NN_Project_Gesture_Recognition_AG_CR.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Gesture Recognition\n",
    "In this group project, you are going to build a 3D Conv model that will be able to predict the 5 gestures correctly. Please import the following libraries to get started."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np\n",
    "import os\n",
    "from scipy.misc import imread, imresize\n",
    "from imageio import imread\n",
    "import cv2 as cv\n",
    "import datetime\n",
    "import os\n",
    "import matplotlib\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline  "
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We set the random seed so that the results don't vary drastically."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "np.random.seed(30)\n",
    "import random as rn\n",
    "rn.seed(30)\n",
    "from keras import backend as K\n",
    "import tensorflow as tf\n",
    "tf.set_random_seed(30)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In this block, you read the folder names for training and validation. You also set the `batch_size` here. Note that you set the batch size in such a way that you are able to use the GPU in full capacity. You keep increasing the batch size until the machine throws an error."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_doc = np.random.permutation(open('Project_data/train.csv').readlines())\n",
    "val_doc = np.random.permutation(open('Project_data/val.csv').readlines())\n",
    "batch_size = 15"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Generator\n",
    "This is one of the most important part of the code. The overall structure of the generator has been given. In the generator, you are going to preprocess the images as you have images of 2 different dimensions as well as create a batch of video frames. You have to experiment with `img_idx`, `y`,`z` and normalization such that you get high accuracy."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def generator(source_path, folder_list, batch_size):\n",
    "    print( 'Source path = ', source_path, '; batch size =', batch_size)\n",
    "    # Read every third frame of the video; So 10 frames out of 30 will be used for Project\n",
    "    img_idx = [x for x in range(0,30,3) ] #create a list of image numbers you want to use for a particular video\n",
    "    while True:\n",
    "        t = np.random.permutation(folder_list)\n",
    "        num_batches = len(folder_list) // batch_size # calculate the number of batches\n",
    "        for batch in range(num_batches): # we iterate over the number of batches\n",
    "            # We will use image size of (120 x 120) with 3 channels\n",
    "            batch_data = np.zeros((batch_size,len(img_idx),120,120,3)) # x is the number of images you use for each video, (y,z) is the final size of the input images and 3 is the number of channels RGB\n",
    "            batch_labels = np.zeros((batch_size,5)) # batch_labels is the one hot representation of the output\n",
    "            for folder in range(batch_size): # iterate over the batch_size\n",
    "                imgs = os.listdir(source_path+'/'+ t[folder + (batch*batch_size)].split(';')[0]) # read all the images in the folder\n",
    "                for idx,item in enumerate(img_idx): #  Iterate iver the frames/images of a folder to read them in\n",
    "                    image = imread(source_path+'/'+ t[folder + (batch*batch_size)].strip().split(';')[0]+'/'+imgs[item]).astype(np.float32)\n",
    "                    \n",
    "                    #crop the images and resize them. Note that the images are of 2 different shape \n",
    "                    #and the conv3D will throw error if the inputs in a batch have different shapes\n",
    "                    height, width  = image.shape[0:2]\n",
    "                    if width != height:\n",
    "                        # Cropped image of above dimension to change the aspect ratio\n",
    "                        image = image[:,21:141,:]\n",
    "                    #Resize the image to (120,120)\n",
    "                    image = cv.resize(image,(120,120),interpolation = cv.INTER_AREA)\n",
    "                    \n",
    "                    #Normalise the pixel values by dividing with 255.0 (max pixel value)\n",
    "                    image_norm = image/255.0\n",
    "                    \n",
    "                    # Batch of 3 channel images normalized and fed into the batch_data array\n",
    "                    batch_data[folder,idx,:,:,0] = image_norm[:,:,0] #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,1] = image_norm[:,:,1] #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,2] = image_norm[:,:,2] #normalise and feed in the image\n",
    "                    \n",
    "                batch_labels[folder, int(t[folder + (batch*batch_size)].strip().split(';')[2])] = 1\n",
    "            yield batch_data, batch_labels #you yield the batch_data and the batch_labels, remember what does yield do\n",
    "\n",
    "        \n",
    "        # write the code for the remaining data points which are left after full batches\n",
    "        batch_size_rem = (len(folder_list) % batch_size)\n",
    "        if batch_size_rem != 0:\n",
    "            batch_data = np.zeros((batch_size_rem,len(img_idx),120,120,3))\n",
    "            batch_labels = np.zeros((batch_size_rem,5))\n",
    "            for folder in range(batch_size_rem):\n",
    "                imgs = os.listdir(source_path+'/'+ t[folder + (num_batches*batch_size)].split(';')[0]) # read all the images in the folder\n",
    "                # For each video, iterate through the images\n",
    "                for idx,item in enumerate(img_idx): #  Iterate iver the frames/images of a folder to read them in\n",
    "                    image = imread(source_path+'/'+ t[folder + (num_batches*batch_size)].strip().split(';')[0]+'/'+imgs[item]).astype(np.float32)\n",
    "                    \n",
    "                    #crop the images and resize them. Note that the images are of 2 different shape \n",
    "                    #and the conv3D will throw error if the inputs in a batch have different shapes\n",
    "                    height, width  = image.shape[0:2]\n",
    "                    if width != height:\n",
    "                        # Cropped image of above dimension to change the aspect ratio\n",
    "                        image = image[:,21:141,:]\n",
    "                    #Resize the image to (120,120)\n",
    "                    image = cv.resize(image,(120,120),interpolation = cv.INTER_AREA)\n",
    "                    \n",
    "                    #Normalise the pixel values by dividing with 255.0 (max pixel value)\n",
    "                    image_norm = image/255.0\n",
    "                    \n",
    "                    # Batch of 3 channel images normalized and fed into the batch_data array\n",
    "                    batch_data[folder,idx,:,:,0] = image_norm[:,:,0] #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,1] = image_norm[:,:,1] #normalise and feed in the image\n",
    "                    batch_data[folder,idx,:,:,2] = image_norm[:,:,2] #normalise and feed in the image\n",
    "                    \n",
    "                batch_labels[folder, int(t[folder + (num_batches*batch_size)].strip().split(';')[2])] = 1\n",
    "            yield batch_data, batch_labels #you yield the batch_data and the batch_labels, remember what does yield do\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note here that a video is represented above in the generator as (number of images, height, width, number of channels). Take this into consideration while creating the model architecture."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "# training sequences = 663\n",
      "# validation sequences = 100\n",
      "# epochs = 25\n"
     ]
    }
   ],
   "source": [
    "curr_dt_time = datetime.datetime.now()\n",
    "train_path = 'Project_data/train'\n",
    "val_path = 'Project_data/val'\n",
    "num_train_sequences = len(train_doc)\n",
    "print('# training sequences =', num_train_sequences)\n",
    "num_val_sequences = len(val_doc)\n",
    "print('# validation sequences =', num_val_sequences)\n",
    "num_epochs = 25 # choose the number of epochs\n",
    "print ('# epochs =', num_epochs)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Model\n",
