train_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from tensorflow.keras import preprocessing\n",
    "from tensorflow.keras.applications import mobilenet_v2\n",
    "from tensorflow import keras\n",
    "import csv\n",
    "import os"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = []\n",
    "\n",
    "for entry in os.scandir('data'):\n",
    "    labels.append(entry.name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "# this are the folder names\n",
    "classes = labels\n",
    "# plug in the path to your data folder\n",
    "base_path = 'data/'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define an image data generator\n",
    "data_gen = preprocessing.image.ImageDataGenerator(\n",
    "    # define the preprocessing function that should be applied to all images\n",
    "    preprocessing_function=mobilenet_v2.preprocess_input,\n",
    "    # fill_mode='nearest',\n",
    "    # rotation_range=20,\n",
    "    # width_shift_range=0.2,\n",
    "    # height_shift_range=0.2,\n",
    "    # horizontal_flip=True, \n",
    "    # zoom_range=0.2,\n",
    "    # shear_range=0.2    \n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Found 443 images belonging to 40 classes.\n"
     ]
    }
   ],
   "source": [
    "# a generator that returns batches of X and y arrays\n",
    "train_data_gen = data_gen.flow_from_directory(\n",
    "        directory=base_path,\n",
    "        class_mode=\"categorical\",\n",
    "        classes=classes,\n",
    "        batch_size=443,\n",
    "        target_size=(224, 224)\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "labels = train_data_gen.class_indices\n",
    "labels = dict((v,k) for k,v in labels.items())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "  \n",
    "with open('label_mapping.csv', 'w') as f:\n",
    "    for key in labels.keys():\n",
    "        f.write(\"%s, %s\\n\" % (key, labels[key]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "((443, 224, 224, 3), (443, 40))"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# load in all images at once\n",
    "xtrain, ytrain = next(train_data_gen)\n",
    "xtrain.shape, ytrain.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Select the convolutional base and freeze the weights"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [],
   "source": [
    "base_model = mobilenet_v2.MobileNetV2(\n",
    "    weights='imagenet', \n",
    "    alpha=0.35,         # specific parameter of this model, small alpha reduces the number of overall weights\n",
    "    pooling='avg',      # applies global average pooling to the output of the last conv layer (like a flattening)\n",
    "    include_top=False,  # we only want to have the base, not the final dense layers \n",
    "    input_shape=(224, 224, 3)\n",
    ")\n",
    "\n",
    "# freeze it!\n",
    "base_model.trainable = False"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Add your own dense layers on top"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 190,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential_6\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "mobilenetv2_0.35_224 (Functi (None, 1280)              410208    \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 100)               128100    \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 100)               0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 40)                4040      \n",
      "=================================================================\n",
      "Total params: 542,348\n",
      "Trainable params: 132,140\n",
      "Non-trainable params: 410,208\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = keras.Sequential()\n",
    "model.add(base_model)\n",
    "model.add(keras.layers.Dense(100, activation='relu'))\n",
    "model.add(keras.layers.Dropout(0.5))\n",
    "model.add(keras.layers.Dense(len(classes), activation='softmax'))\n",
    "# have a look at the trainable and non-trainable params statistic\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Compile and train!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 185,
   "metadata": {},
   "outputs": [],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "import numpy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 191,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/50\n",
      "3/3 - 5s - loss: 4.0330 - categorical_accuracy: 0.0387 - val_loss: 3.4307 - val_categorical_accuracy: 0.1278\n",
      "Epoch 2/50\n",
      "3/3 - 1s - loss: 3.3126 - categorical_accuracy: 0.1419 - val_loss: 3.1706 - val_categorical_accuracy: 0.2406\n",
      "Epoch 3/50\n",
      "3/3 - 1s - loss: 2.9681 - categorical_accuracy: 0.2355 - val_loss: 2.8751 - val_categorical_accuracy: 0.3459\n",
      "Epoch 4/50\n",
      "3/3 - 1s - loss: 2.6681 - categorical_accuracy: 0.3677 - val_loss: 2.5640 - val_categorical_accuracy: 0.4962\n",
      "Epoch 5/50\n",
      "3/3 - 1s - loss: 2.4417 - categorical_accuracy: 0.4323 - val_loss: 2.2964 - val_categorical_accuracy: 0.5263\n",
      "Epoch 6/50\n",
      "3/3 - 1s - loss: 2.1201 - categorical_accuracy: 0.5161 - val_loss: 2.0696 - val_categorical_accuracy: 0.6015\n",
      "Epoch 7/50\n",
      "3/3 - 1s - loss: 1.9905 - categorical_accuracy: 0.5000 - val_loss: 1.8442 - val_categorical_accuracy: 0.6391\n",
