch 197/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2365 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 13s 76ms/step - loss: 0.0507 - categorical_accuracy: 0.9921 - val_loss: 0.2365 - val_categorical_accuracy: 0.9536\n",
      "Epoch 198/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.3597 - categorical_accuracy: 0.9286\n",
      "175/175 [==============================] - 14s 82ms/step - loss: 0.0687 - categorical_accuracy: 0.9875 - val_loss: 0.3597 - val_categorical_accuracy: 0.9286\n",
      "Epoch 199/200\n",
      "18/18 [==============================] - 0s 25ms/step - loss: 0.2116 - categorical_accuracy: 0.9536\n",
      "175/175 [==============================] - 14s 80ms/step - loss: 0.0695 - categorical_accuracy: 0.9868 - val_loss: 0.2116 - val_categorical_accuracy: 0.9536\n",
      "Epoch 200/200\n",
      "18/18 [==============================] - 0s 24ms/step - loss: 0.2104 - categorical_accuracy: 0.9643\n",
      "175/175 [==============================] - 13s 74ms/step - loss: 0.0554 - categorical_accuracy: 0.9911 - val_loss: 0.2104 - val_categorical_accuracy: 0.9643\n"
     ]
    }
   ],
   "source": [
    "import math\n",
    "\n",
    "history = model.fit_generator(\n",
    "      train_generator,\n",
    "      steps_per_epoch=math.ceil(number_of_training_samples//batch_size),\n",
    "      epochs=200,\n",
    "      callbacks=callbacks_list,\n",
    "      validation_data=validation_generator,\n",
    "      validation_steps=math.ceil(number_of_validation_samples//batch_size))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "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": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "\n",
    "acc=history.history['categorical_accuracy']\n",
    "val_acc=history.history['val_categorical_accuracy']\n",
    "loss=history.history['loss']\n",
    "val_loss=history.history['val_loss']\n",
    "epochs=range(1,len(acc)+1)\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Accuracy')\n",
    "plt.plot(epochs, acc, 'bo', label='Training acc')\n",
    "plt.plot(epochs, val_acc, 'b', label='Validation acc')\n",
    "plt.title('Training and validation accuracy')\n",
    "plt.legend()\n",
    "plt.figure()\n",
    "\n",
    "plt.plot(epochs, loss, 'bo', label='Training loss')\n",
    "plt.plot(epochs, val_loss, 'b', label='Validation loss')\n",
    "plt.title('Training and validation loss')\n",
    "plt.xlabel('Epochs')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.python.keras.models import load_model\n",
    "\n",
    "best_model = load_model('MobileNet_Indus_Fruits_7_classes_100_100_1st Experiment.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "35/35 [==============================] - 1s 19ms/step - loss: 0.2459 - categorical_accuracy: 0.9446\n",
      "Test-set accuracy: 94.46%\n"
     ]
    }
   ],
   "source": [
    "from tensorflow.python.keras.models import load_model\n",
    "\n",
    "steps_test=int(number_of_test_samples/batch_size)\n",
    "result = best_model.evaluate_generator(test_generator, steps=steps_test,verbose=1)\n",
    "print(\"Test-set accuracy: {0:.2%}\".format(result[1]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total number of test examples:  (560, 100, 100, 3)\n",
      "Total number of labels: (560, 1)\n"
     ]
    }
   ],
   "source": [
    "# Preparing test data\n",
    "import glob\n",
    "from pathlib import Path\n",
    "from tensorflow.python.keras.utils import to_categorical\n",
    "import cv2\n",
    "\n",
    "apple_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\apple')\n",
    "banana_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\banana')\n",
    "orange_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\orange')\n",
    "watermelon_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\watermelon')\n",
    "mango_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\mango')\n",
    "grape_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\grape')\n",
    "strawberry_cases_dir = Path('D:\\\\TensorFlow Programs\\\\fruit_image_classification_Indus Fruit Dataset\\\\Indus Fruit Dataset_Resized Images_100_100\\\\Test\\\\strawberry')\n",
    "\n",
    "apple_cases = apple_cases_dir.glob('*.jpg')\n",
    "banana_cases = banana_cases_dir.glob('*.jpg')\n",
    "orange_cases = orange_cases_dir.glob('*.jpg')\n",
    "watermelon_cases = watermelon_cases_dir.glob('*.jpg')\n",
    "mango_cases = mango_cases_dir.glob('*.jpg')\n",
    "grape_cases = grape_cases_dir.glob('*.jpg')\n",
    "strawberry_cases = strawberry_cases_dir.glob('*.jpg')\n",
    "\n",
    "test_data = []\n",
    "test_labels = []\n",
    "\n",
    "for img in apple_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [0]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "for img in banana_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [1]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "for img in grape_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [2]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "    \n",
    "for img in mango_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [3]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "for img in orange_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [4]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "for img in strawberry_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [5]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "for img in watermelon_cases:\n",
    "    img = cv2.imread(str(img))\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)\n",
    "    img = img.astype(np.float32)/255.\n",
    "    label = [6]\n",
    "    test_data.append(img)\n",
    "    test_labels.append(label)\n",
    "\n",
    "test_data = np.array(test_data)\n",
    "test_labels = np.array(test_labels)\n",
    "test_labels_one_hot_encoding=to_categorical(test_labels)\n",
    "\n",
    "print(\"Total number of test examples: \", test_data.shape)\n",
    "print(\"Total number of labels:\", test_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "560/560 [==============================] - 1s 1ms/sample - loss: 0.2459 - categorical_accuracy: 0.9446\n",
      "Loss on test set:  0.24591173777090652\n",
      "Accuracy on test set:  0.94464284\n"
     ]
    }
   ],
   "source": [
    "# Evaluation on test dataset\n",
    "\n",
    "test_loss, test_score = best_model.evaluate(test_data, test_labels_one_hot_encoding, batch_size=8)\n",
    "print(\"Loss on test set: \", test_loss)\n",
    "print(\"Accuracy on test set: \", test_score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "560/560 [==============================] - 1s 2ms/sample\n",
      "(560, 1)\n",
      "(560, 1)\n"
     ]
    }
   ],
   "source": [
    "# Get predictions\n",
    "preds = best_model.predict(test_data, batch_size=16,verbose=1)\n",
    "preds=np.around(preds)\n",
    "orig_test_labels=test_labels\n",
    "predicted_labels=np.argmax(preds,axis=1)\n",
    "predicted_labels=np.reshape(predicted_labels,(560,1))\n",
    "print(predicted_labels.shape)\n",
    "print(orig_test_labels.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 3600x2400 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 864x720 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "# Get the confusion matrix\n",
    "from sklearn.metrics import confusion_matrix\n",
    "from mlxtend.plotting import plot_confusion_matrix\n",
    "\n",
    "cm  = confusion_matrix(orig_test_labels, predicted_labels)\n",
    "plt.figure(dpi=600)\n",
    "plot_confusion_matrix(cm,figsize=(12,10), hide_ticks=True,cmap=plt.cm.Blues)\n",
    "plt.xticks(range(7), ['apple', 'banana','grape','mango','orange','strawberry','watermelon'], fontsize=12)\n",
    "plt.yticks(range(7), ['apple', 'banana','grape','mango','orange','strawberry','watermelon'], fontsize=12)\n",
    "plt.xlabel('Predicted Label',fontsize=14)\n",
    "plt.ylabel('True Label',fontsize=14)\n",
    "plt.savefig('MobileNet_Indus Fruits_7 classes_100_100_confusion_matrix.jpeg',dpi=1200)\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}