    "Here you make the model using different functionalities that Keras provides. Remember to use `Conv3D` and `MaxPooling3D` and not `Conv2D` and `Maxpooling2D` for a 3D convolution model. You would want to use `TimeDistributed` while building a Conv2D + RNN model. Also remember that the last layer is the softmax. Design the network in such a way that the model is able to give good accuracy on the least number of parameters so that it can fit in the memory of the webcam."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model 1: TimeDistributed Conv2d + RNN Model (CNN + RNN)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import Sequential, Model\n",
    "from keras.layers import Dense, GRU, Flatten, TimeDistributed, Flatten, BatchNormalization, Activation, Dropout, LSTM\n",
    "from keras.layers.convolutional import Conv2D, Conv3D, MaxPooling2D, MaxPooling3D\n",
    "from keras.callbacks import ModelCheckpoint, ReduceLROnPlateau\n",
    "from keras import optimizers\n",
    "from keras.optimizers import Adam"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_1 (TimeDist (None, 10, 60, 60, 32)    4736      \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 10, 60, 60, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_2 (TimeDist (None, 10, 58, 58, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 10, 58, 58, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_3 (TimeDist (None, 10, 29, 29, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_4 (TimeDist (None, 10, 29, 29, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_5 (TimeDist (None, 10, 29, 29, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_6 (TimeDist (None, 10, 14, 14, 64)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_7 (TimeDist (None, 10, 12544)         0         \n",
      "_________________________________________________________________\n",
      "time_distributed_8 (TimeDist (None, 10, 64)            802880    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_9 (TimeDist (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_1 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 1,276,901\n",
      "Trainable params: 1,276,517\n",
      "Non-trainable params: 384\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Model 1 : Conv2D + RNN with LSTM\n",
    "\n",
    "# Initial layers take the image nD array as input\n",
    "modelC2D_1 = Sequential()\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_1.add(TimeDistributed(Conv2D(32, (7, 7), strides=(2, 2), activation='relu', padding='same'), input_shape=(10, 120, 120, 3)))\n",
    "modelC2D_1.add(BatchNormalization())\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_1.add(TimeDistributed(Conv2D(32, (3,3), kernel_initializer=\"he_normal\", activation='relu')))\n",
    "modelC2D_1.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_1.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_1.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_1.add(BatchNormalization())\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_1.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_1.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_1.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# Flatten the feature map into Vector\n",
    "modelC2D_1.add(TimeDistributed(Flatten()))\n",
    "\n",
    "# Add a fully connected layer with 50% connections dropped out\n",
    "modelC2D_1.add(TimeDistributed(Dense(64, activation='relu')))\n",
    "modelC2D_1.add(Dropout(0.5))\n",
    "\n",
    "# Add second fully connected layer with 50% connections dropped out\n",
    "modelC2D_1.add(TimeDistributed(Dense(128, activation='relu')))\n",
    "modelC2D_1.add(Dropout(0.5))\n",
    "\n",
    "# Add a LSTM layer\n",
    "modelC2D_1.add(LSTM(256, return_sequences=False, dropout=0.5))\n",
    "\n",
    "# Softmax layer for final 5-class output\n",
    "modelC2D_1.add(Dense(5, activation='softmax'))\n",
    "modelC2D_1.summary()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Now that you have written the model, the next step is to `compile` the model. When you print the `summary` of the model, you'll see the total number of parameters you have to train."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_1 (TimeDist (None, 10, 60, 60, 32)    4736      \n",
      "_________________________________________________________________\n",
      "batch_normalization_1 (Batch (None, 10, 60, 60, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_2 (TimeDist (None, 10, 58, 58, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_2 (Batch (None, 10, 58, 58, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_3 (TimeDist (None, 10, 29, 29, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_4 (TimeDist (None, 10, 29, 29, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_3 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_5 (TimeDist (None, 10, 29, 29, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_4 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_6 (TimeDist (None, 10, 14, 14, 64)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_7 (TimeDist (None, 10, 12544)         0         \n",
      "_________________________________________________________________\n",
      "time_distributed_8 (TimeDist (None, 10, 64)            802880    \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_9 (TimeDist (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_1 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_3 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 1,276,901\n",
      "Trainable params: 1,276,517\n",
      "Non-trainable params: 384\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "# Compile the model\n",
    "optimiser = Adam(lr = 0.0001) #write your optimizer\n",
    "modelC2D_1.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['categorical_accuracy'])\n",
    "print (modelC2D_1.summary())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us create the `train_generator` and the `val_generator` which will be used in `.fit_generator`."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_generator = generator(train_path, train_doc, batch_size)\n",
    "val_generator = generator(val_path, val_doc, batch_size)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "model_name = 'model_init' + '_' + str(curr_dt_time).replace(' ','').replace(':','_') + '/'\n",
    "    \n",
    "if not os.path.exists(model_name):\n",
    "    os.mkdir(model_name)\n",
    "        \n",
    "filepath = model_name + 'model-{epoch:05d}-{loss:.5f}-{categorical_accuracy:.5f}-{val_loss:.5f}-{val_categorical_accuracy:.5f}.h5'\n",
    "\n",
    "checkpoint = ModelCheckpoint(filepath, monitor='val_loss', verbose=1, save_best_only=False, save_weights_only=False, mode='auto', period=1)\n",
    "\n",
    "LR = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=2, cooldown=1, verbose=1) # write the REducelronplateau code here\n",
    "callbacks_list = [checkpoint, LR]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The `steps_per_epoch` and `validation_steps` are used by `fit_generator` to decide the number of next() calls it need to make."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "if (num_train_sequences%batch_size) == 0:\n",
    "    steps_per_epoch = int(num_train_sequences/batch_size)\n",
    "else:\n",
    "    steps_per_epoch = (num_train_sequences//batch_size) + 1\n",
    "\n",
    "if (num_val_sequences%batch_size) == 0:\n",
    "    validation_steps = int(num_val_sequences/batch_size)\n",
    "else:\n",
    "    validation_steps = (num_val_sequences//batch_size) + 1"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Let us now fit the model. This will start training the model and with the help of the checkpoints, you'll be able to save the model at the end of each epoch."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Source path =  Project_data/val ; batch size = 15\n",
      "Source path =  Project_data/train ; batch size = 15\n",
      "Epoch 1/25\n",
      "45/45 [==============================] - 103s 2s/step - loss: 1.6569 - categorical_accuracy: 0.1955 - val_loss: 1.5283 - val_categorical_accuracy: 0.2900\n",
      "\n",
      "Epoch 00001: saving model to model_init_2020-03-0819_30_13.737684/model-00001-1.66064-0.19306-1.52826-0.29000.h5\n",
      "Epoch 2/25\n",
      "45/45 [==============================] - 19s 426ms/step - loss: 1.5491 - categorical_accuracy: 0.2859 - val_loss: 1.3456 - val_categorical_accuracy: 0.4400\n",
      "\n",
      "Epoch 00002: saving model to model_init_2020-03-0819_30_13.737684/model-00002-1.54905-0.28507-1.34558-0.44000.h5\n",
      "Epoch 3/25\n",
      "45/45 [==============================] - 20s 437ms/step - loss: 1.4154 - categorical_accuracy: 0.4237 - val_loss: 1.0999 - val_categorical_accuracy: 0.5600\n",