      "Epoch 8/50\n",
      "3/3 - 1s - loss: 1.6861 - categorical_accuracy: 0.6032 - val_loss: 1.6270 - val_categorical_accuracy: 0.7143\n",
      "Epoch 9/50\n",
      "3/3 - 1s - loss: 1.4696 - categorical_accuracy: 0.6645 - val_loss: 1.4306 - val_categorical_accuracy: 0.7519\n",
      "Epoch 10/50\n",
      "3/3 - 1s - loss: 1.3129 - categorical_accuracy: 0.6806 - val_loss: 1.2623 - val_categorical_accuracy: 0.8195\n",
      "Epoch 11/50\n",
      "3/3 - 1s - loss: 1.2549 - categorical_accuracy: 0.7097 - val_loss: 1.1268 - val_categorical_accuracy: 0.7970\n",
      "Epoch 12/50\n",
      "3/3 - 1s - loss: 1.1125 - categorical_accuracy: 0.7452 - val_loss: 1.0086 - val_categorical_accuracy: 0.8346\n",
      "Epoch 13/50\n",
      "3/3 - 1s - loss: 1.0057 - categorical_accuracy: 0.7452 - val_loss: 0.9095 - val_categorical_accuracy: 0.8346\n",
      "Epoch 14/50\n",
      "3/3 - 1s - loss: 0.9015 - categorical_accuracy: 0.7903 - val_loss: 0.8321 - val_categorical_accuracy: 0.8797\n",
      "Epoch 15/50\n",
      "3/3 - 1s - loss: 0.7795 - categorical_accuracy: 0.8129 - val_loss: 0.7686 - val_categorical_accuracy: 0.8947\n",
      "Epoch 16/50\n",
      "3/3 - 1s - loss: 0.6843 - categorical_accuracy: 0.8742 - val_loss: 0.7034 - val_categorical_accuracy: 0.8947\n",
      "Epoch 17/50\n",
      "3/3 - 1s - loss: 0.6835 - categorical_accuracy: 0.8677 - val_loss: 0.6502 - val_categorical_accuracy: 0.9098\n",
      "Epoch 18/50\n",
      "3/3 - 1s - loss: 0.5864 - categorical_accuracy: 0.8871 - val_loss: 0.6010 - val_categorical_accuracy: 0.9398\n",
      "Epoch 19/50\n",
      "3/3 - 1s - loss: 0.5978 - categorical_accuracy: 0.8710 - val_loss: 0.5540 - val_categorical_accuracy: 0.9398\n",
      "Epoch 20/50\n",
      "3/3 - 1s - loss: 0.5178 - categorical_accuracy: 0.9032 - val_loss: 0.5193 - val_categorical_accuracy: 0.9398\n",
      "Epoch 21/50\n",
      "3/3 - 1s - loss: 0.5736 - categorical_accuracy: 0.8774 - val_loss: 0.4956 - val_categorical_accuracy: 0.9474\n",
      "Epoch 22/50\n",
      "3/3 - 1s - loss: 0.4564 - categorical_accuracy: 0.9194 - val_loss: 0.4749 - val_categorical_accuracy: 0.9549\n",
      "Epoch 23/50\n",
      "3/3 - 1s - loss: 0.4238 - categorical_accuracy: 0.9516 - val_loss: 0.4557 - val_categorical_accuracy: 0.9549\n",
      "Epoch 24/50\n",
      "3/3 - 1s - loss: 0.4154 - categorical_accuracy: 0.9161 - val_loss: 0.4291 - val_categorical_accuracy: 0.9549\n",
      "Epoch 25/50\n",
      "3/3 - 1s - loss: 0.4014 - categorical_accuracy: 0.9355 - val_loss: 0.4058 - val_categorical_accuracy: 0.9624\n",
      "Epoch 26/50\n",
      "3/3 - 1s - loss: 0.4121 - categorical_accuracy: 0.9097 - val_loss: 0.3879 - val_categorical_accuracy: 0.9549\n",
      "Epoch 27/50\n",
      "3/3 - 1s - loss: 0.3731 - categorical_accuracy: 0.9452 - val_loss: 0.3737 - val_categorical_accuracy: 0.9549\n",
      "Epoch 28/50\n",
      "3/3 - 1s - loss: 0.3861 - categorical_accuracy: 0.9161 - val_loss: 0.3626 - val_categorical_accuracy: 0.9699\n",
      "Epoch 29/50\n",
      "3/3 - 1s - loss: 0.3074 - categorical_accuracy: 0.9387 - val_loss: 0.3562 - val_categorical_accuracy: 0.9624\n",
      "Epoch 30/50\n",
      "3/3 - 1s - loss: 0.3025 - categorical_accuracy: 0.9323 - val_loss: 0.3452 - val_categorical_accuracy: 0.9549\n",
      "Epoch 31/50\n",
      "3/3 - 1s - loss: 0.3043 - categorical_accuracy: 0.9323 - val_loss: 0.3319 - val_categorical_accuracy: 0.9474\n",
      "Epoch 32/50\n",
      "3/3 - 1s - loss: 0.2437 - categorical_accuracy: 0.9581 - val_loss: 0.3205 - val_categorical_accuracy: 0.9474\n",
      "Epoch 33/50\n",
      "3/3 - 1s - loss: 0.2471 - categorical_accuracy: 0.9613 - val_loss: 0.3090 - val_categorical_accuracy: 0.9549\n",
      "Epoch 34/50\n",
      "3/3 - 1s - loss: 0.2857 - categorical_accuracy: 0.9581 - val_loss: 0.2956 - val_categorical_accuracy: 0.9624\n",
      "Epoch 35/50\n",
      "3/3 - 1s - loss: 0.2121 - categorical_accuracy: 0.9581 - val_loss: 0.2833 - val_categorical_accuracy: 0.9624\n",
      "Epoch 36/50\n",
      "3/3 - 1s - loss: 0.2405 - categorical_accuracy: 0.9645 - val_loss: 0.2799 - val_categorical_accuracy: 0.9699\n",
      "Epoch 37/50\n",
      "3/3 - 1s - loss: 0.2369 - categorical_accuracy: 0.9581 - val_loss: 0.2798 - val_categorical_accuracy: 0.9699\n",
      "Epoch 38/50\n",
      "3/3 - 1s - loss: 0.2542 - categorical_accuracy: 0.9645 - val_loss: 0.2803 - val_categorical_accuracy: 0.9699\n",
      "Epoch 39/50\n",
      "3/3 - 1s - loss: 0.1960 - categorical_accuracy: 0.9710 - val_loss: 0.2784 - val_categorical_accuracy: 0.9624\n",
      "Epoch 40/50\n",
      "3/3 - 1s - loss: 0.2029 - categorical_accuracy: 0.9710 - val_loss: 0.2747 - val_categorical_accuracy: 0.9624\n",
      "Epoch 41/50\n",
      "3/3 - 1s - loss: 0.1992 - categorical_accuracy: 0.9677 - val_loss: 0.2709 - val_categorical_accuracy: 0.9624\n",
      "Epoch 42/50\n",
      "3/3 - 1s - loss: 0.2102 - categorical_accuracy: 0.9677 - val_loss: 0.2653 - val_categorical_accuracy: 0.9549\n",
      "Epoch 43/50\n",
      "3/3 - 1s - loss: 0.1806 - categorical_accuracy: 0.9806 - val_loss: 0.2584 - val_categorical_accuracy: 0.9549\n",
      "Epoch 44/50\n",
      "3/3 - 1s - loss: 0.1743 - categorical_accuracy: 0.9871 - val_loss: 0.2515 - val_categorical_accuracy: 0.9624\n",
      "Epoch 45/50\n",
      "3/3 - 1s - loss: 0.1845 - categorical_accuracy: 0.9710 - val_loss: 0.2436 - val_categorical_accuracy: 0.9624\n",
      "Epoch 46/50\n",
      "3/3 - 1s - loss: 0.1851 - categorical_accuracy: 0.9548 - val_loss: 0.2320 - val_categorical_accuracy: 0.9699\n",
      "Epoch 47/50\n",
      "3/3 - 1s - loss: 0.1388 - categorical_accuracy: 0.9806 - val_loss: 0.2248 - val_categorical_accuracy: 0.9699\n",
      "Epoch 48/50\n",
      "3/3 - 1s - loss: 0.1818 - categorical_accuracy: 0.9613 - val_loss: 0.2235 - val_categorical_accuracy: 0.9699\n",
      "Epoch 49/50\n",
      "3/3 - 1s - loss: 0.1541 - categorical_accuracy: 0.9774 - val_loss: 0.2239 - val_categorical_accuracy: 0.9624\n",
      "Epoch 50/50\n",
      "3/3 - 1s - loss: 0.1270 - categorical_accuracy: 0.9903 - val_loss: 0.2250 - val_categorical_accuracy: 0.9624\n"
     ]
    }
   ],
   "source": [
    "model.compile(optimizer=keras.optimizers.Adam(learning_rate=0.001),\n",
    "              loss=keras.losses.categorical_crossentropy,\n",
    "              metrics=[keras.metrics.categorical_accuracy])\n",
    "\n",
    "# observe the validation loss and stop when it does not improve after 3 iterations\n",
    "callback = keras.callbacks.EarlyStopping(monitor='val_loss', patience=3)\n",
    "mcp_save = keras.callbacks.ModelCheckpoint('mdl_wts.hdf5', save_best_only=True, monitor='val_categorical_accuracy')\n",
    "\n",
    "\n",
    "history = model.fit(xtrain, ytrain, \n",
    "          epochs=50, \n",
    "          verbose=2,\n",
    "          #batch_size=len(xtrain),\n",
    "          batch_size=110,        \n",
    "          callbacks=[callback, mcp_save],\n",
    "          # use 30% of the data for validation\n",
    "          validation_split=0.3)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYIAAAEWCAYAAABrDZDcAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjMuMiwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy8vihELAAAACXBIWXMAAAsTAAALEwEAmpwYAAA6HklEQVR4nO3dd3xV9fnA8c+TPclmhiSMsGUGEEUEJ4jgaOvEVmuLVu2uVdva2tr+fvbX2tpWK1rqXljrQEVFUIbKnoaVBMhmZO99v78/zgUybuAGc3PvzX3er1deyT3fc+59TiDnOec7xRiDUkop3+Xn7gCUUkq5lyYCpZTycZoIlFLKx2kiUEopH6eJQCmlfJwmAqWU8nGaCJRPEZHnROT3Tu6bLSKXuDompdxNE4FSSvk4TQRKeSERCXB3DKr30ESgPI69SuZeEdktIjUi8m8R6SciH4hIlYisEpGYVvsvFJE9IlIuImtEZHSrskkist1+3DIgpN1nXSkiO+3HfiEi452Mcb6I7BCRShHJE5GH2pXPtL9fub38Vvv2UBF5VERyRKRCRD6zb5stIvkOfg+X2H9+SETeEJGXRKQSuFVEponIBvtnHBGRx0UkqNXxY0XkYxEpFZFjIvILEekvIrUiEtdqvykiUiQigc6cu+p9NBEoT/U14FJgBLAA+AD4BRCP9f/2BwAiMgJ4FfgRkACsAN4VkSD7RfFt4EUgFviP/X2xHzsZeAa4A4gDngKWi0iwE/HVAN8EooH5wPdE5Gr7+ybZ4/2HPaaJwE77cX8GpgDn2WP6OWBz8ndyFfCG/TNfBlqAH2P9TmYAFwN32WOIBFYBHwIDgeHAamPMUWANcF2r910EvGaMaXIyDtXLaCJQnuofxphjxpgCYD2wyRizwxjTALwFTLLvdz3wvjHmY/uF7M9AKNaF9lwgEHjMGNNkjHkD2NLqM74LPGWM2WSMaTHGPA802I87LWPMGmPMl8YYmzFmN1YyutBefDOwyhjzqv1zS4wxO0XED/g28ENjTIH9M7+wn5MzNhhj3rZ/Zp0xZpsxZqMxptkYk42VyE7EcCVw1BjzqDGm3hhTZYzZZC97Huvij4j4AzdiJUvlozQRKE91rNXPdQ5eR9h/HgjknCgwxtiAPGCQvazAtJ1ZMafVz8nAT+1VK+UiUg4Mth93WiIyXUQ+tVepVAB3Yt2ZY3+Pgw4Oi8eqmnJU5oy8djGMEJH3ROSovbrof5yIAeAdYIyIDMV66qowxmw+y5hUL6CJQHm7QqwLOgAiIlgXwQLgCDDIvu2EpFY/5wF/MMZEt/oKM8a86sTnvgIsBwYbY6KAJcCJz8kDhjk4phio76SsBghrdR7+WNVKrbWfKvhJYD+Qaozpg1V1dqYYMMbUA69jPbncgj4N+DxNBMrbvQ7MF5GL7Y2dP8Wq3vkC2AA0Az8QkQARuRaY1urYfwF32u/uRUTC7Y3AkU58biRQaoypF5FpwE2tyl4GLhGR6+yfGyciE+1PK88AfxGRgSLiLyIz7G0SGUCI/fMDgV8BZ2qriAQqgWoRGQV8r1XZe0B/EfmRiASLSKSITG9V/gJwK7AQeMmJ81W9mCYC5dWMMQew6rv/gXXHvQBYYIxpNMY0AtdiXfDKsNoT3mx17FasdoLH7eVZ9n2dcRfwOxGpAn6NlZBOvG8ucAVWUirFaiieYC/+GfAlVltFKfBHwM8YU2F/z6VYTzM1QJteRA78DCsBVWEltWWtYqjCqvZZABwFMoE5rco/x2qk3m5vX1A+THRhGqV8k4h8ArxijFnq7liUe2kiUMoHichU4GOsNo4qd8ej3EurhpTyMSLyPNYYgx9pElCgTwRKKeXz9IlAKaV8nNdNXBUfH29SUlLcHYZSSnmVbdu2FRtj2o9NAbwwEaSkpLB161Z3h6GUUl5FRHI6K9OqIaWU8nGaCJRSysdpIlBKKR/ndW0EjjQ1NZGfn099fb27Q3G5kJAQEhMTCQzUNUSUUt3DZYlARJ7BmhP9uDFmnINyAf6GNSdLLXCrMWb72XxWfn4+kZGRpKSk0Haiyd7FGENJSQn5+fkMGTLE3eEopXoJV1YNPQfMPU35PCDV/rUYa0rds1JfX09cXFyvTgIAIkJcXJxPPPkopXqOyxKBMWYd1uyKnbkKeMFYNgLRIjLgbD+vtyeBE3zlPJVSPcedjcWDaLviUr59m1Kql6hrbOHFjTkcrdCnWE/mzkTg6NbW4cRHIrJYRLaKyNaioiIXh9V15eXl/POf/+zycVdccQXl5eXdH5Dq1ZpabFTUev4687vyypn/j/U8+HY6X1/yBXmlte4OqVPGGLZml/Kj13bw3Re2sju//Cu/Z4vNkF1cQ21j82k/91hlPRV17v33dGevoXysJQVPSMRadrADY8zTwNMAaWlpHjdL3olEcNddd7XZ3tLSgr+/f6fHrVixwtWhqV7GGMP3XtrGZ1nFPDBvNLecm4yfn2dVFza12Hj8kywe/zSLhIhgHr5qLH9emcE3lmzgle9OZ2hCxJnfpIdUNzTz9o4CXtqYw/6jVUQGBxAY4MfCxz9n4YSB3Hv5SAbHhp35jewKyutYn1HEuswiPsssprLeSgIJkcEkx4aRFBdGdGgQeWW15JbUkltaS11TC+FB/vx6wRiuSxvslupfdyaC5cA9IvIaMB1rAe0jboznrN1///0cPHiQiRMnEhgYSEREBAMGDGDnzp3s3buXq6++mry8POrr6/nhD3/I4sWLgVPTZVRXVzNv3jxmzpzJF198waBBg3jnnXcIDQ1185kpT/PcF9ms2nec4X0j+M3yPXy89xh/+sZ4BkSd/v9KRV0Tb27P58P0o5w3LJ7vzhpCWFD3/vk3NtvYf7SSX72dzu78Cq6ZNIiHFo4lKjSQtJRYFi3dxHVPbeTl70xnZH9nVgNtq8VmKCyvI7e0lpySWnJKa8gpriW/vJaBUaHMGpHArNQEkuI6Xrgr65vILTl1XG5JLdklNaQXVFLd0MyYAX145NpzWDhxIC02w5K1B1m6/jAfph/lW+clM2tEArmltSffI7e0lvrmljaf0dBko6C8DoD+fUK4fGx/JifHUFrTSE5JDTkltXyRVUJFXROJMaEkx4UxMzWepNgwPkw/yn3//ZKP9x7jf68dT0LkmVYp7V4um4ZaRF4FZgPxwDHgN0AggDFmib376ONYPYtqgdvsSweeVlpammk/19C+ffsYPXo0AL99dw97Cyu770SAMQP78JsFYzstz87O5sorryQ9PZ01a9Ywf/580tPTT3bxLC0tJTY2lrq6OqZOncratWuJi4trkwiGDx/O1q1bmThxItdddx0LFy5k0aJFDj+v9fkqF6kphm3PQVO76oygcBh+KfQ/B3r4zm1vYSVXP/E5F