      "\n",
      "Epoch 00003: saving model to model_init_2020-03-0819_30_13.737684/model-00003-1.41176-0.42534-1.09988-0.56000.h5\n",
      "Epoch 4/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 1.2753 - categorical_accuracy: 0.4904 - val_loss: 1.1247 - val_categorical_accuracy: 0.5500\n",
      "\n",
      "Epoch 00004: saving model to model_init_2020-03-0819_30_13.737684/model-00004-1.27693-0.49321-1.12473-0.55000.h5\n",
      "Epoch 5/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 1.1365 - categorical_accuracy: 0.5749 - val_loss: 0.9253 - val_categorical_accuracy: 0.5800\n",
      "\n",
      "Epoch 00005: saving model to model_init_2020-03-0819_30_13.737684/model-00005-1.13303-0.57919-0.92527-0.58000.h5\n",
      "Epoch 6/25\n",
      "45/45 [==============================] - 20s 443ms/step - loss: 0.9737 - categorical_accuracy: 0.6459 - val_loss: 1.0784 - val_categorical_accuracy: 0.5900\n",
      "\n",
      "Epoch 00006: saving model to model_init_2020-03-0819_30_13.737684/model-00006-0.97368-0.64555-1.07835-0.59000.h5\n",
      "Epoch 7/25\n",
      "45/45 [==============================] - 20s 441ms/step - loss: 0.8422 - categorical_accuracy: 0.6950 - val_loss: 0.9068 - val_categorical_accuracy: 0.6300\n",
      "\n",
      "Epoch 00007: saving model to model_init_2020-03-0819_30_13.737684/model-00007-0.83346-0.70136-0.90679-0.63000.h5\n",
      "Epoch 8/25\n",
      "45/45 [==============================] - 20s 446ms/step - loss: 0.6900 - categorical_accuracy: 0.7734 - val_loss: 0.9116 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00008: saving model to model_init_2020-03-0819_30_13.737684/model-00008-0.67805-0.77526-0.91158-0.72000.h5\n",
      "Epoch 9/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 0.5654 - categorical_accuracy: 0.8251 - val_loss: 1.0026 - val_categorical_accuracy: 0.6400\n",
      "\n",
      "Epoch 00009: saving model to model_init_2020-03-0819_30_13.737684/model-00009-0.57093-0.82202-1.00265-0.64000.h5\n",
      "\n",
      "Epoch 00009: ReduceLROnPlateau reducing learning rate to 4.999999873689376e-05.\n",
      "Epoch 10/25\n",
      "45/45 [==============================] - 20s 435ms/step - loss: 0.3993 - categorical_accuracy: 0.8903 - val_loss: 1.0049 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00010: saving model to model_init_2020-03-0819_30_13.737684/model-00010-0.40244-0.88839-1.00489-0.70000.h5\n",
      "Epoch 11/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.3217 - categorical_accuracy: 0.9244 - val_loss: 0.9026 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00011: saving model to model_init_2020-03-0819_30_13.737684/model-00011-0.32292-0.92308-0.90259-0.71000.h5\n",
      "Epoch 12/25\n",
      "45/45 [==============================] - 20s 443ms/step - loss: 0.2743 - categorical_accuracy: 0.9259 - val_loss: 0.9517 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00012: saving model to model_init_2020-03-0819_30_13.737684/model-00012-0.27422-0.92459-0.95172-0.71000.h5\n",
      "Epoch 13/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 0.2222 - categorical_accuracy: 0.9541 - val_loss: 1.0529 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00013: saving model to model_init_2020-03-0819_30_13.737684/model-00013-0.22271-0.95324-1.05295-0.70000.h5\n",
      "\n",
      "Epoch 00013: ReduceLROnPlateau reducing learning rate to 2.499999936844688e-05.\n",
      "Epoch 14/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.2080 - categorical_accuracy: 0.9320 - val_loss: 1.0719 - val_categorical_accuracy: 0.6700\n",
      "\n",
      "Epoch 00014: saving model to model_init_2020-03-0819_30_13.737684/model-00014-0.20150-0.93665-1.07191-0.67000.h5\n",
      "Epoch 15/25\n",
      "45/45 [==============================] - 20s 435ms/step - loss: 0.1778 - categorical_accuracy: 0.9644 - val_loss: 1.0751 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00015: saving model to model_init_2020-03-0819_30_13.737684/model-00015-0.17873-0.96380-1.07506-0.68000.h5\n",
      "\n",
      "Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.249999968422344e-05.\n",
      "Epoch 16/25\n",
      "45/45 [==============================] - 20s 436ms/step - loss: 0.1636 - categorical_accuracy: 0.9674 - val_loss: 1.0763 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00016: saving model to model_init_2020-03-0819_30_13.737684/model-00016-0.16529-0.96682-1.07627-0.70000.h5\n",
      "Epoch 17/25\n",
      "45/45 [==============================] - 20s 436ms/step - loss: 0.1380 - categorical_accuracy: 0.9733 - val_loss: 1.0895 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00017: saving model to model_init_2020-03-0819_30_13.737684/model-00017-0.13920-0.97285-1.08953-0.70000.h5\n",
      "\n",
      "Epoch 00017: ReduceLROnPlateau reducing learning rate to 6.24999984211172e-06.\n",
      "Epoch 18/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.1509 - categorical_accuracy: 0.9720 - val_loss: 1.1081 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00018: saving model to model_init_2020-03-0819_30_13.737684/model-00018-0.13897-0.97738-1.10808-0.68000.h5\n",
      "Epoch 19/25\n",
      "45/45 [==============================] - 20s 440ms/step - loss: 0.1314 - categorical_accuracy: 0.9733 - val_loss: 1.0942 - val_categorical_accuracy: 0.6900\n",
      "\n",
      "Epoch 00019: saving model to model_init_2020-03-0819_30_13.737684/model-00019-0.13302-0.97285-1.09423-0.69000.h5\n",
      "\n",
      "Epoch 00019: ReduceLROnPlateau reducing learning rate to 3.12499992105586e-06.\n",
      "Epoch 20/25\n",
      "45/45 [==============================] - 20s 441ms/step - loss: 0.1398 - categorical_accuracy: 0.9733 - val_loss: 1.0850 - val_categorical_accuracy: 0.6900\n",
      "\n",
      "Epoch 00020: saving model to model_init_2020-03-0819_30_13.737684/model-00020-0.13855-0.97285-1.08505-0.69000.h5\n",
      "Epoch 21/25\n",
      "45/45 [==============================] - 20s 448ms/step - loss: 0.1362 - categorical_accuracy: 0.9631 - val_loss: 1.0734 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00021: saving model to model_init_2020-03-0819_30_13.737684/model-00021-0.12980-0.96833-1.07344-0.68000.h5\n",
      "\n",
      "Epoch 00021: ReduceLROnPlateau reducing learning rate to 1.56249996052793e-06.\n",
      "Epoch 22/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.1211 - categorical_accuracy: 0.9837 - val_loss: 1.0764 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00022: saving model to model_init_2020-03-0819_30_13.737684/model-00022-0.12282-0.98341-1.07641-0.70000.h5\n",
      "Epoch 23/25\n",
      "45/45 [==============================] - 20s 440ms/step - loss: 0.1364 - categorical_accuracy: 0.9778 - val_loss: 1.0832 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00023: saving model to model_init_2020-03-0819_30_13.737684/model-00023-0.13443-0.97738-1.08321-0.71000.h5\n",
      "\n",
      "Epoch 00023: ReduceLROnPlateau reducing learning rate to 7.81249980263965e-07.\n",
      "Epoch 24/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.1140 - categorical_accuracy: 0.9837 - val_loss: 1.0839 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00024: saving model to model_init_2020-03-0819_30_13.737684/model-00024-0.11490-0.98341-1.08395-0.70000.h5\n",
      "Epoch 25/25\n",
      "45/45 [==============================] - 20s 436ms/step - loss: 0.1297 - categorical_accuracy: 0.9763 - val_loss: 1.0842 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00025: saving model to model_init_2020-03-0819_30_13.737684/model-00025-0.13061-0.97587-1.08418-0.71000.h5\n",
      "\n",
      "Epoch 00025: ReduceLROnPlateau reducing learning rate to 3.906249901319825e-07.\n"
     ]
    }
   ],
   "source": [
    "H1 = modelC2D_1.fit_generator(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                    callbacks=callbacks_list, validation_data=val_generator, \n",
    "                    validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA34AAAFRCAYAAAArTH/eAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4yLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvOIA7rQAAIABJREFUeJzs3XdcU9f7B/DPDYEk7BGWooKAKFi1iIqjioq7rqq1dW/92WodtFarX63W0VqttXUi7llXrYOqOKpVq7irOMCiYhmyQZaSPL8/UqIRCMNAAJ/368WLJPfec557Mk6e3HPPFYiIwBhjjDHGGGOsyhLpOwDGGGOMMcYYY2WLEz/GGGOMMcYYq+I48WOMMcYYY4yxKo4TP8YYY4wxxhir4jjxY4wxxhhjjLEqjhM/xhhjjDHGGKvi3orEz9nZGd98802JthEEAVu3bi2jiKqu7OxsCIKAPXv2lGg7X19ffPrpp2UUVfkpz/0ICAhAo0aN1Pd//vlnWFpaat3m1q1bEAQB169ff+P6fXx8MGnSpDcuhzFW+XE/W364n+V+lrHSeisSP125ffs2Bg8eDCcnJ0gkEtSqVQu9e/fGqVOnAAApKSmYNGkSvLy8YGJiAgcHB/Tp0wd3797VKMfPzw+CIEAQBBgaGsLOzg5t2rTB0qVLkZWVVex4YmJiMHDgQHh5eUEsFsPf37/IbZydndV1F/b3JqRSKWJiYtC9e/cSbXfkyBEsXLjwjequ6C5cuABBEHDy5MkCl8+fPx8WFhbIyMgoVfkjRozA/fv33yTEAn355ZeoX79+vsePHj2KefPm6bw+bYYNGwYDAwOsX7++XOt92xw6dEj9eSASiWBmZoZ69eph1KhRuHbtWonLK+w1VB4++ugjvP/++3qpm5Uc97NF4362cNzPlo4uk9WKLm9f8/5MTEzg5uaGgQMH4vTp0yUub/Xq1TA1NdV9oMVQmr6VE79iOnr0KHx8fBAdHY3AwECEhYXh4MGD8PX1xdixYwGoOojIyEjMnTsXV69exaFDh/Ds2TO0a9cOycnJGuUNGDAAMTExePjwIY4fP46+fftiyZIl8Pb2xtOnT4sVU05ODqytrTFlypRidUYAEBoaipiYGMTExODq1asAgL1796ofi4mJKXC758+fF6t8AHBwcIBEIin2+gBgbW0NMzOzEm1T2TRv3hwNGjRAYGBgvmVEhKCgIAwcOBAmJialKt/Y2Bh2dnZvGmax2djYlOtzlpKSgt27d2P69OlYu3ZtudWrjUKhgFKp1HcYZebevXuIjo7GjRs3sGTJEiQlJaFJkyZYt26dvkNjVRD3s9zPvinuZ1lxhYSEICYmBrdv38batWshk8nQrl07zJ49W9+hlS3SgzZt2tCIESPoq6++IltbW7KwsKAZM2aQQqGgr7/+muzs7Egul9OMGTM0tktLS6MxY8aQXC4niURCjRs3pqNHj2qsc/36dWrevDlJJBJyd3enXbt2Ua1atWjevHnqddLT02nixIlUrVo1kslk1KhRI9q7d69GOQBoy5YtRESUkZFBdnZ21Llz5wL3JykpqdB9TUhIIAD022+/aez/yJEj860bFRVFlpaWNHz48ELLK8zQoUOpffv2JdomKiqKANCpU6fyLWvWrBmNGzeOpk2bRvb29uTk5ERERBs3biQfHx8yMzMjuVxO3bt3p4iICPV2WVlZBIB2796tcX/t2rX00UcfkYmJCdWoUYN++OGHfPV98sknGvfHjx9Ps2bNIltbW7K2tqZRo0ZRZmamep3c3FwKCAgga2trMjU1pYEDB9J3331HJiYmWve7qH24c+cOAaC9e/dS586dSSaTkaurK+3cuVOjnIiICGrfvj1JJBKqWbMmrV69Ot9+vO7nn38miURCCQkJGo8fO3aMANC1a9eIiCg2Npb69+9PTk5OJJVKqW7duvTzzz9rbDN16lRq2LCh+v5PP/1EFhYWGuts2rSJnJ2dSSqV0nvvvUd79uzRqCc7O5uGDx+uXsfV1ZXmzJlDL168UJcJQONv8eLFRETUuHFj+uyzz9R1ZWVl0aRJk8jBwYGMjIzonXfe0XhfpaenEwAKCgqiDz/8kExMTKhmzZq0fPnyQtvrVcuXL6f33nuP0tPTycTEhG7cuJFvnfPnz1P79u3JxMSETE1NydfXl65fv65efujQIfL19SWpVEqWlpbUrl07ioqKIiKiPn36UM+ePTXKW7VqlcbrKa/NN27cSG5ubiQSiSgyMpLOnz9P/v7+JJfLydTUlJo1a0YnT57UKCs7O5tmzJhBzs7OZGRkRE5OTvTFF18QEdEHH3xAvXv3zrc/Pj4+Wl9PSUlJNGzYMLKxsSGJRELNmjWj06dPq5f//fffBIB+/fVX6tixI8lkMnJzc1O/Pwtz8OBBAkDJycn5lo0dO5akUin9+++/6v0q7Wto/fr11LhxY/V7sUePHvTPP/+o61IqlTRr1iyqVasWGRkZka2tLXXp0oUUCoVGrE2bNiWpVEpOTk40ZswYddxTp07NV3dR+15VcD/L/Sz3s9zPFrefzesr8mIuyN69e6lBgwZkZGRE9vb29Nlnn1FWVpZ6+dWrV6lt27Zkbm5OJiYm5OnpSb/88otG27m7u5NEIiEbGxtq27YtPX36tND69LGvCxcuJAB0+fJl9WOTJ0+mOnXqkFQqpZo1a9LEiRPp2bNnRPSyr3z1L++1+dtvv1HLli3J0tJS/X3j9TqLapNz585R27ZtydjYmOzs7Kh///7qvlfb60YbvSV+5ubm9MUXX9C9e/coKCiIAFCXLl3o888/p3v37tHGjRsJAB05ckS9Xd++falWrVr0+++/U1hYGE2cOJEMDQ3pzp07RESUmZlJ1apVoy5dutD169fp/Pnz5OPjQzKZTN0hKZVK8vPzozZt2tDZs2fpwYMHtGbNGjI0NKSQkBB1Xa92SPv37ycAdPbs2RLv64MHDwgAnT9/XmP/C+qQiIg++eQTMjc31/hiUxzaOiQANHv27HyPF9UhmZqa0oQJEygsLIxu3rxJRERr166lw4cPU0REBF2+fJk6d+5Mnp6e6g+wwjokR0dHWr9+PYWHh9N3332Xr00K6pAsLCzoiy++oLt379KhQ4fI1NSUFixYoF5n/vz5ZG5uTtu3b6d79+7RokWLyNLSssgOqah9yOuQ3NzcaO/evRQeHk6TJk0iIyMjevjwIRERKRQK8vT0pObNm9OlS5foypUr1KZNGzI1NdXaIaWkpJCxsTEtXbpU4/F+/fpRkyZN1PcjIyPp+++/p6tXr9I///xDQUFBJJFIND5Ei+qQzp49S4Ig0Ndff0337t2jHTt2UPXq1TU+8J49e0Zz5syhS5cuUWRkJO3Zs4esra3pu+++IyLVl7FPP/2UPDw8KCYmhmJiYigjI4OI8ndI48aNIzs7O9q/fz/dvXuXZs2aRYIg0IULF4jo5Yd0tWrVaOPGjRQREaH+kL106ZLW54yIqH79+hQUFERERIMHD87XzqGhoWRkZETDhg2jy5cv071792jz5s3qD/ADBw6QSCSiadOm0c2bN+nWrVu0atUqdaJR3MTPxMSE/P396dKlS3Tnzh3KyMigo0eP0tatWyksLIzu3r1LU6ZMIZlMRo8ePVJv+8EHH5CjoyPt3LmTIiIi6Ny5c+oO6tixYyQWiykmJka9/o0bNwiARuL6us6dO5ObmxuFhITQ7du3afTo0SSVSikyMpKIXnZwderUof3791N4eDhNmDCBJBKJOuEtiLbELzIykgDQqlWriOjNXkNr1qyhI0eO0IMHDyg0NJQ6dOhA77zzDuXm5hKR6guVtbU1BQcH06NHj+jKlSu0ePFi9efjb7/9RiYmJrRmzRoKDw+nCxcukK+vL3Xp0oWIVElMz549yd/fX113dnZ2oftdlXA/y/0s97Pczxa3ny0q8btw4QIJgkDTp0+nu3fv0oEDB8jBwYHGjRunXsfFxYVGjBhBd+7coYiICDp48CD9/vvvRET0xx9/kJGREe3cuZMePnxIN2/epJUrV2pN/PSxr1lZWSSVSmnatGnqx2bPnk3nzp2jyMhI+v3338nFxYXGjx9PRKqk/ttvvyVjY2P1c5eWlkZERDt37qR9+/bR/fv36ebNmzRw4EBycHBQLy+qTS5fvkwymYwWLlxId+/epWvXrtH7779P77zzDr148ULr60YbvSV+r76RiIg8PT2pfv36Go81aNCApk6dSkRE4eHhBIAOHz6ssc67776r/uUuMDCQTExMNH4ZzHuC8zqkU6dOkUQioZSUFI1yhg8frvGl79UO6dtvvyUAlJiYWKL9zM3NpU6dOlGTJk00OhhtHdKqVasIAMXFxZWoLm0dkoeHB/3000/5Hi+qQ/Ly8iKlUqm13ujoaI1fRwrrkD7//HON7WrVqkVz5szRqO/1DunVD+i8ffTz81Pft7GxoW+++UZjnZ49exbZIRW1D3kd0ooVK9Tr5OTkkJGREW3cuJGIVF+M84725Pn333/J0NBQa4dEpHqt1atXT33/6dOnZGRkRIGBgVq3GzFiBPXq1Ut9v6gOqWfPntSxY0eNMvI+FLX9qjd37lxq1KiR+v60adPIy8sr33qvdkgJCQlkYGBAmzZt0ljH39+funfvTkQvP6SnT5+usY6Tk1O+5/F1f/75J8lkMvUH5okTJ8jCwkLjQ65Xr17k6+tbaBmNGjWi/v37F7q8uImfoaFhsd6ftWvXpmXLlhGR6ggJAAoODi5wXaVSSa6urrRw4UL1Y59++ik1bdq00PLzyvzjjz/UjykUCvLw8KAJEyYQ0cvPvzVr1qjXycrKIgMDA9q6dWuhZWtL/IiITExM8r2nX1Xc19DrHj9+rJHszp07lxo2bKhOBF/XuHFjjaNMRES3b98mABQeHk5ERP3796du3boVWXdVw/0s97Pcz3I/m6eofraoxK9Xr17Utm1bjce2bt1KBgYG9PTpU8rNzSVDQ8NCR1Rs3ryZ7OzsipWYEOl3X728vKhfv36Fbr9582aytLRU33/9e0Jhnj9/TlKplH799Vd1OdrapE+fPvk+w9LS0sjAwICOHz9ORMXvW1+lt3P8GjZsqHHfwcEBDRo0yPdY3jj8sLAwAEDr1q011mndujVu376tXqdevXqwsrJSL69fvz4sLCzU90NDQ/H8+XNUr14dpqam6r+tW7ciPDy8wFiJqMT7p1AoMGTIENy/fx/79u2DSFS8ps6r601P/n7V3bt3SzUDVpMmTfLFceXKFfTs2RPOzs4wMzODu7s7AODRo0day3p1ViwAqF69OuLi4kq9TVxcHBITE+Hr66uxTvPmzbWWWZJ9eLV+IyMjyOVydf1hYWGoVq0anJ2d1etUq1YNtWvXLrL+sWPH4s6dOzh37hwAYNOmTZBIJPjoo4/U6+Tm5mLu3Llo0KABbGxsYGpqii1bthTZzq8KCwtDixYtNB5r1apVvvV+/vln+Pj4wNbWFqamppg/f36J6gFU54IpFIp87882bdqo3595Xn9eq1WrVuRrYc2aNfjggw/U5zq0bdsWlpaW+OWXX9TrXLlyBR06dChwe4VCgRs3bqBjx47F3qfC1KpVK985HjExMRg9ejTq1KkDCwsLmJqa4uHDh+p2vHLlCgwMDNC+ffsCyxQEAaNHj8a6detARMjOzsbWrVsxevToQuO4ffs2DAwMNJ5jkUiEVq1aaW1zqVQKGxubIttcGyLS+Gwo7WsoNDQUPXr0UL8X69WrB+Dle3HAgAFISEiAs7MzRo4ciR07diAzMxOA6jm9du0a5s+fr/FZ3rRpUwAo9PP8bcL9bMG4ny3eNtzPFq0q9bPa3L59u8B6FQoF7t69CwMDA0yZMgUff/wx2rdvj3nz5uHvv/9Wr/v+++/DxsYGzs7OGDhwIIKCgvKdk/sqfe7r6/3bjh070LJlSzg6OsLU1BRjxoxBSkoK0tLStJZz//59fPzxx3B1dYW5uTmsrKyQk5Ojft6LapPQ0FBs2bJF4zPU0dERCoXijfo3vSV+hoaGGvfzZt56/bGiJk549Ql6/ckqiFKphIWFBa5fv67xFxYWhuDg4AK38fDwAPCyUyzK8+fP8eGHH+LixYs4ffo0nJycirUdoJptyNLSEjY2NsXepqy8fvJzamoqOnToAKlUik2bNiE0NBTnz58HUPRJ6UZGRhr3i/PcFmebknbcJdkHbfUX57VWmGbNmqFhw4bqCUrWrVuHAQMGaMwKNW/ePPz4448ICAjAiRMncP36dQwYMKBEJ/8XJ8YNGzbg888/x7Bhw3D06FFcu3YNAQEBJarnVa/XV1AMJX0tJCcnY8+ePdixYwfEYjHEYjEMDQ3x+PHjfJO8FLa/xfmiJxKJ8n35fPHiRb71CpoU4KOPPsLVq1fxww8/4Ny5c7h+/Trq1KlTonYcMWIEoqKicOrUKezZswcKhULjS0pBCtofXbS5NpGRkcjMzISrqyuA0r+GkpOT0bFjR5iamqrfi2fOnAHw8r3o6uqK8PBwrF69GpaWlpg5cyY8PT0RFxcHUo1Ywddff53v8zw8PDzfF4a3EfezBeN+tmTbcD9buKrSz5a23lcfX7RoEcLCwtCrVy9cuXIF3t7e6llkrayscOPGDezcuRPOzs5YtmwZ3Nzciny/l/e+ZmVl4cGDB+r+7eTJkxg0aBA6d+6M3377DdeuXcPSpUsBaH8/EhE6deqExMRErF69Gn/99ReuX78OU1NT9XZFtYlSqcTYsWML7N8GDBhQqv0DKtGsnl5eXgCg/mKQ5+zZs+plXl5eCAsLQ0pKinr57du3kZqaqr7v4+ODlJQUZGdnw83NTeOvZs2aBdbdsWNH2NnZYf78+QUufzVDz8zMRI8ePRAWFoYzZ84UWmZBnjx5gm3btqFv377F/uWyPN26dQvJyclYtGgR2rRpg7p16yIhIUEvsdjb28PGxgYXLlzQePyvv/7Sup2u9sHLywv//vuvxi92MTEx+Oeff4q1/dixY7F7924cOHAA9+7dw5gxYzSWnzlzBr169cKQIUPQqFEjuLm5lfgXHi8vL/WvnXlev3/mzBm0aNECn376Kby9veHu7p5vH4yMjKBQKLTW5eHhAQMDA/zxxx/5ys97f5bW5s2b1R+Qr374HT9+HBcuXMCtW7cAAI0bN8axY8cKLEMsFqNhw4Y4evRoofXY2dkhOjpa47G8Gfm0USgUOHfuHCZNmoRu3bqhfv36sLGxwePHj9XrNG7cGAqFAidOnCi0HFtbW/Tu3RuBgYEIDAzExx9/rHWKaC8vL+Tm5qq/UAGqjuLcuXNv3ObaLFq0CDKZTD2VfGlfQzdv3kRKSgq+/fZbre9FmUyGbt26YcmSJbh58yZiYmJw5MgR9XMaFhaW77Pczc1N/YW6OK9fpsL9rP5xP/sS97OayrKfLYqXl1eB9RoYGKh/tAEAd3d3TJgwAb/++iumTZuG1atXq5cZGhqiXbt2mD9/vjoJ2rVrV4H16Wtfly1bhufPn6Nfv34AVJ99zs7OmDVrFpo0aQJ3d3dERUVpbFPQc/f48WM8fPgQs2bNQocOHeDp6QmlUon09HSN9bS1iY+PD27evFlg/5Y3wqI0/Zu4RGvrkaurK/r164fx48djzZo1qFWrFlatWoVbt25h+/btAFTDgmbNmoVBgwZh/vz5yMrKwmeffQaZTKYup127dvD398cHH3yAb7/9Fg0bNkRycjLOnz8PqVRa4NAqY2NjbNy4Eb1794a/vz+mTp0KDw8PZGRk4OjRowgMDMS9e/eQnp6Orl274smTJzhw4ABEIhFiY2MBABYWFhpxZGVlITY2FkqlEgkJCThz5gwWLlyI6tWrl+g6O3nXXElKSsKzZ8/U91899F23bl18+umnb3zBUxcXFxgaGmL58uWYMGECIiIiMH369Dcq801MmTIF3333Hdzc3ODt7Y0DBw7gjz/+0Prrm672oUuXLvDw8MDAgQOxbNkyiEQiBAQEFHt67YEDB+Lzzz/H8OHD0bhxY3h7e2ss9/DwwOHDh/Hnn39CLpcjMDAQt27dQo0aNYod45QpU+Dn54dvvvkGH374Ia5du4YVK1bkq2fv3r0IDg6Gu7s79u3bl+8XeRcXFzx69AhXr15FzZo1YWJiovFaBlRTTo8ZMwaff/45LCwsUK9ePWzduhUnTpzQSExKY+3atejbt2++a9XUr18f7777LtauXYvly5dj+vTpaNWqFYYPH44JEybA3NwcoaGh8PDwgLe3N/73v/+hT58+cHZ2xqBBgyAWi3H27Fl06NABzs7O8Pf3x4oVKxAUFIQ2bdrg6NGjOHjwYJHxGRgYwM3NDZs2bULjxo2RlZWF6dOnw8DAQL1Ow4YN0bNnT4wYMQI//PADmjRpgvj4eFy+fBmffPKJer2xY8eiU6dOePHihfpXxcI0bNgQXbp0wahRo7B69Wo4ODhg2bJlePToEaZMmVLCVi7Y06dPkZ2djczMTNy7dw/r1q3DgQMHEBgYCEdHRwClfw3Vrl0bYrEYP/74I8aPH4/w8HB8+eWXGtutXLkSMpkMPj4+MDMzw5EjR/DixQv1kNBvvvkGPXv2hKOjIz766CMYGxvj/v372LVrFzZv3qyu+9SpU7h79y7kcjnMzc3z/ULMVLifLRj3s9zPFqaq9LN57t27l++x2rVrY9q0aWjRogW++uor9fDqgIAAjB49Gra2toiPj8f8+fPRq1cvODs7IyEhASEhIfD09AQA7Nq1C/Hx8WjZsqX6x4S4uDj18teVx74mJiYiNjYWOTk5ePDgAbZv344NGzZgzpw5ePfddwGonrtHjx5h27ZtaN68OU6ePIkNGzZolOPi4oLs7GwcO3YM3t7ekMlkcHBwgLm5OVavXg1HR0fExsbi888/13jtFtUms2bNQsuWLTFy5EiMHz8eVlZW+Oeff7Bv3z7MmjULjo6OxXrd5FOiMwJ1pKCTrtu3b09Dhw7VeKxTp040cOBA9f3U1FT1NNNGRkYFTjN99epV8vX1JSMjI6pduzbt2LEj3zTTmZmZNG3aNHJ2diZDQ0Oyt7enTp060YkTJ9Tr4JWTzvPcvHmTBgwYQI6OjmRoaEg1atSg3r17qydXOHXqVL6pVfP+NmzYoLH/eY+LxWKysbGh1q1b05IlSzSmUS6Owup7fZ3SzDZW0MnT27dvp9q1a6un+f7jjz8IAO3YsYOICj/p/PUTflu2bEljx44ttL6C6v/qq6/Iw8NDfT83N5emTp1KVlZW6mmm58yZQ3