6TG869vpvHqllx+/94+AvyFh68ax1UTB566m2xphvQ3yKtsYunRVF7fXU5dUwvJcWHklNTSNzKYn14yjK/3LcS/YAsMng6Dp4Ff50+wALWNzdbFtKTWurjZL47ZJTUUltdhMxATFsj/XHMO885p2/cj63gVN/1rE00tNl68fTrjBkV1+jnHq+r5LLOYXXnlJz8jr6yWppZT16sgfz8SY0NJjAnj4PHqkxfhlLgwZgyLo7ax5eRFu7Smsc37x0cEkRQbxsj+kXwjbTCTBkcjDZWQsRLqy2HE5RyRBB5dmcF/t+dz4jIZ5O/H9Ohy5gbtpi6kL3vDp9HoZyVhfz/hnEFRzBqRQGrfiC7d2dtshme/yOaPH+4nIjiA/7lmHHPHDaC8tpEc++83t6SWCYOjmTXC4bxxZyQi24wxaY7KXPZEYIy58QzlBrjbVZ/vTtOmTWvTz//vf/87b731FgB5eXlkZmYSFxfX5pghQ4YwceJEAKZMmUJ2dnZPhataMwZ2L4MPH4C6UvBr9ydia4bVv4PoZBi9gOZRC6iOn0h0eIhLw6ptbOb7r24nOiyQP31jAn5+ws3Tkzl/WDw//c8ufrRsJ49+fIBAPz+GthzmZw2PM8qWxWDglyaA66On0WfStSROXUjG7g1kr3+RSSs+x19a3TRF9INR82H0Aki5APwDMcawPbeMlzbm8llWMUVVDW3iigkLJCkunCnJMVw7aRBJceHMHplAfETHO9rhfSN5/Y4Z3Lx0E9c/tYEpKbEkx4aRHBdGclw4QQF+fJFVzNqMIvYftdbLiQgOIDkujFEDIrlsbH/7vtb+/fuE4G+vFjPGcKi4hvUZRazPLOa93UfoExJIclwYl584zl41kxwXTkSw/d+1phj2L4d178KhNWCz19Wv+BkDBkzkz6MX8IPbLqG0spphRZ8QcfgD5Phea0VpgIAQGHax9TsbORdCY87q39fPT7h95hBmpcbzk9d3cedL24kMDqCqoW37wvdmDzvrRHA6vWJkcWunu3PvKeHh4Sd/XrNmDatWrWLDhg2EhYUxe/Zsh+MAgoNP/eH4+/tTV1fXI7GqVspy4L0fw8HV1h3ygr9D31Ft96kphgMrYN+7mE1PEbDhcRpNDEdHzaf/9G9A8kzw7/4/q9+9u5dDxTW8fPt0YsODTm5PiQ/n9Ttm8PwX2XyZc4y5JS9ySemr1Pj34dkBvyZ+0BAuYxNjMlfAup/Bup8xAkgNiuDI4Fk8VDSGFRVDmB2ayfXsZPyOVwnc+gwmJJrDcbN4sXw8r5QMJyg4jEvH9GNY3wj7RTWcpLgwokK7NsI9JT6cZXecy6MrM8g6Xs2O3DKq6k9d7AL9hbTkWH4+dySzUhMYM6CPU20gIsKwKGFYzJfcGvkuhK6ynuaOYX11prkBMBCTAufeCaOvgrBY2P8e7HsXPnmYJB4myfoUSD4P5j4CI+dBRb61z7534cD71k1DygUwZiGMnA+R/U59Tut9C7bTSb8YUrHqzFvCDMYYJEwQrIdPEUEC7wEePOPvo6t6XSJwh8jISKqqHK/4V1FRQUxMDGFhYezfv5+NGzf2cHQ+wmaD/M1QdRSGzobQaIe7NTbbKCyvo2+f4FN15NVFsOsVWPMIiB8tc/+PZVzO6/8p4PaZhSyYMPDUG4THw+RvUjbyBu5c+imDitaxIHgb0/f/Bw68ZN0RjpwPAyd2/PC+YyDpXIfVL80tNpZtzeM/W/MZ2S+SWSMSmDk8nqiwQN7bXchrW/K4e84wzhsaAzlfwLE9J4/1B74d0AylS6E0CyYuos9lD3NbWKx9j2vBPAJHdsGhTyFhFDJ0DgMDQ/hFs43Je46yLuMcvpdxARU1tzLT70vm27ZwUd1qfiPL+UV4KJJ6GQHJ51mx12N163DYtcMB8bOqnfqNAxESY8L46/XW78cYQ3lNI8cyNhNwZAeDY4IJDiiwjisAyhNg+MUQ3EmbQl05ZK6EfcshcxU011n/BiPmQkTfM8cWHGnta4/tpPN/aH1VFlqJ3y/Quvi3fs+YFEiZaSWGwu2wd7l1oX/vx/DeT6x/68HT4PB6qxys/wNTboWAIDojnObCnHTumc/pLGgi6AZxcXGcf/75jBs3jtDQUPr1O3UnMHfuXJYsWcL48eMZOXIk557rmn9In9TSBNnrrT++/e9Dtf3Wzy8Qhl6IGbWAT0waq/PNyca6E/XYw4PL+engDGbbNhJ6ZDMYGyb1Mj4f+QseWldJ1vE9xIQF8v1Xd/Dx3mP87qqxRIdZf7zF1Q0sWrqJQ8U27lj0A85JjOampesYWLyBB/tmMWDfu7DzJccxh8Xbq18WwpBZGP9AVu87ziMf7ifreDUj+0WyIv0Iy7bm4ScwPjGanKJybut3kJ80fASProCaTrpQRyfDLW/DsDkdy0Ss5NQuQQUF+LFwwkAWThiIMYaMY9Wsz5zA7rKvkXxOApNtewjc9651h7z/nbP6ZzopJsWqQhl9FQycBAVbkX3vErNvOTHluZ0f5x9sndPoBTDyCrC1WHfg+96FQ2ut6pzIATD5FmufpPO676msz0CY+p3T7yMCg6ZYX5c8BMf3nbr7//xvp7aPWgDxw7snrm7mdWsWn6mx2Bf41PkaA+n/hfV/gaaatmW1ZdBQAYHhkHqpdRGISoT971O/+21CqnNpMcIRScDfz49Afz8C/IRAP0N4nXU7u982mPQ+swifeA3PHwxn4+EyhsSHc9/ckVw8uh9L1hzkb6sziYsI4v++PoGR/SK5eelGCsrrWPrNqcxMjQegvLaRbz2zmT2Flfzj+nHMGxZKk83Gl/kVbDxUwtbDxYxu2sPFbGJc9UaCbLU0B4RTbPrQ0NRCoL8f0WGBhAZZTwuNzTbqmlqob2whwlZJBLXWeY64zEoiSTM6tl+ExrikWgqwnrhqS87u2OY6OPhJ2wu3fxC0NFrfh9ov8kNnW3XurZVknrqoVuSB+AMGjM2eWBZaX4OmgJ8HTqbcVA+Brm0/ctbpGos1EXghrzjf2lIoOWjd+Z3txaki33rEzvzI6qXTd0zb8sAwKwEMuwgCrZ4bB4uqeeSD/Xy89ygzI47ywJBMRoeU4te+B0fCKMqS5/LaoWBe3pRDflkdseFB/PDiVG6ankSg/6mLSnpBBT9etpPM49XEhAXS2Gzj2dumMW1IbJu3rKpv4rZnt7A9t4yZqQlszymjuqEZfz9hQmIUwQH+5JbWUlxRwfmSzhy/ncQGNjK6fyQp8eEdYzwhKBxSL7MumB5yUTlrdeWQ8RHkb7GqOVIvg5A+Zz7OGDiy03ryE38YfWXH6hx1WpoIehmPPl9jYPfr8OH9Vq+bE3XmJ+74nLmQ2Wyw9d+w6iHrzu+iX8H0O0/btbG+qYW/fpzB0s8OExroz/dmD+Pb5w85eYd9Oi02w5cFFQxNCKdPiOPGz/qmFh5deYAP0o/y9xsnMTnJce+Q2sZmfrxsJ/uPVnH+8HhmpcYzY1h8m0bV+qYW8svqOFpRz4TBUUR28plKdSdNBL2Mx55veS68+yOr103iNEj7ttU4eeBDqwonKMK6Axyz0OqLH9xuhGnlEasueucrVuPa0Dmw4DGrCuA09hRW8JNluzhwrIrr0wZz79yRDrsvKuXL3DKOQPkQmw02PwWrH7Ye1ef9Cabebt3BT7wRmhvh8DrY9w7sXwF73rQaAIdfDKOutJ4c9i63ev0AxKXC1U/ChBtP++jf3GLjqXWHeGxVBtFhQTx721TmjHSip4hSqg1NBOqrW/Ub+OLv1l3+lX+F6MFtywOCIPUS6+vKxyB3o9Xdb9+7Vtc8gP7jrSqg0QshYeRpP66irokNB4t5et0htueWM/+cAfz+6nHEhHfeJU8p1TlNBOqr2fuOlQTSbof5j5658c7PH1LOt77mPgJHv7QaC09T/dPcYmN3QQXr7KNGd+aV02IzRIcF8rcbJrJwwkBdp0Gpr0ATQTcoLy/nlVde6TD7qDMee+wxFi9eTFiY8zMceoyiDHj7Lkical3Uu3oxFoEB4x0WFZTX2S/8p2ZxFHu/+rtmD+OC1AQmJUW36d2jlDo7mgi6QWfTUDvjscceY9GiRd6XCBqqYdkiq9/3N54/7UjJE2obm9mSXUaufXBXTmkteaW11LSbr72p2XC00pqGo3+fEOaO688FqdZIW63+Uar7aSLoBq2nob700kvp27cvr7/+Og0NDVxzzTX89re/paamhuuuu478/HxaWlp48MEHOXbsGIWFhcyZM4f4+Hg+/fRTd5+Kc4yB5d+3Bvvc8jZEnXlhuaYWG9c9tYH0AmuSs+AAP5LjwkiKDadPSLv/hgJjBvThwhEJDO/iLI5Kqa7rfYngg/uteufu1P8cmPdIp8WPPPII6enp7Ny5k5UrV/LGG2+wefNmjDEsXLiQdevWUVRUxMCBA3n//fcBaw6iqKgo/vKXv/Dpp58SHx/fvTF3F0fdizc+afX8ufg3MPRCp97myTUHSS+o5A/XjOPiUf3oGxnscQuqKOWrel8icLOVK1eycuVKJk2aBEB1dTWZmZlccMEF/OxnP+O+++7jyiuv5IILLnBzpE4oy4HnroQKB/PAjJwPM3/s1NvsP1rJPz7JZMGEgdw8Pbmbg1RKfVW9LxGc5s69JxhjeOCBB7jjjjs6lG3bto0VK1bwwAMPcNlll/HrX//aDRE6qakeXv8m1FfAhfdZM0ieEBRuzaDoRJVNc4uNe/+zmz4hgfx2ofunCFdKddT7EoEbtJ6G+vLLL+fBBx/k5ptvJiIigoKCAgIDA2lubiY2NpZFixYRERHBc8891+ZYj6sa+uDn1twuN7wKo6447a7F1Q28uT2faycndhjR+/T6Q3xZUMETN01uM4++UspzaCLoBq2noZ43bx433XQTM2bMACAiIoKXXnqJrKws7r33Xvz8/AgMDOTJJ58EYPHixcybN48BAwZ4TmPx9hdh+/NwwU/PmATAWjRl+a5C/rYqk+/NHsbtM4cSGuRP1vEqHvs4k3nj+jN//IAzvo9Syj10riEv1C3nW5ZtLbqROK3t7KCFO+Hfl0HyDFj05hnXsN2ZV87VT3zODVMHU1bbyEd7jtG/Twg/uWwEr27OJbu4hpU/vrDHF+NWSrWlcw2ptqqPwzNzoeoIhMWdXCilOHIksctuwS88Ab727zMmAWMMv39vL/ERwfzqyjFEBAewJbuUP7y/j5+/sRuAv90wUZOAUh5OE4GvaWmGN74NdWVwxZ+teX/S34LtLxBrhGbx58OpzzI3OJYz1eh/mH6UrTll/M8155xcDHxqSixv3XUeK748ypGKOha2XuZRKeWRek0iMMb4xMCjr1yV98nvrOUdr34SJt6EmfodnvpkL5s/eYvrIr8kM2wKj673J3nvWn5++SiuOKe/w99rY7ONRz7cT2rfCK5LS2xTJiLaJqCUF+kViSAkJISSkhLi4uJ6dTIwxlBSUkJIyFmuUrV3ubWGatq3YeJN1De1cN9/d/POzkIWTpjP7K8/wOUBfpyTUcT/rtjP3a9sZ1JSNL+8YjRpKW1X43pxYw45JbU8e9tUAnS+H6W8Wq9oLG5qaiI/P5/6+no3RdVzQkJCSExMJDCwi6taFWfC03MgYQTc9gHHaw3ffXEbu/LKuffykdw1e1ibJNpiM7yxLY9HV2ZwvKqBy8f24765oxiaEEF5bSMX/mkN4xOjeOHb03p18lWqt+j1jcWBgYEMGTLE3WF4rsYaWHYL+AfaJ4gL5sevbyTzWBVP3TKFy8f273CIv59w/dQkFkwYyL/XH2bJ2oOs3reOm6Yn0dhso7K+iV9cMVqTgFK9QK9IBOo0GmutJFC0H255E6IHk3W8ms+zSrj38pEOk0BrYUEBfP/iVG6YlsTfVmfw8qZcWmyG69MGM3qAE4uOK6U8niaC3qyhGl69AbI/g4X/gGEXAfDq5lwC/ITr0gaf4Q1OSYgM5vdXn8Ot5w3hrR353D5zqKuiVkr1ME0EvVV9Bbz0dSjYBtf+C8Z/w9rc1MJ/t+dz+dj+Z9W/f3jfCO69fFR3R6uUciNNBL1RbSm8eA0c2wPfeBbGXHWy6IP0I5TXNnHT9CQ3BqiU8iSaCHqb6iJ44SooyYIbXoYRl7cpfmVTLilxYcwYGuemAJVSnkY7gPcmlYXw3BVQeghuWtYhCWQcq2JLdhk3TU/SRWGUUifpE0FvUZ4Lzy+AmmKrd1DyeR12eWVTLkH+fnx9ivONxEqp3k8TQW9QchCeXwiNVfDNdyCx45iRukarkXjuuP66LoBSqg2XVg2JyFwROSAiWSJyv4PyKBF5V0R2icgeEbnNlfH0SkUH4NkroKkWvvWuwyQA8N7uQqrqm7WRWCnVgcueCETEH3gCuBTIB7aIyHJjzN5Wu90N7DXGLBCRBOCAiLxsjGl0VVy9gs0Ghdth7zuw4yXwC4DbVkDfztcoeGVzLsMSwpk+JLbTfZRSvsmVVUPTgCxjzCEAEXkNuAponQgMECnWPAURQCnQ7MKYvFvBNtj1Gux7D6oKwS8Qhs6GuY9A/PBOD9tbWMmO3HIevHKMTgmhlOrAlYlgEJDX6nU+ML3dPo8Dy4FCIBK43hhja/9GIrIYWAyQlOSjVRuFO2HppeAfBMMvhtEPWb2CQqPPeOi/1h8iOMCPr00e5OoolVJeyJWJwNGtZ/upTi8HdgIXAcOAj0VkvTGmss1BxjwNPA3W7KPdH6qHs7XAez+yVhO7exOEOV+982V+BW/tKODOC4cRHaaNxEqpjlzZWJwPtO6nmIh159/abcCbxpIFHAZ0/oL2tiyFwh0w93+7lASMMfz+/b3Ehgdx15xhLgxQKeXNXJkItgCpIjJERIKAG7CqgVrLBS4GEJF+wEjgkAtj8j6VhbD6YWvCuHFf69KhH+89xqbDpfz4klT6hHRx/QKllM9wWdWQMaZZRO4BPgL8gWeMMXtE5E57+RLgYeA5EfkSqyrpPmNMsati8kof3Ae2Jpj/KHShobepxcYjH+xnWEI4N07z0XYVpZRTXDqgzBizAljRbtuSVj8XApe5MgavduBD2LccLnoQYrs27fPLG3M4VFzDv7+VpktJKqVOS68QnqqxBlb8DBJGwXk/6NKhFXVN/G11JucNi+OiUX1dFKBSqrfQKSY81ZpHoCIPbvsQArrW2+eJT7Mor2vil/N1KUml1JnpE4EnKsuBjU/CxEWQPKNLh+aV1vLc59l8bXIiYwdGuShApVRvoonAE637PxA/mPOLLh/68qZcWozhp5eNcEFgSqneSBOBpynOgp2vwtTbIaprI4FtNsM7Owu4cEQCA6JCXRSgUqq30UTgadb8DwSEwMyfdPnQjYdKOFJRz9WTdCoJpZTzNBF4kqPpkP5fOPdOiEjo8uFv7iggIjiAy8b0c0FwSqneShOBJ/n0DxAcBed9v8uH1jW28MGXR5g3rj8hgf4uCE4p1VtpIvAU+VvhwAorCYTGdPnwlXuPUtPYwjU6w6hSqos0EXiKTx62Zhc9986zOvytHQUMjArh3CFx3RyYUqq300TgCQ6vh0NrrAbi4MguH15U1cD6zGKumjQIPz8dQKaU6hpNBJ5g7R8hcoDVZfQsvLurkBab4VrtLaSUOguaCNztyC7IXg/n3gWBZ9f3/60dBYwb1IfUfl1/mlBKKU0E7rbhnxAUAZO/eVaHZx2v4suCCq6ZlNjNgSmlfIVOOudOlYWQ/gZM/W6HtYff3J7PgWNVJMeGkxwXRlJsGAOjQ/Fv1wbw5vYC/P2EhRMG9mDgSqneRBOBO23+FxgbTL+jzeaG5hZ++VY6dU0tbbYH+guDY8JIigsjJS6cpNgw3tpRwAWp8SREBvdk5EqpXkQTgbs01sDWZ2DUfIgd0qZoa3YZdU0tPH3LFMYOiiKnuIac0lqyS2rILaklp6SWrdllVDc0A/Cr+WPccQZKqV5CE4G77HoV6sthxj0ditZmFBHoL5w/PJ7w4AAGRYdyXrt9jDGU1jRSUtNIat+IHglZKdU7aSJwB5vNaiQeNAUGT+9QvC6jiKkpsYQHd/7PIyLERQQTF6FVQkqpr0Z7DblD5kdQehBm3N1hQfqjFfXsP1rFrBFdn3ROKaXOhiYCd9jwBEQNhtFXdShal1kEwIWaCJRSPUQTQU8r3GkNIJt+B/h3rPpZm1FE38hgRvXXwWFKqZ6hiaCnbXsWAsMdDiBrsRk+yyxm1ogEXXReKdVjNBH0pOZG2PsOjLoCQjouLL8rv5yKuiZtH1BK9ShNBD3p0BqoK4NxX3dYvC6jCBG4YHh8z8allPJpmgh6UvobEBINwy5yWLw2o4jxidHEhAf1bFxKKZ+miaCnNNbC/vdhzEII6HihL69tZFdeufYWUkr1OE0EPSVzJTRWd1ot9FlWMTYDF47QaiGlVM/SRNBT0t+AiH6QMtNh8bqMIvqEBDAhMbpn41JK+TxNBD2hvhIyVsLYa8DPv0OxMYa1GUXMTI0nwF//SZRSPUuvOj1h//vQ0gDjvuawOONYNccqG7R9QCnlFpoIekL6GxCdBIlTHRavzTgOoOMHlFJu4VQiEJH/ish8EelS4hCRuSJyQESyROT+TvaZLSI7RWSPiKztyvt7hZoSOPip9TTgYLTwgaNVPL3uMKMH9GFA1NmtWayUUl+Fsxf2J4GbgEwReURERp3pABHxB54A5gFjgBtFZEy7faKBfwILjTFjgW90IXbvsPdtMC0Oq4XSCyq44ekN+An848aJPR6aUkqBk4nAGLPKGHMzMBnIBj4WkS9E5DYRCezksGlAljHmkDGmEXgNaD/d5k3Am8aYXPvnHD+bk/Bo6W9C/EjoN67N5h25Zdz4r42EBQXw+h0zGN5XJ5lTSrmH01U9IhIH3Ap8B9gB/A0rMXzcySGDgLxWr/Pt21obAcSIyBoR2SYiHWdisz57sYhsFZGtRUVFzobsfpWFkPN5h2qhzYdLWbR0E7HhQSy741xS4sPdGKRSytc5tUKZiLwJjAJeBBYYY47Yi5aJyNbODnOwzTj4/CnAxUAosEFENhpjMtocZMzTwNMAaWlp7d/Dc23+F2DgnFODyLbnlvGtZzYzMDqEl79zLv2jQtwXn1JK4fxSlY8bYz5xVGCMSevkmHxgcKvXiUChg32KjTE1QI2IrAMmABl4u+oi2LTEehqIG3Zy89L1hwgP9mfZHTOI12UmlVIewNmqodH2hl0ARCRGRO46wzFbgFQRGSIiQcANwPJ2+7wDXCAiASISBkwH9jkZk2f77C/Q3ACzf3FyU0NzC2sPFHHpmP6aBJRSHsPZRPBdY0z5iRfGmDLgu6c7wBjTDNwDfIR1cX/dGLNHRO4UkTvt++wDPgR2A5uBpcaY9C6fhaepKIAt/4aJN0L88JObNx0qpaaxhUtG93VjcEop1ZazVUN+IiLGGAMnu4aeca5kY8wKYEW7bUvavf4T8Ccn4/AO6/4PjA0uvK/N5lX7jhES6Mf5ut6AUsqDOJsIPgJeF5ElWA2+d2Ldyav2Sg/Bjpcg7dvWaGI7Ywyr9x1n5vAEQgI7zjeklFLu4mwiuA+4A/geVm+glcBSVwXl1db8EfwC4YKfttm8/2gVBeV1/ODi4Z0cqJRS7uFUIjDG2LBGFz/p2nC83PH9sHsZnPd9iOzfpmjV3mMAzBml7QNKKc/i7DiCVOB/saaKONnx3Rgz1EVxeadP/wBBETDzxx2KVu0/zoTB0fSN1HEDSinP4myvoWexngaagTnAC1iDy9QJhTth33KYcTeExbYpOl5Zz668ci7V3kJKKQ/kbCIINcasBsQYk2OMeQhwvAK7r/rk9xAaAzM6Dq/4ZL81hdLFo/v1dFRKKXVGzjYW19unoM4UkXuAAkBvb0/I3QhZH8Mlv4WQqA7Fq/YdY1B0KKP668RySinP4+wTwY+AMOAHWHMDLQK+5aKYvIsxsPphaz3iaYs7FNc3tfBZVjGXjO6LOFiPQCml3O2MTwT2wWPXGWPuBaqB21welTc5tAZyPoN5f4KgsA7Fn2cVU99k02ohpZTHOuMTgTGmBZgiejvbkTHwycMQNRimOH5AWrXvGBHBAUwfGuuwXCml3M3ZNoIdwDsi8h+g5sRGY8ybLonKWxz4AAq2wcJ/QEDHSeRsNms08awR8QQH6GhipZRncjYRxAIltO0pZADfTQQ2mzVuIHYYTLjJ4S67Cyo4XtXAJVotpJTyYM6OLNZ2gfb2vAnH0uFr/wZ/x7/Gxz/JIiI4gIt0NLFSyoM5O7L4WTquLoYx5tvdHpE3MAbWPAJ9x8LYax3usuFgCav2HePnc0cSHXbGiVqVUsptnK0aeq/VzyHANXRcbcx3lByEkky48jHw69jebrMZ/rBiL4OiQ/n2+UN6Pj6llOoCZ6uG/tv6tYi8CqxySUTeIH+z9T3pXIfFb+0oIL2gkseun6hTTiulPJ6zA8raSwWSzrhXb5W3GYKjIH5kh6K6xhb+9NEBxidGsXDCQDcEp5RSXeNsG0EVbdsIjmKtUeCb8rdA4hSH1UJL1x/iaGU9f79xEn5+OvRCKeX5nK0a0klyTmioguN7YdSVHYqOV9Xz5NqDXD62H9OG6AAypZR3cKpqSESuEZGoVq+jReRql0XlyQq2WesRJ07tUPTXjzNobLZx/7zRbghMKaXOjrNtBL8xxlSceGGMKQd+45KIPF3eFut74pS2m0trWbYlj1tmJDMkPtwNgSml1NlxNhE42s/Zrqe9S/5mq5E4NKbN5k/2H8dm4FszUtwTl1JKnSVnE8FWEfmLiAwTkaEi8ldgmysD80jGWA3FgztWC63NKCI5LowUfRpQSnkZZxPB94FGYBnwOlAH3O2qoDxWyUGoK4PEaW02NzS3sOFgCReOSHBTYEopdfac7TVUA9zv4lg834mBZIPbJoKt2WXUNbUwK1UTgVLK+zjba+hjEYlu9TpGRD5yWVSeqpOBZOsyigj0F2YMi3NTYEopdfacrRqKt/cUAsAYU4YvrlncyUCytRlFpCXHEh7sm+3nSinv5mwisInIySklRCQFB7OR9monBpK1ax84VlnP/qNVXDhSq4WUUt7J2VvYXwKficha++tZQMeV2nuzEwPJ2vUYWpdRBKDtA0opr+VsY/GHIpKGdfHfCbyD1XPId5wYSDYorc3mtRlFJEQGM3qAzsKhlPJOzk469x3gh0AiViI4F9hA26Ure7f8zZAwCkKjT25qsRk+yyrm4lH9ENEJ5pRS3snZNoIfAlOBHGPMHGASUOSyqDzNiYFk7eYX2p1fTnltk7YPKKW8mrOJoN4YUw8gIsHGmP1Ax8n42xGRuSJyQESyRKTTcQgiMlVEWkTk607G07NKsuwDydq3DxQjAhcMj3dTYEop9dU521icbx9H8DbwsYiUcYalKkXEH3gCuBTIB7aIyHJjzF4H+/0R8NxxCXmOB5KtzTjO+EFRxITrmsRKKe/lbGPxNfYfHxKRT4Eo4MMzHDYNyDLGHAIQkdeAq4C97fb7PvBfrKonz5TfcSBZRW0TO/PKuWfOcDcGppRSX12XR0AZY9aeeS8ABgF5rV7nA9Nb7yAig4BrsBqdO00EIrIYe3fVpKQeXiHTGMj+vMNAss+yirEZmKXzCymlvNzZrlnsDEfdaNoPQnsMuM8Y03K6NzLGPG2MSTPGpCUk9PCFt2AblGTC6IVtNq/LKCIyJICJg6N7Nh6llOpmrpwTIR8Y3Op1Ih3bFdKA1+xdL+OBK0Sk2Rjztgvj6prtz0NgGIz72slNNpthXWYRM4fHE+DvylyqlFKu58qr2BYgVUSGiEgQcAOwvPUOxpghxpgUY0wK8AZwl0clgYZqSH8Txl4LIX1Obv4sq5gjFfVcPra/G4NTSqnu4bInAmNMs4jcg9UbyB94xhizR0TutJcvcdVnd5s9b0FjNUy+pc3m57/IJj4iiHnnaCJQSnk/l06XaYxZAaxot81hAjDG3OrKWM7K9hcgfgQMPtXGnVNSwycHjvP9i1IJDvB3Y3BKKdU9tIK7M8f3W91GJ90CraaPeGFDDv4i3Dy9h3svKaWUi2gi6MyOF8EvACbceHJTTUMzr2/NY945A+jXJ8SNwSmlVPfRROBIcyPsehVGXgERp7qrvrmjgKr6Zm49L8V9sSmlVDfTRODIgRVQWwKTv3lykzGGF77I5pxBUUxOinZfbEop1c00ETiy/QXoMwiGnZpl+4uDJWQer+Zb56XolNNKqV5FE0F75blw8BOYeDP4neoV9Ozn2cSFB3Hl+AFuDE4ppbqfJoL2drxsfZ+06OSmvNJaVu8/xo3TkggJ1C6jSqneRRNBa/UVsPkpSL0UYpJPbn5xYw5+Itx8rnYZVUr1PpoIWtvwT2sBmjm/OLnJGMPbOwq4ZHRfBkSFujE4pZRyDU0EJ9SWwoYnYPQCGDjp5OacklqOVzXodNNKqV5LE8EJn/3Vmldozi/bbN58uBSA6UNi3RGVUkq5nCYCgKqjsPlfMP466Du6TdGmw6XEhgcxLCHCTcEppZRraSIAWPdnaGmE2fd3KNqcXcK0lFgdO6CU6rU0EZTnwrbnrO6isUPbFBWW15FXWsc0rRZSSvVimgjW/tGaXfTCn3co2pJttQ9oIlBK9Wa+nQiKs2Dnq5B2O0QldijefLiUyOAARg/o4+BgpZTqHXw7Eex6xXoauOAnDos3Hy4lLSUGfz9tH1BK9V6+nQiKDkDsMIjo26GopLqBzOPVTBsS54bAlFKq5/h2IijOgPhUh0VbsssAbR9QSvV+vpsIWpqg9JC1JrEDmw+XEhLoxzmDono4MKWU6lm+mwjKssHW3OkTwebsEiYNjiEowHd/RUop3+C7V7niTOu7gyeCyvom9hZWarWQUson+HAiyLC+xw3vULQtpwyb0fmFlFK+wYcTQSZE9IPQ6A5Fmw+XEuAnTEqK6fm4lFKqh/lwIsg4bUPx+MQoQoN0NTKlVO/nm4nAmE67jtY1trA7v1zHDyilfIZvJoKaYqgvd/hEsCOvjKYWo+0DSimf4ZuJoOREj6GOTwSbD5ciAlNStH1AKeUbfDMRnOwx5DgRjO7fhz4hgT0clFJKuYePJoJMCAiBqMFtNjc229ieW8b0oVotpJTyHT6aCDKspwG/tqf/ZUEF9U02bR9QSvkU300EnbQPAExN0USglPIdLk0EIjJXRA6ISJaIdFgQWERuFpHd9q8vRGSCK+MBoKkeynIc9hjafLiE4X0jiIsIdnkYSinlKVyWCETEH3gCmAeMAW4UkTHtdjsMXGiMGQ88DDztqnhOKj0ImA5PBC02w9bsMn0aUEr5HFc+EUwDsowxh4wxjcBrwFWtdzDGfGGMKbO/3Ah0XC+yu3Uy2dz+o5VUNTRr+4BSyue4MhEMAvJavc63b+vM7cAHLozHciIRtJts7kT7gM44qpTyNQEufG9HC/0ahzuKzMFKBDM7KV8MLAZISkr6alEVZ1jdRoPC2mzefLiUxJhQBkaHfrX3V0opL+PKJ4J8oHVH/USgsP1OIjIeWApcZYwpcfRGxpinjTFpxpi0hISErxaVgx5Dxhg2Hy7VpwGllE9yZSLYAqSKyBARCQJuAJa33kFEkoA3gVuMMRkujMVijFU11K594GBRDSU1jdo+oJTySS6rGjLGNIvIPcBHgD/wjDFmj4jcaS9fAvwaiAP+KSIAzcaYNFfFRGUhNNV0eCI41T6gM44qpXyPK9sIMMasAFa027ak1c/fAb7jyhjaODHHULsngs2HS0iIDCYlLszBQUop1bv51sjikizre6tEYIxhk719wP5UopRSPsW3EkFxBgT3sZaotMsvq+NIRb22DyilfJbvJYL4VGh156/jB5RSvs7HEkHHHkObD5cSFRrIiL6RbgpKKaXcy3cSQUMVVBZ0HFGcXcrUlBj8/LR9QCnlm3wnEThoKD5eWc/h4hqtFlJK+TTfSQTFHRPB5mwdP6CUUr6TCEbNh8VrIG7YyU1bDpcSGujP2IF93BeXUkq5mUsHlHmUoDAYOKnNpi3ZZUxOjibQ33fyoVJKteezV8DK+ib2H60kLVnbB5RSvs1nE8GO3HJsRtcnVkopn00EW7NL8fcTJiZFuzsUpZRyK59NBFuySxkzoA8Rwb7TTKKUUo74ZCJobLaxM6+ctJQYd4eilFJu55OJYE9hBfVNNm0fUEopfDQRbM0uAyAtWZ8IlFLKJxPBluxSkuPC6NsnxN2hKKWU2/lcIjDGsC2nTMcPKKWUnc8lgsPF1kL1U7WhWCmlAB9MBCfbB7ShWCmlAB9MBFuyS4kJC2RYQri7Q1FKKY/gc4lga04ZaSm6UL1SSp3gU4mgqKqBw8U12j6glFKt+FQi2JZjLUSj7QNKKXWKTyWCLdllBAf4MW5glLtDUUopj+FTiWBrdikTB0cTFOBTp62UUqflM1fE2sZm0gsrdX4hpZRqx2cSwc7cclpsRmccVUqpdnwmEQQG+HHRqL5M1onmlFKqDZ9ZlWVqSixTb9VqIaWUas9nngiUUko5polAKaV8nCYCpZTycS5NBCIyV0QOiEiWiNzvoFxE5O/28t0iMtmV8SillOrIZYlARPyBJ4B5wBjgRhEZ0263eUCq/Wsx8KSr4lFKKeWYK58IpgFZxphDxphG4DXgqnb7XAW8YCwbgWgRGeDCmJRSSrXjykQwCMhr9Trfvq2r+yAii0Vkq4hsLSoq6vZAlVLKl7kyETia8N+cxT4YY542xqQZY9ISEhK6JTillFIWVw4oywcGt3qdCBSexT5tbNu2rVhEcs4ypnig+CyP9RTefg4av/t5+zlo/GcnubMCVyaCLUCqiAwBCoAbgJva7bMcuEdEXgOmAxXGmCOne1NjzFk/EojIVmNM2tke7wm8/Rw0fvfz9nPQ+LufyxKBMaZZRO4BPgL8gWeMMXtE5E57+RJgBXAFkAXUAre5Kh6llFKOuXSuIWPMCqyLfettS1r9bIC7XRmDUkqp0/O1kcVPuzuAbuDt56Dxu5+3n4PG383EuilXSinlq3ztiUAppVQ7mgiUUsrH+UwiONMEeJ5GRJ4RkeMikt5qW6yIfCwimfbvHrvcmogMFpFPRWSfiOwRkR/at3vTOYSIyGYR2WU/h9/at3vNOYA175eI7BCR9+yvvSZ+EckWkS9FZKeIbLVv85r4AUQkWkTeEJH99r+HGZ52Dj6RCJycAM/TPAfMbbftfmC1MSYVWG1/7amagZ8aY0YD5wJ323/n3nQODcBFxpgJwERgroici3edA8APgX2tXntb/HOMMRNb9b33tvj/BnxojBkFTMD6t/CsczDG9PovYAbwUavXDwAPuDsuJ+JOAdJbvT4ADLD/PAA44O4Yu3Au7wCXeus5AGHAdqyBj15zDlij9VcDFwHvedv/IyAbiG+3zZvi7wMcxt4xx1PPwSeeCHBycjsv0M/YR17bv/d1czxOEZEUYBKwCS87B3u1yk7gOPCxMcbbzuEx4OeArdU2b4rfACtFZJuILLZv86b4hwJFwLP26rmlIhKOh52DryQCpya3U91PRCKA/wI/MsZUujuerjLGtBhjJmLdWU8TkXFuDslpInIlcNwYs83dsXwF5xtjJmNV694tIrPcHVAXBQCTgSeNMZOAGtxdDeSArySCLk9u56GOnVivwf79uJvjOS0RCcRKAi8bY960b/aqczjBGFMOrMFqt/GWczgfWCgi2VjrgVwkIi/hPfFjjCm0fz8OvIW1zonXxI917cm3P0kCvIGVGDzqHHwlEZycAE9EgrAmwFvu5pjOxnLgW/afv4VV7+6RRESAfwP7jDF/aVXkTeeQICLR9p9DgUuA/XjJORhjHjDGJBpjUrD+z39ijFmEl8QvIuEiEnniZ+AyIB0viR/AGHMUyBORkfZNFwN78bRzcHdjSg822lwBZAAHgV+6Ox4n4n0VOAI0Yd1V3A7EYTX8Zdq/x7o7ztPEPxOr+m03sNP+dYWXncN4YIf9HNKBX9u3e805tDqX2ZxqLPaK+LHq13fZv/ac+Lv1lvhbncdEYKv9/9HbQIynnYNOMaGUUj7OV6qGlFJKdUITgVJK+ThNBEop5eM0ESillI/TRKCUUj5OE4FSPUhEZp+YBVQpT6GJQCmlfJwmAqUcEJFF9rUIdorIU/bJ56pF5FER2S4iq0Ukwb7vRBHZKCK7ReStE3PLi8hwEVllX89gu4gMs799RKv56V+2j8JWym00ESjVjoiMBq7HmvBsItAC3AyEA9uNNQnaWuA39kNeAO4zxowHvmy1/WXgCWOtZ3Ae1khxsGZi/RHW2hhDseYEUsptAtwdgFIe6GJgCrDFfrMeijUpmA1YZt/nJeBNEYkCoo0xa+3bnwf+Y58jZ5Ax5i0AY0w9gP39Nhtj8u2vd2KtO/GZy89KqU5oIlCqIwGeN8Y80GajyIPt9jvd/Cynq+5paPVzC/p3qNxMq4aU6mg18HUR6Qsn18hNxvp7+bp9n5uAz4wxFUCZiFxg334LsNZYay/ki8jV9vcIFpGwnjwJpZyldyJKtWOM2Ssiv8JaGcsPawbYu7EWFRkrItuACqx2BLCmEV5iv9AfAm6zb78FeEpEfmd/j2/04Gko5TSdfVQpJ4lItTEmwt1xKNXdtGpIKaV8nD4RKKWUj9MnAqWU8nGaCJRSysdpIlBKKR+niUAppXycJgKllPJx/w/LPaA5G4nLLAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(history.history['categorical_accuracy'])\n",
    "plt.plot(history.history['val_categorical_accuracy'])\n",
    "plt.title('model accuracy')\n",
    "plt.ylabel('accuracy')\n",
    "plt.xlabel('epoch')\n",
    "plt.legend(['train', 'test'], loc='upper left')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Save your model for later"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 171,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_x = xtrain[-2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 172,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_x = test_x.reshape(1,224,224,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 173,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 224, 224, 3)"
      ]
     },
     "execution_count": 173,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_x.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 174,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_y = ytrain[-2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "32"
      ]
     },
     "execution_count": 175,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_y.argmax()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "result_dict = {}\n",
    "for count,value in enumerate(classes):\n",
    "    result_dict[count] = value"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "model = keras.models.load_model('mdl_wts.hdf5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [],
   "source": [
    "result = model.predict(test_x)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [],
   "source": [
    "sorted_index_array = np.argsort(result[0]) \n",
    "sorted_array = result[0][sorted_index_array] "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0.   , 0.002, 0.001, 0.   , 0.   , 0.   , 0.001, 0.   , 0.002,\n",
       "        0.   , 0.01 , 0.   , 0.002, 0.03 , 0.   , 0.001, 0.   , 0.   ,\n",
       "        0.001, 0.   , 0.001, 0.012, 0.   , 0.   , 0.001, 0.001, 0.   ,\n",
       "        0.   , 0.   , 0.003, 0.011, 0.004, 0.909, 0.001, 0.001, 0.   ,\n",
       "        0.002, 0.   , 0.   , 0.   ]], dtype=float32)"
      ]
     },
     "execution_count": 199,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "np.round(result,3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[21 13 32]\n"
     ]
    }
   ],
   "source": [
    "print(sorted_index_array[-3 :])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 201,
   "metadata": {},
   "outputs": [],
   "source": [
    "predictions = [labels[k] for k in sorted_index_array[-3 :]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 202,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['29. Buxus sempervirens', '21. Fraxinus sp', '39. Schinus terebinthifolius']"
      ]
     },
     "execution_count": 202,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 167,
   "metadata": {},
   "outputs": [],
   "source": [
    "image = test_x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 168,
   "metadata": {},
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "Invalid shape (1, 224, 224, 3) for image data",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-168-56ed2357daf6>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mget_ipython\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrun_line_magic\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;34m'matplotlib'\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;34m'inline'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      2\u001b[0m \u001b[1;32mfrom\u001b[0m \u001b[0mmatplotlib\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mpyplot\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 3\u001b[1;33m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mimshow\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mimage\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0minterpolation\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;34m'nearest'\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m      4\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mshow\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\keras\\lib\\site-packages\\matplotlib\\pyplot.py\u001b[0m in \u001b[0;36mimshow\u001b[1;34m(X, cmap, norm, aspect, interpolation, alpha, vmin, vmax, origin, extent, filternorm, filterrad, resample, url, data, **kwargs)\u001b[0m\n\u001b[0;32m   2722\u001b[0m         \u001b[0mfilternorm\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mTrue\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mfilterrad\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;36m4.0\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mresample\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mNone\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0murl\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mNone\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2723\u001b[0m         data=None, **kwargs):\n\u001b[1;32m-> 2724\u001b[1;33m     __ret = gca().imshow(\n\u001b[0m\u001b[0;32m   2725\u001b[0m         \u001b[0mX\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mcmap\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mcmap\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mnorm\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mnorm\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0maspect\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0maspect\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   2726\u001b[0m         \u001b[0minterpolation\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0minterpolation\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0malpha\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0malpha\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mvmin\u001b[0m\u001b[1;33m=\u001b[0m\u001b[0mvmin\u001b[0m\u001b[1;33m,\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\keras\\lib\\site-packages\\matplotlib\\__init__.py\u001b[0m in \u001b[0;36minner\u001b[1;34m(ax, data, *args, **kwargs)\u001b[0m\n\u001b[0;32m   1436\u001b[0m     \u001b[1;32mdef\u001b[0m \u001b[0minner\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0max\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdata\u001b[0m\u001b[1;33m=\u001b[0m\u001b[1;32mNone\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1437\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mdata\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 1438\u001b[1;33m             \u001b[1;32mreturn\u001b[0m \u001b[0mfunc\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0max\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0mmap\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0msanitize_sequence\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0margs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   1439\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   1440\u001b[0m         \u001b[0mbound\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnew_sig\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mbind\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0max\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m*\u001b[0m\u001b[0margs\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\keras\\lib\\site-packages\\matplotlib\\axes\\_axes.py\u001b[0m in \u001b[0;36mimshow\u001b[1;34m(self, X, cmap, norm, aspect, interpolation, alpha, vmin, vmax, origin, extent, filternorm, filterrad, resample, url, **kwargs)\u001b[0m\n\u001b[0;32m   5521\u001b[0m                               resample=resample, **kwargs)\n\u001b[0;32m   5522\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m-> 5523\u001b[1;33m         \u001b[0mim\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mset_data\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mX\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m   5524\u001b[0m         \u001b[0mim\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mset_alpha\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0malpha\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m   5525\u001b[0m         \u001b[1;32mif\u001b[0m \u001b[0mim\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mget_clip_path\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m \u001b[1;32mis\u001b[0m \u001b[1;32mNone\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32m~\\anaconda3\\envs\\keras\\lib\\site-packages\\matplotlib\\image.py\u001b[0m in \u001b[0;36mset_data\u001b[1;34m(self, A)\u001b[0m\n\u001b[0;32m    707\u001b[0m         if not (self._A.ndim == 2\n\u001b[0;32m    708\u001b[0m                 or self._A.ndim == 3 and self._A.shape[-1] in [3, 4]):\n\u001b[1;32m--> 709\u001b[1;33m             raise TypeError(\"Invalid shape {} for image data\"\n\u001b[0m\u001b[0;32m    710\u001b[0m                             .format(self._A.shape))\n\u001b[0;32m    711\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mTypeError\u001b[0m: Invalid shape (1, 224, 224, 3) for image 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": [
    "%matplotlib inline \n",
    "from matplotlib import pyplot as plt\n",
    "plt.imshow(image, interpolation='nearest')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "from PIL import Image\n",
    "import numpy as np\n",
    "\n",
    "w, h = 512, 512\n",
    "data = np.zeros((h, w, 3), dtype=np.uint8)\n",
    "data[0:256, 0:256] = [255, 0, 0] # red patch in upper left\n",
    "img = Image.fromarray(data, 'RGB')\n",
    "img.save('my.png')\n",
    "img.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.save('imageclassifier/models/leaves_model.h5')"
   ]
  }  
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "keras",
   "language": "python",
   "name": "keras"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}