K5vJDWKt4+5J10HhoaqrFd9erVNSbfCA8Pp3bt2pFEIiEnJydauXJlkdNMv2rUqFH5Jt7IEx8fTz179iRTU1OSy+U0efJkmjp1qsZJvMWZZnr9+vVUq1Ytkkgk1LJlS9q9e7fGSc1ZWVk0dOhQsrS0JAsLCxo6dCgtXrxY40TlzMxM6tOnD1laWhZrmml7e3syNDQsdOrl/fv3a8Sorc3OnDlDAOjPP/8scHneDHN575uzZ89SmzZtSCaTkampKbVo0ULjsg8HDhwgHx8fkkgkZGlpSe3bt6cnT54QkWqClZkzZ5KDgwOZmprS0KFD6fvvvy/wcg6vCw0NJR8fH5JKpeTi4kJBQUH59isrK4u++OILcnJyUn9+fPnll/nKcnd3p3fffbfA/X1d3uUcrK2tycjIqNDLObx+Eru9vb3WaZ9fn6La2NiY6tatS6NGjcpX1pu8hrZs2UIuLi4kkUjIx8eHTp8+rfF5sW3bNmratClZWFiQTCajBg0a5JuU5vjx49SmTRsyMTEhExMT8vLyoilTpqiXx8TEUIcOHcjMzKzAz6KqivtZ7me5n+V+Nk9RbZbXVxT0d/DgQSJ6eTmHvPfzq5dzSE1NpQ8//JBq1qxJRkZGZGdnRwMHDqTY2FgiUs1c3bp1a7K2tiapVEoeHh75Zl19XXntq0wmo9q1a9OAAQM0+k8i1feCqVOnklwuJxMTE+rZsydt2LCBAFB8fLx6nbzPTODl5Rx+//138vT0JIlEQp6ennTo0CGNvrc4bXL58mXq2rUrmZubk1QqJXd3dxo3bpx6orvCXjfaCESlOKOasQpqwIABePToUb6hFoxVBtnZ2ahevToWLFigvmA1Y4xVJNzPMlZ5VZqhnoy97tGjRwgODkabNm0gCAL279+PXbt2ISgoSN+hMVYiCoUCcXFx+PHHH2FgYIAhQ4boOyTGGON+lrEqpuKd2cwAqE4MfXUK19f/tm3bpu8Q9c7AwADbtm1DixYt0LhxY+zatQvr16/HsGHD9B0aYyVy584dVK9eHdu2bcOmTZuKHqPPGHtj3M8WjftZxqoWHupZQeXm5uLhw4eFLre3t1df04wxxhhjJcP9LGPsbcOJH2OMMcYYY4xVcTzUkzHGGGOMMcaqOE78GGOMMcYYY6yKq/SzekZHRwMA5HI5EhIS9BxNxcXtox23j3bcPtpx+2inq/apVq2aDqJ5u3AfWTzcPtpx+2jH7aMdt492umif4vaPfMSPMcYYY4wxxqq4cjnit3LlSly9ehUWFhZYsmRJvuVEhA0bNuDatWuQSCQYP348ateuXR6hMcYYY4wxxliVVy5H/Pz8/DBjxoxCl1+7dg2xsbFYvnw5xowZg3Xr1pVHWIwxxhhjjDH2ViiXI36enp54+vRpocsvX76M1q1bQxAE1KlTBxkZGUhOToaVlVV5hMcYY4wxxthbhYiQnZ0NpVIJQRDKrJ64uDjk5OSUWfmVXXHbh4ggEokglUpL/XxViMldkpKSIJfL1fdtbGyQlJTEiR9jjDHGGGNlIDs7G4aGhhCLyzYdEIvFMDAwKNM6KrOStE9ubi6ys7Mhk8lKV1epttKxgq4hX1gmGxISgpCQEADAokWL1AmjWCzWSB6ZJm4f7bh9tOP20Y7bRztuH8YYq3iUSmWZJ31Mt8Ri8RsdPa0Qz7aNjY3GNKaJiYmFHu3z9/eHv7+/+n7edjxVrHbcPtpx+2jH7aMdt492fDkHxhireMpyeCcrO2/yvFWIyzn4+PjgzJkzICLcv38fxsbGPMyTMcYYY4wxxnSkXBK/ZcuWYebMmYiOjsa4ceNw8uRJHDt2DMeOHQMAvPvuu7Czs8P7NQBwAAAgAElEQVTEiROxZs0ajBo1qjzCYowxxhhjjOlJamoqNm7cWOLtBg8ejNTU1BJvN2nSJBw6dKjE21UV5TLUc9KkSVqXC4LAyR5jjBUD5eYC9/8GZTyDILcH5A6AqVmZDNkhpRJISwbi40CJcRBkpkC1GoCNHQRRhRgwwsoQ/X0ZMDaF4FpX36EwxqqotLQ0bN68GcOGDdN4XKFQaJ3wZMuWLWUcWdVUIc7xY4wxVjh68QK4cx105Tzo+kUg85nq8bwVJDLA1h6Q20OQO6j+29oDtg6qJM1IUnjZWZlAQhwQHwtKiAMSYkHxcarHEuKA3Bcv1827YWQEODhBcKgBVKsBwVH1H7aOEHjmtiqBcl9AuTMQyMmBaOYSCJY2+g6JMVYFLViwAI8ePUKHDh1gaGgIY2Nj2Nvb4/bt2zh9+jRGjBiB6Oho5OTkYOTIkRg0aBAAoFmzZggODkZGRgYGDRqEpk2b4vLly3BwcMD69euLNevl2bNnMW/ePCgUCjRs2BALFy6ERCLBggULcOzYMYjFYrRu3Rr/+9//cPDgQfzwww8QiUQwNzfHvn37yrppygQnfoyxSotycoDUJCA1GUhNAqUmA1kZgJklBAsrwMIasLACzC0hVLKZy+h5DhB2TZXs3bgEZGUCMhMIDZtCaNwckNsDCXGqZC3+v/9x0aCwa8Dz59CYK9nCGpDbQbB1AMwsgOREUHwskBgHPEvXrFhmrCq7Wg0IDXwAucN/RxbtgMwMUPRjICYKFBMFiggDLv3xsi6xGLCvrkoEHWtAqFYDcKwJsrAon0ZjOiOIDSH6v+lQLvoCylWLIApYAMHQUN9hMcbKiHJnICgqUqdlCjVcIPpotNZ1ZsyYgXv37uH48eM4f/48hgwZgpMnT6JmzZoAgCVLlsDKygpZWVno1q0bunbtCmtra40yIiMjsWLFCixevBhjx47FkSNH0KdPH631ZmdnY/Lkydi1axdcXV0xceJEbN68GX379kVwcDDOnDkDQRDUw0mXLVuGbdu2wdHRsVRDTCuKyvVNiDH2VqDsLCA58WUyl5fcpSSDUpNe3s/KLLyMV+8IAmBqrkqALK00kkIhLzm0sAK0HBnTRik1AhG98XBLyskGbl1RJXs3Q4GcbMDEDIJ3CwiNWwB1G2p++XZywes1EhGQlqJKCuNj1UfuKCEOFB6mWmZlo0roarkBtvaqxM5WdaQQxqZa9+P1YX+UnQnE/AuKeQzEPFElhI8igCvn1JfqeSoWQ7R8JwRDozdqH1a+BCdniIZPgnL1ItC2VcDQCTwLIGOsTDVq1Eid9AHA+vXrERwcDACIjo5GZGRkvsSvRo0aqF+/PgCgQYMGiIqKKrKeBw8eoGbNmnB1dQUA9OvXD5s2bcLw4cMhkUgQEBCA9u3bq68k4OPjg8mTJ6N79+7o0qWLTvZVHzjxY4yVO1IogKT4V45Yxb68nRAHpBfwa5qhEWD5X5JWvRYEz3f/S9isVYmcpRVgYQPIZEBaquZRwP8SRUpNBlKSQE8eqhIgpRL5ryJacvEAIDZUxWOZl1AWkFxaWgGmFhrnx1F2JuhGKOjqBeDWZeD5c8DMAkKzNqpkr847JTpaKQiCOpEtj3OzBKkx4OIOwcVd43F6ngPE/guKiYLJ8yxkcdJXKQmNW0Do9iHo8C9ALVcIbbvpOyTGWBko6shceTE2NlbfPn/+PM6ePYuDBw9CJpOhb9++BV7DTiJ5+aOtgYEBsrOzi6ynoGuIA6rr5B0+fBh//vknDhw4gA0bNmD37t349ttvcfXqVZw4cQIdO3bEsWPH8iWglQEnfowxnaLcXNVwy4xnQEb6y2Tu1cQuKR5QKl9uZGAAWNuqzk1r1Ex15Mna9r+E7r+kSWZS/KMN1nLVH5DviJg6TqUCeJYGpCT/lxQmAbm5pdpnEyNDZERHvUwuY56A7v6d/1y8vH01s1Ttk1QGPLirOo/OwhpCS38IjVsC7p4QRJX7XDnBSALUrA2hZm2YyOXI4uscVlpCjwGgqEjQrnWgarUgeNTXd0iMsSrCxMQEz549K3BZeno6LCwsIJPJEBERgatXr+qsXjc3N0RFRSEyMhIuLi7Yu3cvfH19kZGRgaysLLRv3x7e3t5o1aoVAODhw4fw9vaGt7c3jh8/jujoaE78GGNVDz1LQ07kPShjo1WJTGaG6n/GM1Dms5dJXt7jOYX80mZmAdg6QKjtATRt8/KcM7k9YCUv90lBBJEBYG6l+kPhCWJxFJbY0POc/448vpJcvnok8lk6BL8uqiN7tevyTJmsQhJEIohGToFyYQCUqxdBNHMpBBs7fYfFGKsCrK2t0aRJE7Rr1w5SqRRyuVy9zM/PD1u2bIG/vz9q164Nb29vndUrlUqxdOlSjB07Vj25y+DBg5GSkoIRI0YgJycHRITZs2cDAL755htERkaCiNCqVSt4eXnpLJbyJFBhxzoriejoaACAXC5HAv+iXChuH+24ffKj3FzQ6SOg33aokrtXSWSAsYnqz8RUdV6Y7OVtGJsCJiYQjE0BGzvVzJLSomfYqqz49aOdrtqnWrVqOoim4lm5ciWuXr0KCwsLLFmypMB1bt++jY0bN0KhUMDMzAxff/11scrWZR9JsU+gXBAA2DpA9MW3ECSlOye2IuL3sHbcPtpV1vbJzMzUGFpZVsRiMXJLOaLmbVDS9inoeStu/8hH/Bhj+dDdm1DuWAtEPwY834XVx6OQCpEqsZOZVLoZMhmryPz8/NC5c2esWLGiwOUZGRlYt24dvvrqK8jlcr3NKCc4OEE0OgDKn+aBNi0HRgfwZC+MMVaJ8Lc3xpgaJcaDdq8HXTkH2NhBNH4G0KgZjGxtIVTCXzMZqww8PT3x9OnTQpf/+eefaNasmXoIlIUeL48hvOMDodcg0P4tqsleOn2gt1gYY6wwM2bMQGhoqMZjo0aNQv/+/fUUUcXAiR9jlQRlpIP2bQbdvqb68uXrB9T20Mkv7vTiOejoPlDwHgCA0HMAhI69tV74mzFWPmJiYpCbm4s5c+YgKysLXbt2RZs2bfQWj9ClL/D4H9DezaDqzhDq6+68G8YY04UFCxboO4QKiRM/xio4IgJdPA36ZT2QkQ7UbQA6FwI6fUQ1WUozPwi+fhDsS37+ExEBNy5CuSsISIiD0LglhH7DeeIGxioQhUKByMhIzJo1C8+fP8fMmTPh7u5e4DkdISEhCAkJAQAsWrRIfZRQLBZrTJrwpihgLpKmj4Vi3RJYLQ6C2NFJZ2Xrg67bp6rh9tGusrZPXFwcxOV06kZ51VNZlaR9JBJJqV9v/CwwVoFR7L9Qbl8N3LkBuNSBaPJcCDVcQFmZoGsXQH+dBh3eBTq0E3Cpo0oAm7wHwazooWAU8wTKXYHA7WtAtZoQTZkHoV7DctgrxlhJ2NjYwMzMDFKpFFKpFPXq1cOjR48KTPz8/f3VFxwGoJ5woiwmn6AxX4AWTEXivKkQzVisuqZjJVVZJ+coL9w+2lXW9snJyYFBOcyozZO7aFfS9snJycn3euPJXRirxOjFC1DwHlDwbsBQAmHg/0Fo3Uk93b8gM4bQoj3Qoj0oORF06YwqCdyxFrRrHeDlrUoCGzbLN/MeZWWCDu0CnfgNMJJC+Gg0hDZdeMIWxiooHx8frF+/HgqFArm5uYiIiEC3bvq/kLpg6wDRmC+gXDYbyqAfIPq/6XxJEsYYq8D4mx5jFQzdvQnl1lVA3L+qo3f9R6kuZF4IwcoGQqfeQKfeoCcPQRf/UP0Ffg+SyiB4t1CdD1invipB3LsRSEtRXSy892AI5pbltm+MsfyWLVuGsLAwpKenY9y4cfjwww/Vv/527NgRTk5OaNSoEQICAiASidCuXTvUrFlTz1GrCPUaQug3QnVx90M7IfQYoO+QGGOMFYITP8YqCEpPBf2yHvTXKdV1sj6bU+JJEwQnZwhOzqDeg4H7t1RHAa+eB50/AUikqouru9SB6JOZEFzcy2hPGGMlMWnSpCLX6dGjB3r06FEO0ZSc0L67arKXgztBNWpDeNdX3yExxqood3d3hIeHIzY2FrNmzUJgYGC+dfr27YtZs2ahYcPCT18JDAzEoEGDIJOprjE8ePBg/PzzzzqbNXnSpEnw9/fH+++/r5PydIUTP8b0jJRK1WQtezYCOdkQun4IoVu/N5pRUxCJgLoNINRtABowFrgZCroZCni8A6F5Ox6OxRjTGUEQgMHjQTFRqiGf0xdDqF4xjkgyxqomBweHApO+4lq3bh369OmjTvy2bNmiq9AqNE78GNMj+vcxlFtXAhFhgLsnRIPGQ6im2y9MgpEE8GkFwaeVTstljLE8gqERRP83Hcr5U6BcOR+iGUsgmJjqOyzGWDGtuxyHyORsnZbpYiXFKB97revMnz8f1atXx7BhwwAAS5YsgSAI+Ouvv5Camorc3Fx88cUX6NSpk8Z2UVFRGDp0KE6ePImsrCxMmTIF4eHhcHNzQ3b2y/348ssvcePGDWRnZ6Nbt24ICAhAUFAQ4uLi0K9fP1hZWWHPnj1o1qwZgoODYW1tjTVr1mDXrl0AgI8//hijR49GVFQUBg0ahKZNm+Ly5ctwcHDA+vXr1YmjNmfPnsW8efOgUCjQsGFDLFy4EBKJBAsWLMCxY8cgFovRunVr/O9//8PBgwfxww8/QCQSwdzcHPv27Sthq2vHiR9jekA5OarZOI/tB6TGEIZOgNCiPR+JY4xVWoK1XJX8ff8VlIGLIZo4mz/TGGNa9ezZE7Nnz1YnfgcPHsS2bdswevRomJmZISkpCd27d0fHjh0LvW7x5s2bIZPJEBISgrCwMHTu3Fm9bNq0abCysoJCoUD//v0RFhaGkSNHYu3atdi9ezesra01yrp58yZ++eUXHDp0CESE999/H82bN4eFhQUiIyOxYsUKLF68GGPHjsWRI0fQp08frfuXnZ2NyZMnY9euXXB1dcXEiROxefNm9O3bF8HBwThz5gwMDQ2RmJgIQHXO97Zt2+Do6IjU1NQ3aNmCceLHWClRWjLoVDCQ+Qx48Rx48QL0Igd48eK/+8+B589f3n718RfPAUA17LLf8GJdfoExxio6wa0ehI/HgLauBJ0+AqFdxTq/hTFWsKKOzJWV+vXrIyEhAbGxsUhMTISFhQXs7OwwZ84cXLx4EYIgIDY2FvHx8bCzK/gawxcvXsSIESMAAJ6enqhXr556WV4iqVAoEBcXh/DwcHh6ehYaz6VLl9C5c2cYG6suT9OlSxdcvHgRHTt2RI0aNVC/fn0AQIMGDRAVFVXk/j148AA1a9aEq6srAKBfv37YtGkThg8fDolEgoCAAHTs2BFt27YFoJrFefLkyejevTu6dOlSjBYsGU78GCsFuvc3lIHfA2mpgMwYMDQCDA3/+//fbSMJIDMBDI0gGP33uNgQ+O+2UK8hhDr19b0rjDGmU0LrTqDrf4H2bQa94wPB1kHfITHGKrBu3brh8OHDePr0KXr27Il9+/YhMTERwcHBMDQ0RLNmzZCTk6O1jIKOBj5+/Bhr1qzB4cOHYWlpiUmTJmkMAy0IERW6TPLK5bEMDAyKLEtbeWKxGIcPH8aff/6JgwcPYt26ddi9eze+/fZbXL16FSdOnEDHjh1x7NixfEcl3wQnfoyVACmVoCO7Qb/tAOwdIZr0NQQnZ32HxRhjOheT/hwvFISaliWbaEoQBIgGfQLlnE+h3LICoslzCx2ixRhjPXv2xOeff46kpCTs3bsXBw8ehFwuh6GhIc6dO4cnT55o3b5Zs2bYv38/WrZsibt37+LOnTsAgPT0dMhkMpibmyM+Ph6nTp1C8+bNAQCmpqZ49uxZvqTK19cXkydPxqeffgoiwu+//47ly5eXet/c3NwQFRWFyMhIuLi4YO/evfD19UVGRgaysrLQvn17NG3aFL6+qtmQHz58CG9vb3h7e+P48eOIjo7mxI8xfaC0FCiDlgJh1yE0bQNh8P9BkBrrOyzGGNM5hZIwM+QxHM2M8I1/ySecEmxsIfQdrhryefYYhNadit6IMfZW8vDwQEZGBhwcHGBvb48PPvgAQ4cORZcuXeDl5QU3Nzet2w8ZMgRTpkyBv78/PD090ahRIwCAl5cX6tevj7Zt26JmzZpo0qSJepuBAwdi0KBBsLOzw549e9SPv/POO+jXrx+6desGQDW5S/369Ys1rLMgUqkUS5cuxdixY9WTuwwePBgpKSkYMWIEcnJyQESYPXs2AOCbb75BZGQkiAitWrWCl5dXqeotjEDajmlWAtHR0QAAuVyOhIQEPUdTcXH7aFdU+9D9W6qhnc/SIXw8GsJ7nd6qX7D59aMdt492umqfatWq6SCat8ub9JG/3U1C0JWnWNChJrzsSv4jFymVUC6dBTyKgOjrnyFY25a4jPLC72HtuH20q6ztk5mZqT6XrSyJxWLk5uaWeT2VVUnbp6Dnrbj9Y7lNt3X9+nV89tlnmDBhAn799dd8y+Pj4zF37lwEBARgzpw56tltGNMnUiqhPPwLlN/PBIykEM34HqLWnd+qpI8x9nbq5GYJC6kBfvm7dF9oBZEIoiGfAkollFtXaT13hjHGWNkrl6GeSqUSQUFBmDlzJmxsbDB9+nT4+PjAyclJvc6WLVvQunVr+Pn54datW9i+fTsmTJhQHuExViBKT1UN7bx9DUKT9yAM/gSCjId2MsbeDhKxCL3rWWPjtXjcjc9CXduir1f1OsHOEULvwaBd60B/nYbQvG0ZRMoYY/oxY8YMhIaGajw2atQo9O/fX08RaVcuiV9ERIR63C4AtGjRAqGhoRqJ35MnTzB06FAAqjG5ixcvLo/QGCsQhYdBuXYx8CwNwsD/g9CGj/Ixxt4+XepYYV9YEnb9nYDZ7WqUqgyhXTfQ5T9BOwNBno0gWFjpOErGGNOPBQsW6DuEEimXxC8pKQk2Njbq+zY2NggPD9dYp1atWrh48SK6du2KS5cuISsrC+np6TAzM9NYLyQkBCEhIQCARYsWQS6XA1CNj827zfLj9tEur31IqUTmr9vwbNtaGNg5wGLW9zCs7aHv8PSOXz/acftox+1TeUnFIvSqZ43N1+NxPyELdeSlOOonMoBo6EQo534G5bZVEP3fdP4hjbEKgIdfV05v8ryVS+JXUICvf+gPHjwY69evx+nTp1GvXj1YW1vDwMAg33b+/v7w9/dX3887mbaynlhbXrh9tJPL5Yh/+A+U65cBf1+G0LglaOgEpMqMAW43fv0UgdtHO57cpXLrUscS+8MS8cutBMz0K+VRP0cnCD0GgPZtAq6cA3xa6ThKxlhJiUQi5ObmQizmSf4ri9zcXIhEpZ+ipVyeaRsbG43JWhITE2FlpTnUw9raGgEBAQCA7OxsXLx4sVxmGmIMAJ7f/RvK774C0lMgDBgLwa8r/yLNGGMAjA0N0KOeNbbdSEBEYjbcbKSlKkfo2At05RyU29dA5NEAgpm5jiNljJWEVCpFdnY2cnJyyvQ7j0QiKfIC7G+z4rYPEUEkEkEqLd1nMFBOiZ+rqytiYmLw9OlTWFtb4/z585g4caLGOmlpaTA1NYVIJML+/fvRti2fAM7Kh/LsMSRvXQVYyyH68jsItbRfL4Yxxt4273tY4dc7SfjlVgJmtHEqeoMCCAYGEA2bCOU3U0A7AyGMnqrjKBljJSEIAmSykg/fLikeFaNdebZPuSR+BgYGGDFiBObPnw+lUom2bduiRo0a2LVrF1xdXeHj44OwsDBs374dgiCgXr16GDlyZHmExt5yFBUJ2rYaRu94I3fEZAjGpvoOiTHGKhxjQwP0qGuNHTcT8E9SNmpbl/Kon5MzhK79QAd3gJq0gtComY4jZYwxVphyG9Tr7e0Nb29vjcdenerU19cXvr6+5RUOY6AXL1SXazAxhcXkOUh6odB3SIwxVmG972GF3/476vdl69Id9QMAoWtf0LULUG5dBZG7FwQT/sGNMcbKQ7ldwJ2xioYObAX+fQTR0AkQ8fTijDGmlamRAd6va4ULUc/wMDm71OUIYkOIhk0E0lNAu4N0GCFjjDFtOPFjbyW6dwt07FcIrTtDaNBE3+Ewxlil0N3DGjKxCL/cSix6ZS2EWm4QOn0AOncCdOuqjqJjjDGmDSd+7K1DmRlQrv8BsHWA0G+4vsNhjLFKw0xigG4eVjj/OB2PU99slj6h+0eAgxOUW34GZWXqKELGGGOF4cSPvXVoZyCQnAjRiMkQpGU/mxVjjFUlPetZQyIWsPvvNzzqZ2ikGvKZnKi6vh9jjLEyxYkfe6vQlfOgCychdO0LwbWuvsNhjLFKx1xigK51rHD2URqevOlRP9e6ENr3AJ0OBt37W0cRMsYYKwgnfuytQSlJUG5dAdRyg/D+R/oOhzHGKq1e9axhZCBg9xue6wcAQq9BgK0DlJt+AuWUftIYxhhj2nHix94KRATl5p+BnByIRk6GIC63K5kwxphWK1euxKhRozB1qvYLmkdERKB///7466+/yimywllIxehSxwpnHqUhOu35G5UlSCQQDZ0AxMeCft2mowgZY4y9jhM/9lagM0eBvy9D6DMMgmMNfYfDGGNqfn5+mDFjhtZ1lEoltm3bhkaNGpVTVEXrXc8aYpGA3bd1cNTP4x0Ifl1AJ34DRdzRQXSMMcZex4kfq/IoLhr0SxDg2QhC2676DocxxjR4enrC1FT7RcyDg4PRrFkzmJubl1NURbOUidHJ3RKnI1MRk/5mR/0AQOgzFLCSQ7lxOej5m507yBhjLD9O/FiVRgoFlEFLAbEhRMM+gyDilzxjrHJJSkrCpUuX0LFjR32Hkk/vetYwEATs0cVRP6mxashn3L+gAzzkkzHGdI1PdGJVGgXvBiLvQxgdAMHKRt/hMMZYiW3cuBEDBw6EqBg/XIWEhCAkJAQAsGjRIsjlcgCAWCxW39YlOYCe72Ri/9+xGNfaHY7m0jcrsLU/0m5fRdbxAzD36wyjeg10EmdRyqp9qgpuH+24fbTj9tGuPNuHEz9WZdHDcNChXRCatoaoaWt9h8MYY6Xy4MED/PjjjwCAtLQ0XLt2DSKRCE2bNs23rr+/P/z9/dX3ExISAAByuVx9W9e61DbGr38DgWcjML6ZwxuXR90/Aq6cR/KyuRD970cIEokOotSuLNunKuD20Y7bRztuH+100T7VqlUr1nqc+LEqiXJyoAz6ATCzhDBgnL7DYYyxUluxYoXG7caNGxeY9OmL3NgQHVwtcPxBCvrVt4GtieEblZc35FO5dBbo160Q+o/UUaSMMfZ24xOeWJVE+zYBsU8gGv4ZBBPtkyYwxpg+LVu2DDNnzkR0dDTGjRuHkydP4tixYzh27Ji+Qyu2Pl6qofR7dXCuHwAI9Rq+nOUzPEwnZTLG2NuOj/ixKoduXwOdPAShfXcInhVn6nPGGCvIpEmTir3uJ598UoaRlJ6tiSHa17bE8Qep6FvfBnLjNzvqBwBCn2Ggv69AufFHiP63vFyGfDLGWFXGR/xYlUIZ6VBu/BFwrAHhgyH6Docxxt4afbysQUTYF5akk/IEqQyiYROBpzGgX7fopEzGGHubceLHqhTathpIT4Vo5BQIRvzrMGOMlRd7UyO0rW2BY+EpSMrK1UmZQt0GEPy6gk4cBN2/rZMyGWPsbcWJH6sylBf/AIWehdD9Ywi1XPUdDmOMvXX6etlAQYR9Ybo51w/478LuNnZQbvwRlJOts3IZY+xtw4kfqxLoyUPV0T7XuhA699F3OIwx9lZyNDOCn4s5joanID1HoZMy1UM+42NB+3nIJ2OMlRYnfqzSo38fQ7l0FiCRqoZ4GhjoOyTGGHtrdfewxnMF4c9HaTorU/B4B0Lbbv8N+byls3IZY+xtwokfq9QoJgrKJV8BIgOIpn4DwfbNLx7MGGOs9FysJKhlIcGpSN0lfsB/Qz5tHaDcuJyHfDLGWClw4scqLYp9AuWSmYAgqJI+h+r6Dokxxt56giDAr7Y57iVkITrtue7KlUghGvrfkM99m3VWLmOMvS3K7Tp+169fx4YNG6BUKtG+fXv06tVLY3lCQgJWrFiBjIwMKJVKDBgwAN7e3uUVHqtkKC4ayu9nAkolRAHzITg66Tskxhhj/2njbI4t1+NxKjIVAxva6qxcwaM+hPbdVUM+vZtD8HhHZ2UzxlhVVy5H/JRKJYKCgjBjxgz88MMPOHfuHJ48eaKxzt69e9G8eXN89913mDRpEoKCgsojNFYJ0dMYKL//ClDkqo70Vaup75AYY4y9wsbYEA0cTHA6Mg1KIp2WLfQe/HLIZ3aWTstmjLGqrFwSv4iICDg4OMDe3h5isRgtWrRAaGioxjqCICAzMxMAkJmZCSsrq/IIjVUyFB+rOqcv9zlEU+dBqF5L3yExxhgrQFsXczzNeIE7T3WbnAkSKUTDPgMSn4L2bdJp2YwxVpWVS+KXlJQEGxsb9X0bGxskJSVprNOvXz+cPXsW48aNw8KFCzFixIjyCI1VIpQQpzqnLzsbosnzIDi56DskxhhjhfCtYQapWISTkak6L1uo4wWh3fugU0dAd2/qvHzGGKuKyuUcPypgmIcgCBr3z507Bz8/P3Tv3h3379/HTz/9hCVLlkAk0sxNQ0JCEBISAgBYtGgR5HI5AEAsFqtvs/wqe/so4mORvGw2hOwsWM39CYauHjotv7K3T1nj9tGO20c7bp+3k1QsQouaZjj3KB1jfOwhEev2t2ah9xDQ35eh3Lgcojk/QZDKdFo+Y4xVNeWS+NnY2CAxMVF9PzExMd9QzpMnT2LGjBkAgDp16uDFixdIT0+HhYWFxnr+/v7w9/dX309ISAAAyOVy9W2WX2VuH0pKgPL7GcCzdIimzEWqhQ2g432pzO1THrh9tOP20U5X7VOtWjUdRMPKU1sXc5z8JxUXnzxDa2dznZYtSCQQDfsMysXTQXs3QRg4TqflM8ZYVVMuQz1dXV0RExODp0+fIjc3F1K8EtMAACAASURBVOfPn4ePj4/GOnK5HLduqS7K+uTJE7x48QLm5rrtJFjlQymJquGd6akQTZoDwdld3yExxhgrpvr2xpAbi3G6DIZ7AoDg7gmhfQ/Q6SOgOzfKpA7GGKsqyuWIn4GBAUaMGIH58+dDqVSibdu2qFGjBnbt2gVXV1f4+PhgyJAhWLNmDQ4fPgwAGD9+fL7hoOztQilJqks2pCarkr7auh3eyRhjrGyJBAF+LhbYF5aI5KxcWMl0/7VD6DUIdDMUyk0/QfT1zxAkUp3XwRhjVUG5XcfP29s733X5+vfvr77t5OSEefPmlVc4rIKjtGQol84CUhIh+mwOBLd6+g6JMcZYKfi5mGPP7USceZiGnvWsdV6+IJFANHSCasjnwR0Q+g7XeR2MMVYVlMtQT8ZKgtJSoFwyC0h8CtHE/0Fw99R3SIwxxkqphoUE7jZSnCqj4Z7Af7N8vtcRdPwAKCqyzOphjLHKjBM/VqFQeprqSF9CLEQTZkGoU1/fITHGGHtDbV0sEJmcg8jk7DKrQ+gzFDAxg3LLCpBSUWb1MMZYZcWJH6tQlJt/BuKiIfpkJoS6DfQdDmOMMR14r5YZDATgdGRamdUhmJhB+HAkEHkf9MfvZVYPY4xVVpz4sQqDrv0FXP8LQs8BEDwb6TscxhhjOmIuFcOnuin+iEyFQpn/2r66IjRrA3g2Au3bDEpJLHoDxhh7i3DixyoEys6EcsdaoHotCP499R0OY4wxHWvrYoHkbAVuxGaUWR2CIEA08P8AhQLKnYFlVg9jjFVGnPixCoEO7FDN4Dn4EwjicptsljHGWDnxqW4CUyMRTpXhcE8AEOwcIXT7ELhyHnQjtEzrYoyxyoQTP6Z39OgB6MRBCK07QXCtq+9wGGOMlQFDAxHeq2WOv6LSkfmibCdfETr1BqrVhHL7alB2VpnWxRhjlQUnfkyvSKmAcssKwMwcwgdD9B0OY4yVu5UrV2LUqFGYOnVqgcvPnj2LgIAABAQEYObMmXj48GH5BqhDbWtb4LmCcP5xepnWI4gNIRo8HkiKB/22vUzrYoyxyoITP6ZXdCoYeBQBof8oCMam+g6HMcbKnZ+fH2bMmFHocjs7O8yZMwfff/89+vTpg7Vr15ZjdLpVx0aKamaGZT7cEwAEN08IrTuDQg6CHj8o8/oYY6yi48SP6Q0lJYB+3QJ4vQuhyXv6DocxxvTC09MTpqaF//Dl4eGhXu7u/v/s3Xd803X+wPHXN6N7J520rEIZgiBWpiCFihMFz3VOnOce5/3Ow8MtJ3qH4xznRsXFOQ7FdVxlCKICylSgbOiCJulu0yb5fn5/pBRqoaQlbdryfj4epcl3vvMhzTfv72f1xW7vvKNVappGVq9oNu6rZn+lq+3Pd8FVEBmF/rbM7SeEEJL4iYDR570KHg+Gy29G07RAhyOEEB3eokWLOOmkkwIdxjE5rVcUAEt2lbX5ubTwCLRLrofd21CLv2zz8wkhREcmwyeKgFDrVsLP36NNvRItPinQ4QghRIe3ceNGFi9ezCOPPHLEbXJycsjJyQFg1qxZWK1WAEwmU8PjQLNa4aRuNr7dXcXNp/Vr8xt/6qyplK5ejmv+u8ROPAejNaHJNh2pfDoiKZ/mSfk0T8qnee1ZPpL4iXannDXo770MKd3RJk0JdDhCCNHh7d69m5dffpnp06cTGRl5xO2ys7PJzs5ueG6z2QCwWq0NjzuCU9PCeO6HIlZsyaOfNbTNz6cuuhb10G3YXpyF8Zam/Sk7Wvl0NFI+zZPyaZ6UT/P8UT4pKSk+bSdNPUW7UwveB0cxhitvQTOZAx2OEEJ0aDabjX/84x/cdtttPl/cO7rR3SMJMmos3tH2zT0BtPgktMm/hzU/oNb+0C7nFEKIjkZq/ES7Unt2oHI+Qxs7Ca3PwECHI4QQAffMM8/w66+/UlFRwU033cTFF1+M2+0GYNKkSXz00UdUVlby2muvAWA0Gpk1a1YgQz5mYWYjI1MjWb67nOtOTsBsbPv70Fr2+agflqC/9wqG/ieihYS1+TmFEKIjkcRPtBule9DfeRHCI9F+d3WgwxFCiA7hrrvuanb9TTfdxE033dRO0bSfrN5RfLu7nNUFVYxKO3LzVX/RTCYMV96K/sS9qPnvol16Q5ufUwghOhJp6inajVr6NezM9c7ZF972F3khhBAd15CkcGJDjO3W3BNAS++PdtpZqEVfoHZtbbfzCiFERyCJn2gXqtSO+s9cGDAEbfi4QIcjhBAiwIwGjXE9o/ipoJJyp7vdzqtNvRKiotHnvojyyNx+QojjhyR+ol2oD14DlwvDFUefs08p1U5RCSGECKSs3tG4dVi2u6LdzqmFhWO49AbYsx216PN2O68QQgSaJH6izakNq1E/fYd27iVoCc2PSLdgs4NbFuygolbuwgohRFfXKzaEnjHBLN7Zfs09ATh5DAzORH36Lspe3L7nFkKIAJHET7QpVetEf/clSE5DO2Nqs9tW1Hp4b72NggoXH2yQ+V6EEOJ4kNU7iq12J3llte12Tk3TMFz2B1AK/f2XpaWJEOK4IImfaFNqwQdg3+9t4nmUOfvmb3JQ49IZmhzOl7kl7GnHLwFCiLbj0eVLtTiycT2jMWiweGd5u55XsyainX8ZrFtJ7cpl7XpuIYQIhHabzmHt2rXMmTMHXdeZOHEiU6ZMabT+zTff5JdffgGgrq6OsrIy3nzzzfYKT7QBlbcT9b/5aKeejpYxqNlty5xuPt/iYEyPSP6QmchNC3bw+k/7eSgr9ah9AttTudNNrt3JFlsNubYadpfW4mnld1pLmIl+1lD6WUPJsISQEhWEoQO9VtE1lNS4ybXVNLxviyrq6BETTEb9e6+vJYTwIKPfzqeUYn+Viy02Z/15a7DV7OD183t1qL9l0XHEhZoYmhTOkp1lXD7E2qLPwWqXhw1F1awprCIy2MjlQ+JbdG5t4nmo5TlUvvU86oFnj3qDUgghOrN2Sfx0Xef1119nxowZWCwWpk+fTmZmJqmpqQ3bTJs2reHxV199xc6dO9sjNNFGlK6jz30RwiJ8mrPvk18d1HkUvx9sJSrExO8HW3ntp/2szq/ilNSIdoi4KZdHsav0QJJX/6W50gWAQYMeMcEMS4kgyNi6L7MFFXV8u6ucr7eWAhAeZKCvJZR+1hD6WULpaw0lKth/X8hF11fn0dnhqCXXXtPwvt1f5X3PGjVvf6r+8aHsKq1ldUFVw36pUUHeGxDWEPpZQ+keHYzR4Nv7utrlYVt9UrnF5iTXXkOZ09tHN8io0ScuhEn94qnzKIJNkviJw8vqHc3s7wrYuK+aE5PCj7idrhTbHU7WFFaxpqCKLbYaPMr7mawrGJUWSe+4EJ/PqxmNGC66Fs8/H0Zb/CXa6ef74+UIIUSH1C6J37Zt20hKSiIxMRGA0aNHs2rVqkaJ36G+++47Lr744vYITbQR9e1/YccWtGvvRouIanZbR42bL3NLGNczitToYADOyojl662lvPHzPoYmh2NuZXLlc7xKsb/S5f2ybPd+gd3hcOKqb6IWG2qinzWEM/rE0M8aSrolhBDTsbeU1pUir7zOWzNS/6X5w412DrSMS4k0k2EJJaP+S3mv2BBMPn4hF12bUoqiShe5thq22L21aztLnLh17/r4MBMZ1lDO7RdLhiWE3nEhBB/ynq2q87DVfrBWblV+Jd/Uz6cWbNToawmpf995awbjQk14dMXestqG2sOtNid7ymo5UOndLSqIk1PCybB49+keE4zJoGG1WrHZpN+uOLIRqRGEmgws3lneJPGzV7tYU1jF2sIq1hZVNwz+lR4XzJQBcZyUEk5adDB/+HQH8zc5+OOY5gcR+y1t8MkEnTSCus8/QI3KOuo1S4jOyuVRGDR8vrEnup52SfwcDgcWi6XhucViYevWw0+cWlxczP79+xk0qPmmgaLjUvl7UB+/Cf1PRBs5/qjbf/SLHbeuuHSwtWGZyaBx3ckJPLw4jy9yHUwZYGnmCMdm/iY7n23ejr3aWzNyoJbinH6xZFhDyLCEYg0ztUkzNYOm0T06mO7RwWSne5fVuHS2OQ7WMq4rqmLJrvKG2AbEh3LNsAR6xfp+V7u91bh0Fu8sY9mucupa2Ra2R0ww5/aLbdHd++NBrq2GL7aU8HNhFeX1X4BDTN737Hn94+pr7ryJWnPCg4wMTQ5naLL3S/bhEsnPNjsaEklLmImqOh1n/YLIIAMZ1lBG94gkw+L9O4mQGmrRSsEmA2N6RLJ8dwXXDEvw1uoVVLKmsIo9ZXUAxIYYyUwJ56TkcIYkhxMT0vg9fnqfaL7cUsKVQ+OJD29Zk83Iabdjv/sq1IIP0H5/o99elxCBcuBG3Va7k20OJ1vtTnaXOgFIjgyiW1QQ3SKDSI0Obngsn+FdX7skfocbLetIX6K/++47Ro4cicFw+NqUnJwccnJyAJg1axZWqzdZMJlMDY9FU+1VPnp5KY5//Q1DaBhx9zyC0dp8f4t9FbUs3LaFswcmMrhX47u0k6xWFu6s4t8bHfzu5F7EhgX5Pd5lO+zM+bmYU7rHMK23hROSI0m3hGEyBnbco7RkyKp/rJRiX0Utv+6r5JfCcv67uZh7vt7NFZmpTDsljSA/1Dweja/vn8JyJx+vK2TBxiIq6zz0sYYTH9Xy/zddKVbsLeebHWUM7RbFRUNTGNvb0mHvUrb135fbo7Nkm51/ry3gl6IKwoKMnJZuYXByFAOTIuhlCfdLTXB8PAzudfB5rVtna3ElvxRVsGlfJZHBJgYlRzIwMZLUmBCfb4bI57PwRVavaHK2l3H1x1vRFZgNGgMTQsnqHc2w5HB6xAQ3+56b3C+WL7aU8PmWEq4ZltCic5u690YbdwZqyZeo8WejJR++RZIQHZGuFIUVLrbaa9hm9yZ5O0qcDTdew8yGhpuDAPnldeSV1bEqr7LROAXRwUZvElj/kxrlTQoTI8wd9vorWqZdEj+LxYLdbm94brfbiY2NPey2K1as4LrrrjvisbKzs8nOzm54fqD5kDQlal57lI9yu9GfeRDsxRj+72+UYICjnPPlH4tQSnF+n4jDxnfF4Bju2F3CPxdv4dYRyX6Nt6iijke/3kV6XDBPTB5IRakDcFJa4vTrefzBBJwYCyfGRnFOejhv/LSPt1buZdGWfdw+Mpl+1tA2PX9z7x+lFL/sr2HBFgcr8yoBGN09ksn94uhn9T05+K3KunhytpfyxZZS/vrFZhLCTZydEcvp6TEd7q5kW/19lTvd/HdbKV/llmKvcZMcaeaGzAQm9I4mzHygDJyUOtruPZtkhqS0YCamBR9c6KnCbq868k6/4a/ySUlpWRM+0bkMTAjlzL4xmI0aw5LDOSEhrFHz5KNJjAhidPdIFm4r5ZLBlkP+RnyjnXcZ6sel6B++gfGOB1oavhDtQleK4ioX2xxOttm9P9sdTqpc3tYYQUaN9LgQzugbQ9+4EPpYQkmONB920CS3rthX6SK/vJa88jryy+soKK9jZV4lZYfMp2wyQEJ4ENYwE3GhJuIO/A5t/Nwc4Jvm4ujaJfFLT0+nsLCQ/fv3ExcXx4oVK7jjjjuabFdQUEBVVRUZGRntEZbwI6UU6v1XYMsGtOvuRuvd76j77KusI2d7KZP6xJAQcfhmOalRwZzTL5YFm0s4q6//mv3VeXSeWJYPGtw7thvBJgMVfjly24sKNnLX6BTG9ojixZVF3Pvf3UzuH8vlQ+L90u/QV3UenW93lfP5lhJ2ltQSGWTggoEWzsqIwRp27CPjRQQZmTLAwuR+cazMr+TzzQ7eXFPM++ttTOgdzbn9Yhv6hLaWW1fsLq0l11bDVruzoRljS8VGltA9XCPDGtKigVGOZFeJkwVbSvi2vqnskKQwbhmRxLCUcBn5VXRZBk3j5uFJx3SMKQPiWL67gv9tK+P8AXEt2leLjEY752LUR2+ifl2DNvCkY4pFiNZSSlFe66GgvI78ijoKK1zepKyijsKKuoaaPJMBesaEMK5nFH0sIfSJCyGtBdcgk0FrqN0b/pt1FbUe8svrGpLCokoXjmo3vxbX4Khx4z7MND2RwcbGCWF9UtinwkCIXktiuLlFN3Pak1KKKpdOnUehlEJXoBQovI+9zxU69cvVIcvrtzl4LO8y7xMa+sGrhn8OXabIMEdwbN9mfNcuiZ/RaOTaa69l5syZ6LpOVlYWaWlpzJs3j/T0dDIzMwFYvnw5o0ePliG/OyG15EvUt1+jnfk7DCOzjr4DMG+DHYOmcdGg5vvvXTLYypKd5bz20z5mZnf3y/vj1dX72FFSy4zTUkmM8H8T0vZwcrcInju3F2+vKeazzSWszKvk1hFJzY6I5w/2ahdfby3lv1tLKav10D06iFtHJHFaz6g2+UA3GjRGpUUyKi2SHQ4nn28pIWd7GV9tLeWk5HAm94vlJB8TIlu1q2G0y1xbDdscB5vCRAcbiWxlTeLG/TWUOd3Awf52vx0Y5Wg8umJ1fiULtpSwYV81QUaNrF7eBLd7THtdEoTo3PpaQhmUEMpnmx2c0y+2xU2gtQmTUUu/Rv/3GxjufwbNeOTPhJIaN0YNokLabWYs0cVUuzwUlLsoqPDWtBVU1DU8PlCDB95RmZMig0iJNDM0KYyUqCDS40LoGRPcZrVskcFG+seH0j++aYsipRQVdTqOaheOGvfBn+qDj/eU1lLidNcnRPsa9o0JMZIYYSYh3ExiRNAhj83Eh5v90m3Boysq6zxU1Hoor/X+rqg7+Lhh2YHHdR4qaz2tnp7rWF1xsuKi/u0zgr2mDtcBrxMpKCgApKnn0bRl+ahN67xNPAedjOHW+9AMR//yXFBex62f7+CcfrFcf3LiUbf/emsJ/1q5jz+PTWFM92MbcW3RjjKe/b6QC0+wcOVQbx/Ezv7++WVfNc//WEhBhYvT06O5ZliCX+dms1qtrNi8lwVbSvhudzm6gsxuEUzuH8uJiWHtfrOm1Olm4dZSvswtocTpoVtUEOf2iyWrVzShZu9F0OnW2X5gzsX6kVodNd7kzGTQSI8LbhgxtZ81hIRwc6tfh8ViYeOuwob5HXPtzkYjbFrr52zMqJ+q49ARNqvqPHyzo4wvtpRQVOnCGuZt0jqpT0yrE9GORpp6Bs7xeI1cmVfBzKX5/HF0Mqf1ivZpn0PLR/20Av2lWWhX3ILhtDMPu32NS+eWBTsIMxv45zm9unz/p678/lH1/eMKKuqodevUehR1Hm/NT51bUXvgcTPLNIOROpcbT0MtkMKj1/+uf64r0PVDnysObWSiAfHhJlIigxoGX0mJDCIlKoiE8M7Zx86jK8pqPdSZwsjNt7Gvso59lS72VbnYX+miuMrVKNkyaGAJNXmTwYggooKNB8vdo3Ad8vjActeh/zf1P4erjTzAZNCIDDYSVX+zNyrYSGTQwcchJgOa5o1FwzsmSZPHGhjQGrY7cOPZu83Bc2mHLD+4rPFzgIzUBELcvnefOBxfr49ym0ocE7WvAP2lJyApFcP19/iU9AF8sMGG2aBx4UDfRus8PT2Gr3JLefPn/WSmRLS6ZmlXiZN/rSxicGIYl53YdQabOCExjGfO7sUHG2zM3+Tgp4Iqbh6eyPDUyFYf89B5DL9flM/GwgpCTQbOzojlnH6xJEcGrqY0JsTExYOtTB1o4bs93uamL6/axztrixmWEk5eeR27S2sbml4kRZgZlBhGhsU7T12vWP/eJdU0jeT6i/X4+i+aB+bUO5B45tpq+G6Pt0GxUYOesSEkR5pZnV+F060zID6Uq4bGMzItslNe4IXoKDK7RdAtKohPNzsY1zOq5Td0ho2CjBNQn76LOmUsWljTVhT/3mirr9mAJTvLmJge46foRVsrqXGTa/dOR7PV4WSbvYbKuuab+QcZNYKNGkFGA0Gm+t9GjSCjRkSQkbCQINwuDUN9YmA0aA0JgfHAbwMH19f/Dg8yklKf4CVFdNxmkK1lNGjEhZqwWqNIMtc1We/RFfZqN/uq6hPCSm9CuK/KxdrCKqrqPASZDAQZNMxGjWCjAXN9uYeaDEQFH/x/OPB/cmC7iGADkUFGokJM9YmdgahgEyEmrcO1LLTGhGKzHVvi5ytJ/ESrqeoq9OcfA4OG4bYZaKFhPu23p6yWb3eVM3VgHDE+NIED74fH9ZkJzMjZy6ebHVw8qOVJW1Wdh1nL8gkPMvKnMSld7st1sMnA1SclMLp7JM/9UMTMpfmM6xHF9ZkJRB+lKZJSiv1VLu8UEnZvU8hD5zFMiwnh+pMTmJge3eIBE9qS2agxvlc0p/WMYovNyYItDjbsq6ZHTDAXnmCpr9ELOerrbwtBRkOTZjIlNe6GGsEttho27qtmZFoEk/vF0cci01YI4Q8GTWPKgDhe+LGIDUeZEP5wNE3DcPF16I/9EfXVR2i/u7rR+vzyOj7b7CCrVxR7y+r4YIONcT2jZGCLDqja5WkYACXX7mSrvQZbtbflh0HzThs0pnsUfS0hdI8JJsR0MJEIrk/yzIajJwpduUa0LRkNGgkRZhIizAw+euMv4QeS+IlWUboH/dW/Q3EhhrsfQYv3vUP+B+ttBJsMTG1hx/vBieGMSovko412JvaOxtKCAUSUUjz3QyH7Kl08lt3d54SzM+prCWX2mT355Fc7/95oY21RFTdkJjK2R2TDxevAxfDApPFbbDWUOr0jeB0YEeyc+om/M6yh9O+e1Ghk3o5G07T6JKtboENpVmyoiRFpkYxIa31NrBDi6Mb3iuKddcXM3+RoVb9nrUcftFFZqJxPUePOaHSNe/2nfZgN3httu0treXDRXv67rZRz+7Xsmib8r6iijjWFVQ1JXl5ZXcMgGkkRZgbEh9LXEkqGJaRRk3shjhdd99uvaFPq47dg489oV9yC1m+wz/vtLHHy3Z4KLh5kaVWH+GuGxbM6v5K31xZz92jf+/t8trmE7/dWMu2keE5I8K1msjMzGzUuGWxlVFokz/1QyOzvCliyMxxLmIktNid7yw42g0yJDGJocri3D5ollJ6xwU06V3e0ZhFCCNGcIKO3Wfr7623sKauleytGANamXoX6aQXq47fQbroXgFV5lfxUUMW1wxKIDTURE2JkcGIY/95oZ2LvmIY+xqL9bbXX8Nf/7aHWo4gONtLXEsKpPaLIqB/tUgbhEUISP9EK+nffoBbOR8s654gd34/kvfU2ws0Gzu/fujujiRFBnD8gjo9+sXN2RqxP89dt2l/NW2v2MzItgiktrGXs7LrHBDNrUg8+31LCu+uKMRs1MiyhjEqLoJ/Ve+ezqwwgIoQQhzq7bwwf/2Ln000Obh/Z8nlgtVgL2hkXoBa8j9r6K67e/Xntp32kRgVxdoZ3LmJN07hyaDx//u9uPt/i4KJWdEMQx66woo5HF+cRE2rigaxUukUGyQ1LIQ7Dp1tTX375JeXl5W0di+gE1LZfUe+8AAOGoF1yfYv23WqvYWVeJecPiDumCbgvPMFCbKiJ11bvQz/KoLSlTjdPLi8gPtzMHSOTj8sLgdGgcf6AON69KIN3LuzLgxPS+P2J8QxLiZCkTwjRZUWFmJjYO5olO8spqR/Rt6W0M6ZCjAV93mvM/9VOUaWLGzITMRsPXkv6WUMZnhrBf351UHHIpNeifZQ63Ty0aC8KeCgrjdSo4OPyWi+EL3xK/DZs2MCtt97KrFmzWLFiBS6Xq63jEh2Qshejv/g4xMVj+MOfm53f6HDeW2cjMsjA5P6xxxRHqNnAVUPjybU7WbrzyDckPLpi9vICKus8/GVcN79Ob9AZmY0dbyQrIYRoS+f1j8OjK77MLWnV/lpwCNoFV2ErKuajDcWMTItgaHLTPoOXn2il2qXzn187bl/orqjGpfPo4jwcNW5mjE8lJapzzssrRHvxKfG79957efHFFxk6dChffPEFN954Iy+99BK//vprW8cnOgjlrPGO4Ol2YbjtfrTwlg1Osam4mp8Lq7hgoMUvo0KO7+UdhevttcXUuA4/DPP7622s31fNH05JpFesjJgohBDHm5SoIIanRvBVbglOd/ND9h+JNuI03h58Ccrj4ZpBh5+2oWdsCON6RrFgS0nDfKGibbl1xd+X57OjxMmfT+3mU9cPIY53PvdCjoyM5Mwzz2TmzJk89NBDbN++nYcffphbb72VTz75BKfT2ZZxigBSuo4+5xnI343hhv9DS05t8THeW2cjOsTI2f2OrbbvAIOmcf3JiThq3Hz8S9M7rKvzK/nwFzvZ6dFky/xKQogO7MUXX+T666/nnnvuOex6pRRvvPEGt99+O3/605/YsWNHO0fYuU0dEEdFnc4328tatf8vxU6WR/Rhyp4lJHz3+RG3+/2JVjy64sONMqx/W1NK8a+VRfVz1iZxSmpEoEMSolNo0fBTGzZs4MUXX+Shhx4iOjqa2267jdtuu42dO3fyt7/9ra1iFAGmFnwAP3+PduE0tMEnt3j/DfuqWL+vmgtPsBDix6GT+8eHclrPKOZvcrCv8uDEoPsq63h6RQG9YoO5MVMmhhFCdGzjx4/nvvvuO+L6NWvWUFRUxD//+U9uvPFGXnvttXaMrvPrHx9KP2sIn2124NGb7xf+Wx5d8crqfcSHmZhqcaK+/hhVcvjmnMmRQZzeJ4aF20obXZOE/72/wUbO9jIuGWxhUh+5uSuEr3wa1fPtt99mxYoVhIWFMW7cOGbPnk1c3MHREfv27cs111zTZkGKwNFXLUd9/gHa6Ilop5/f4v2VUry3zkZcqIkz+/r/w/mqk+L5YW8Fb64p5t6x3XB5dJ5cVoBScO/YbjJHjxCiwxs4cCD79+8/4vrVq1czbtw4NE0jIyODqqoqSkpKiI31TwuKrk7TvANcPbmsgJV5lYzq7ntXha+3lrK7tJZ7x6YQGnol+rrvUf+Zi3btXYfd/uJBFhbtKOP99TbuasGUQ23N5VHsKHHi8ijcujr4W/f+brTss8lGzAAAIABJREFUN+sMGlw5MrJlNQVt6L9bS5m3wdui5/eDZRRVIVrCp8TP5XLxpz/9iT59+hz+ICYTs2bN8mtgIvDU3p2oN5+B9P7e+fpaMTDI2qJqfi2u4Q+nJBJk9P9lwxpm5ncnWHhvvY2N+6pZtrucbQ4n943rRnKkdPIWQnR+DocDq/XgF1yLxYLD4ZDErwVGpkaSGGHmP5scPid+ZU43764v5sSkMEalRaJpUWgTz0P99xPUxHPRejT9TmQJM3NORizzNzm4YKCF7jEtnz/Q36pdHh5elMdmW02L9jNqYDJouHTFz0VOZk7odkwjcvvDyrwKXlpVxMkp4dw8PEkGLBOihXxK/KZOnUpQUOMv0ZWVldTV1TXU/HXr1s3/0YmAUa469NdmQ2g4hlumo5nNLT+GUry7rpj4MBOnp0e3QZReUwbE8b9tpTy5PJ8yp4epA+IYkdaywWeEEKKjUoeZtuZIX3hzcnLIyckBYNasWQ0Jo8lkapQ8Ho8uy3Tx9JIdFNYFMTglqtG6w5XPG99sw+lW/Dm7P/GWMAD0q27G9v0ijJ+8RexjLx72/+GGsdEs3L6aDzeX8/i5A9ruBfmgqs7N/fN/Zau9hnvGp9PTEorZYMBk1DAbDJhN9b+NGmajAbNBw2T0PjfUv7af9pbyx/m/8I/v9/HUlEEEBaglzcbCcv7xXSH9EiJ4YspgQv0wUJy/yN9X86R8mtee5eNT4vf3v/+dm2++mYiIg51nHQ4HL730kvTt66LU/HegYA+GOx5Ei2rdXeVV+ZVstTu5dUQS5jao7Tsg2GRg2rAE/r68gBMSQrlyaHybnUsIIdqbxWLBZjs4YIjdbj9ibV92djbZ2dkNzw/sZ7VaGx3jeDQy0UREkIG3ftjBX8Y1HqTst+Wzze7ks41FTO4fS6SqxmarPrjxeZfheudFbP9bgDZs9GHPdX7/WN5bb2PF5r1kBGi0yRqXziOL97LZVsOfTk1hTDcz8JsRRz31Py7vmsONR9ojFP56el8e/m8uD3y+gT+OSWlICttLfnkd9y7cTWyIkb+cmkRVWQlV7RpB8+Tvq3lSPs3zR/mkpPjWtNynb+MFBQV079690bLu3buTn5/f8shEh6c2r0f971O08We1ajAXAF0p3ltvIynCzITebVfbd8CY7pH8ZVw3po9LxWiQph9CiK4jMzOTb7/9FqUUubm5hIWFSTPPVggxGTizbyw/7K2ksOLIg6/oyjugS1SIkUsP04dMO/V0SOmO/tGbqCPMazy5fyzRwUbeWVfst/hbwunWeWyJN+n74+gUxnSPOvpOzZjUP4GrhsazbHcFb69p39dUUuPm4cV7MQAPTUgjJsSnOgshxGH4lPhFRUVRVFTUaFlRURGRkdKcrqtR1VXeqRvik9EubP2APT/srWBnSS2XDLZiaodETNM0RqVFEhng/gdCCNFSzzzzDDNmzKCgoICbbrqJRYsWsXDhQhYuXAjASSedREJCAnfccQcvv/wy119/fYAj7rzO6ReL0aDx6SbHEbdZsrOcLbYarhoaT3hQ02uKZjRiuPg6KC5CLVpw2GOEmY1cOMjCuqJq1he1b91UrVvnsSV5/Fpcw92jUxjb89iSvgMuGBjHWX1j+M8mB19sKfHLMY+m2uXh0SV7Ka1xc39WqvTdF+IY+XTbJCsri9mzZ3PppZeSmJhIUVER8+bNY8KECW0dn2hn6v1XoNSB4d4n0IJbN+m5rdrF22uL6RYVxGl+uuAIIURXddddhx8h8gBN0yTZ85O4UBPje0XxzY4yLhsST9RvbhZWuzy8tWY/GZaQZluraCecBCeeglrwASrzVDRLQpNtzuwbw6ebHMxdW8yTZ4S1y0AktW6dx5bm8cv+au4clcw4P16DNU3jhkzv/Lmvrt5HXJiJUW3Yn96tK55YVsDOklr+eloqfS0yQbsQx8qnGr8pU6YwduxY5s6dy/Tp03nnnXcYO3YsU6ZMaev4RDtSP32H+mEx2tkXo/Xu16pj7Cxx8uevd1Na4+HW4UnS7FIIIUSHcn7/OOo8iq9zm9Zazdtgp8zp4cZTEo/aj81w2U0A6HNfOOwAPEFGA5cOtpJrd7Iyr9I/wTej1q3zt6V5bCiq5o6RyYzv5f9uFkaDxj1jUuhrCeGp7wrYXNyykUJ9pZTi+R8KWVtYxa0jksjsJhO0C+EPPtX4GQwGzjvvPM4777y2jkcEiCp1oM99EXr0QTvn4lYd4+eCSp5YVkC42cDjk7rTK7Z1NYZCCCFEW+keE8zJKeF8nlvClIFxDVMN5ZXVsmCzg4np0T7VLmmWeLSpV6E+eAX14xK0kVlNtpnQO5pPfnXw7jobmd0i2uxmaJ1H52/f5rOuqJo7RiWT1YZ964NNBmaMT+Xehbt5bGkeT0zqQbco/zbBfGedjcU7y7nsRCvZ6TJBuxD+4vNQi263mz179rBx48ZGP6LzU0qhv/VPcNViuP6PaKaWd5xeuK2UR5fkkRxp5u9n9pCkTwghRIc1ZUAcZU4PS3eWA97r4Ks/7SfEZGjRyNBa1lmQ3h817zVURVmT9UaDxuVDrOwuq2XZ7nK/xX+oOo/O40vzWVdYxW0jk9plQLXoEBMPZqWhAY8s3kup83DjgbZcQXkdTyzL56Nf7JzRJ4aLB1n8clwhhJdP3/A3b97MU089hcvloqamhtDQUJxOJxaLheeff76tYxRtTC39Cjb+jHbZH9CSUo++wyF0pXh3nY2PfrEzLDmc/xubQlgHmltHCCHaw+eff86gQYPo2bMnubm5PP300xiNRu644w4yMjICHZ74jcGJYfSKDWb+Jm8N37IdDtYWVnH9yQktGjVSMxgxXHUb+iN3oT54De2Ge5psM7p7JL1jg3l/vY0x3aMwG/1X6+fy6Mz6Np+fC6u4bURSu9aOJUcGMWN8KjNy9vDYkjwey+5OSCvn+Ct1upm3wcZ/t5ZiNmr8/kQrF51gkQnahfAzn/5C33rrLc477zzmzJlDaGgoc+bM4Xe/+x2TJk1q6/hEG1NF+agP34ATTkIbf3aL9q3z6Dz1XUHDnbkZ41Ml6RNCHJe++OILEhK8A3y8//77nHvuuVxwwQW8+eabgQ1MHJamaUwZEEdeeR0/7K3gn9/uoHt0EGdltHyaDC2lO9rZF6FWLkWtX9VkvUHTuGJIPEWVLnK2l/ojfMCb9D2xLJ+fCqq4ZXgSp/dp/yaR/ayh/OnUFLY7nPxjeT4evWlfx+bUuHQ+2GDjD5/u4OutpUzqE8NL56Vz6WCrjBEgRBvweR6/s89unBRMmTKFL774wucTrV27ljvvvJPbb7+d+fPnH3abFStWcPfdd/PHP/6RZ5991udji9ZRHg/6G0+DKQjDtDtadGetvNbDg9/sZdnuCq4eGs/NwxPlQ1oIcdyqrq4mLCyMmpoadu3axVlnncWECRMoKCgIdGjiCE7tEYUlzMTTKwopLK/lhszEVk8/pJ19oXduv3f/hXJWN1k/LCWcgfGhzNtop9atH2vouDyKJ5cXsCq/iptOSeSMvoHrBzciNZIbMhNZlV/FK6v3HXagm9/y6Iqvt5Zw82fbeX+9jZOSw3ju3F7cNDyJ2FCZp0+ItuLTX9eBi1l4eDgxMTHk5eURERGB0+n06SS6rvP6668zY8YMLBYL06dPJzMzk9TUg80KCwsLmT9/Po8++igRERGUlTVtKy/8S335IezMRbvxz2gxvrejL6yo45HFeymucvOnMf6bI0gIITori8XCli1b2Lt3LwMGDMBgMFBdXY3B0Lqmb6LtmQwa5/WPZc7PxUzoa+XEpPBWH0szmb1NPp+4F/XJXLTL/tB4vaZx5dB4pv9vD19sKeGCE1rfd82tK/6+PJ+VeZXcmJnYqlpKfzs7I5biKhef/OogPtzMhUd4fUopVuZV8vbaYvLK6xgQH8q94+IZEB/WzhELcXzyKfEbMWIEa9as4dRTT2XChAk8/PDDGI1GRo0a5dNJtm3bRlJSEomJiQCMHj2aVatWNUr8vvnmG8444wwiIrxD9kZHt33n5OOZ2rkV9fkHaCNOw3DKqT7vt7m4hplL81BK8cjENAYmyIe1EEJcccUVPPXUU5hMJu65x9vP6+eff6ZPnz4Bjkw054w+sdir3Vw7phc4K47pWFp6f7QJ56IWfY4aPg6tz4BG6wcmhHFySjgf/2rnjL4xh50c/mjcuuIfy/P5Ma+SGzITOKdf4JO+A64cGo+t2s3ctcVYw0xNppPYXFzDm2v2s6m4hm5RQdw3rhvDUyOkH58Q7cinxG/atGkNjydPnkzfvn2pqalhyJAhPp3E4XBgsRy8+2OxWNi6dWujbQ40h7n//vvRdZ2LLrqIoUOH+nR80TKqthb99acgOq7JXcnmrNhTztMrCokL9Y7mleLn4ZuFEKKzGjZsGC+//HKjZSNHjmTkyJEBikj4ItRs4LqTE7FGBGM7xsQPQJtyBWrND+hvP4/h/mfQzOZG668YEs/dX+1i/iYHlw+Jx6MrnG6dapdOjVunxlX/2OWp/31wWbVLZ09ZLZuKa7j+5ATO7Rd3zPH6k0HTuGNkEo4aN8/94P2ucGJSOPnldcxdu5/v91YSG2Lk5uGJnJ4eI91DhAiAoyZ+uq5z55138tRTT2Gu/wDr379/i05yuPbev73Do+s6hYWFPPjggzgcDh544AFmz55NeHjjphc5OTnk5OQAMGvWLKxWq/eFmEwNj0VTh5ZP+atPUbMvn9iH/0lQ955H3VcpxQdr8nlhWQGDkiN5/NyBxIaZj7pfZyLvn+ZJ+TRPyqd5x0P5HOgCERMTg9Pp5LPPPsNgMDB58mRMrZgiR3ROWkgohitvQX/2YdSXH6Kdf1mj9b3jQji1RyQf/WLns80OnG7fBkMJMmqEmgyEBRm4MTOxQ9X0HcpsNDB9XDemL9zN49/mMyotksU7ywgyGvj9iVbO7x9HqFmaPwsRKEe9GhkMBgwGAy6XqyHxaymLxYLdbm94brfbiY1t/KEVFxdHRkYGJpOJhIQEUlJSKCwsbNJMJjs7m+zs7IbnNpsNAKvV2vBYNHWgfNTGn9G//Agt+3zKU3rCUcrMoyteXb2Pr7aWMqZ7JHeNTsZTXYatad/1Tk3eP82T8mmelE/z/FU+KSkpfoimbTz77LPcfffdxMTE8Pbbb1NYWIjZbOaVV17h9ttvD3R4oh1pg05GGzke9dVHqMwxaN16NFo/7aQEIoOMBBk1wsxGQs0G74/JQJjZ+9OwzGwk1GTw6xQQbS0iyMgDWWn8+b+7WbKzjDP7xnDJICsxMmiLEAHn01/h2WefzdNPP83UqVOJi4trVFt3oN9ec9LT0yksLGT//v3ExcWxYsUK7rjjjkbbDB8+nOXLlzN+/HjKy8spLCz06djCd6qyHP3Nf0JyGtoFVza7bZ1HZ4ejlg832lhdUMUFA+O4cmg8BmmLL4QQTRQXF5OSkoJSilWrVjF79myCgoK47bbbAh2aCADt4uu9N1rfeg7DX55AMxzszxcfbuam4UkBjK7txYebeeqsnrg8ioSIrtVCSIjOzKfE74033gBg/fr1TdbNmzfvqPsbjUauvfZaZs6cia7rZGVlkZaWxrx580hPTyczM5MhQ4awbt067r77bgwGA1dccQWRkZEtfDniSJRSqHdfgspyDHfcj2YOarSuqNLFFlsNuXYnubYadpY4cetg0OCmUzrGqGFCCNFRmc1mampqyMvLw2KxEBUVhcfjweVyBTo0EQBaZBTapTegXpuNWvwl2sTJgQ6p3cm0DEJ0PD79VfqS3B3NsGHDGDZsWKNll1xyScNjTdO4+uqrufrqq4/5XKIp57L/oVYvR5t6JdVJPdlaWEWuraYh2Suv9QAQbNToawnhvP5xZFhD6W8NlQ9vIYQ4ijFjxvDII49QU1PDmWeeCcDOnTsbJnUXxx9t+DjUj0tR/5mLGjoCzSLvBSFEYMk3+uNAcX4RX32yiNzM68n1DCL/w60c6E6eGhXEKd0iyLCG0M8aSvfoYBlpSwghWmjatGmsW7cOo9HIoEGDgIM3NMXxSdM0DJffjP7gbehzX8Bw50MydYEQIqB8SvweeOCBI35YPfzww34NSPiPR1d8kVvCO6tt1PY6lyizRkaEmdN6RpFhDaWPJYSIVswjJIQQoqkhQ4Zgs9nIzc0lLi6O9PT0QIckAkyzxKNdcCXq/VdQPy5BG5kV6JCEEMcxnxK/CRMmNHpeWlrK4sWLGTt2bJsEJY7dnrJanv+hkC02J8McW7llcDTW8Vlyt1EIIdpASUkJzzzzDFu3biUiIoKKigoyMjK48847iYvrWPOtifaljT8LtfJb1LzXUCcMQ4uMPvpOQgjRBnxK/MaPH99k2ciRI3nxxRe58MIL/R2TOAZuXfHJL3bmbbQTajZwZ/EixhWuJnHGh9jLj31yWiGEEE29+uqr9OjRg+nTpxMSEoLT6eT999/n1Vdf5d577w10eCKANIMRw1W3oT96F+qD19BuuCfQIQkhjlOtnkUzLi6O3bt3+zMWcYy22Z3c89Uu3l1vY2RaBM+lFXPaL19jOO9ytKDgQIcnhBBd1pYtW7jqqqsICQkBICQkhCuuuILc3NwARyY6Ai2lO9rZF6NWLkWtXxXocIQQxymfavwWLVrU6HldXR0//vgjGRkZbRKUaJlat84HG2zM3+QgOsTEfeO6MTwlDP3BxyClO9qo8YEOUQghurTw8HDy8vLo2bNnw7KCggLCwsICF5ToULSzfodavRz93X9hyDgBLUTeG0KI9uVT4rds2bJGz4ODg+nXrx/nnHNOmwTV1Sml/NbX7pf91Tz/QyEFFS6y06O5ZlgCEUFG9G+/hn35GG79a6OJY4UQQvjfeeedx6OPPsqECROIj4+nuLiYJUuWNJq2SBzfNJPZ2+TziXtRn8xFu+wPgQ5JCHGc8Snxe/DBB9s6juPGR7/Y+XCjjd6x3ukTDkyjYAkzt+g41S4Pb68p5qutpSRGmHlkYhpDksIBULW1qM8+gD4DYMjwtngZQgghDpGdnU1SUhLLly9nz549xMbGctttt7F58+ZAhyY6EC29P9qEc1GLPkeNmoDWq2+gQxJCHEd8SvyWLl1Kz5496dGjR8OyXbt2sWfPHsaNG9dmwXU1Hl3x+ZYS4kLNeJRiwZYS3Ju8M+pZQk1kHJII9okLIdh0+C6YPxdU8sKPRdir3UzuH8sVQ+IJOWRb9c1nUObA8Ic/yyieQgjRTgYNGtQwhx+Ay+Xib3/7m9T6iUa0KZd7m3x+8AqGvzwp12khRLvxKfGbN28eTz75ZKNlVquVJ598UhK/Fvi5oIqSGjc3jevGyLRIXB6dnSW1bLHVkGt3kmur4fu93pE3DRr0jAkmwxrqrRm0hBAVbOSNn/ezeGc5qVFBzJrUg/7xoY3OoaoqUF9/AkOGo/UdGIiXKYQQogXWrl3LnDlz0HWdiRMnMmXKlEbrbTYbL7zwAlVVVei6zmWXXcawYcMCFK04VlpIGNrUq1BvPitz+wkh2pVPiV9NTU2TDuphYWFUVVW1SVBdVc6OUqJDjGR2iwDAbDTU1/IdTN7KnG5ybc76ZLCGb3eV8/XWUgA0vAnhxYMsXDzIgtnYtEZQffkROKsxTL2yXV6TEEKI1tN1nddff50ZM2ZgsViYPn06mZmZpKamNmzz8ccfM2rUKCZNmkReXh6PP/64JH6dnDYqC7XkS9THb6GGjkQLCT36TkIIcYx8SvxSU1P54YcfGD16dMOylStXNrowieaVOt2syqtkcv84TIYjN+uIDjFxSmoEp6R6k0NdKfLK68i11ZBfXse4nlH0ig057L7KUYxa9DnayCy0bj0Ou40QQgj/2bhx4xHXud3uo+6/bds2kpKSSExMBGD06NGsWrWq0fVV0zSqq6sBqK6uJjY29hijFoGmGQwYLr0BfdafUV99hCY3a4UQ7cCnxO/yyy/n8ccfZ8WKFSQlJVFUVMSGDRuYPn16W8fXZSzZWYZHwcT06BbtZ9A0ukcH0z366PPwqc/eA0A7//JWxSiEEKJl/vWvfzW73mq1Nrve4XBgsVganlssFrZu3dpom4suuojHHnuMr7/+mtraWu6///7WByw6DC29P9rILNTC+ahTT0eLTwp0SEKILs6nxK9///7Mnj2b5cuXY7PZ6NOnD9OmTTvqBU14KaXI2V5GP2uITwlcq85RsAe1YjFa9mQ0S3ybnEMIIURjL7zwwjHtr5Rqsuy3g3189913jB8/nsmTJ5Obm8tzzz3H7NmzMRiaNvfPyckhJycHgFmzZjVcp00mk1yzmxGo8vHccBf2tT9g/vRdYv7yeLuf31fy/mmelE/zpHya157l41Pi53K5iImJadTh3O1243K5MJtbNg3B8SjX7mRvWR23jmi7u3n6f+ZCSAja2Re12TmEEEL4l8ViwW63Nzy32+1NmnIuWrSI++67D4CMjAxcLhcVFRVERzdtQZKdnU12dnbDc5vNBnhrHg88Fk0Frnw0OOtCav8zl+Jl36ANGBKAGI5O3j/Nk/JpnpRP8/xRPikpKT5td/j5An7jscceY8eOHY2W7dixg5kzZ7Y8suNQzvZSgo0ap/aIbJPjq22bYO2PaGdcgBYR1SbnEEII4X/p6ekUFhayf/9+3G43K1asIDMzs9E2Vqu1oS9hXl4eLpeLqCj5rO8qtNPPh/gk9HmvoTyeQIcjhOjCfKrx27NnD337Np5ktE+fPuzevbtNgupKnG6dZbsqGNMjkjCz0e/HV0qhf/wWRMeiZZ/n9+MLIYRoO0ajkWuvvZaZM2ei6zpZWVmkpaUxb9480tPTyczM5KqrruLll1/miy++AOCWW26Rud+6EM0chOGia9Ff/Bvq26/Rss4JdEhCiC7Kp8QvLCyMsrIyYmJiGpaVlZURHNw2/dW6khV7Kqhx62Snxxx949ZYvxq2/Yp2+c1owYcf7VMIIUTHNWzYsCbTMxw66XtqaiqPPvpoe4cl2tPQETBgCGr+u6hTxkrrHSFEm/CpqeeIESN49tln2bNnD7W1tezZs4fnn3+ekSNHtnV8nV7O9lJSIs0MjPf/HD1K96D/521ISEE79XS/H18IIYQQbU/TNAyXXA/O6oYRuoUQwt98qvG79NJLefvtt7nvvvtwuVwEBQWRlZXFpZde2tbxdWoF5XX8sr+GK4fGt0mzHPXDUsjfjeEPf0Yz+fRfKYQQQogOSOvWA+20s1BLvkKNOxMttWegQxJCdDE+1fgFBQVx/fXXM3fuXF599VUee+wxTCYTd955Z1vH16l9s6MMgwZZvfzfZEO56lCfvgs9+sDJY/x+fCGEEEK0L+38yyAsHP2DVw871YcQQhwLn6uJysvLWb58OUuXLmXXrl0MGDCAadOmtWFonZtHVyzaUcaw5HAsYf6f8kIt+QocxRim3SGd/IUQQoguQAuPRDv/ctR7L8Ga72HY6ECHJIToQppN/NxuN6tXr2bJkiWsW7eOpKQkxowZw/79+7n77rsPO4eQ8FpTWIWjxs2NpyT6/diqugr15b9h4NAOO+ePEEIIIVpOG3cGaulX6P9+A8PgTDRzUKBDEkJ0Ec0mfjfccAMGg4HTTjuNiy++mN69ewOwcOHCFp9o7dq1zJkzB13XmThxYqPJ4AGWLFnC3LlziYuLA+DMM89k4sSJLT5PR5GzvYzoYCOZKRF+P7Za+B+orMBwwdV+P7YQQgghAkczGjFcegP67BmohfPRzrk40CEJIbqIZhO/Hj16sHnzZrZt20ZycjIJCQlERLQ8kdF1nddff50ZM2ZgsViYPn06mZmZpKamNtpu9OjRXHfddS0+fkdT5nSzKr+CczJiMRv92wxTlTpQ//sU7ZSxaD3S/XpsIYQQQgSe1v9EGDYa9eWHqNET0WItgQ5JCNEFNJv4PfTQQxQXF7N06VIWLFjAnDlzOPHEE6mtrcXj8fh8km3btpGUlERiorfZ4+jRo1m1alWTxK+rWLKzHLdOm8zdp76YBx432pTL/X5sIYQQQnQMhgunoa9fhfr4TbTr7wl0OEKILuCog7vEx8dz4YUXcuGFF7J582aWLl2Kpmn83//9H1lZWVxxxRVHPYnD4cBiOXi3ymKxsHXr1ibb/fjjj2zatInk5GSuvvpqrFZrk21ycnLIyckBYNasWQ3bmEymw27f3pRSLN69h4FJkQzr082vx3YX5mFftpDQ088nauCJLdq3o5RPRyXl0zwpn+ZJ+TRPykeIltPik9DOmIr64t+o8Wej9RkQ6JCEEJ1ciyZ/69+/P/379+eaa65h5cqVfPvttz7td7ghiX87EuXJJ5/MmDFjMJvNLFy4kBdeeIEHH3ywyX7Z2dlkZ2c3PLfZbABYrdaGx4GUa6thp72aW4Yn+T0efc5zYDJTm31+i4/dUcqno5LyaZ6UT/OkfJrnr/JJSUnxQzRCdB7aWReivvsG/YNXMdz3DzSDT7NwCSHEYbXqEyQoKIhTTz2V++67z6ftLRYLdru94bndbic2NrbRNpGRkZjN3mkPsrOz2bFjR2tCC7ic7WUEGTXG9oz063FV7kbUqmVop5+PFh179B2EEEII0alpwSFoF06D3dtQ3y8KdDhCiE6uXW4dpaenU1hYyP79+3G73axYsYLMzMxG25SUlDQ8Xr16dafs/1fr1lm2u5wx3SMJMxv9dlzlrEGf8yzEJ6Gd+Tu/HVcIIYQQHZs2fByk90d98jaqpjrQ4QghOrEWNfVsLaPRyLXXXsvMmTPRdZ2srCzS0tKYN28e6enpZGZm8tVXX7F69WqMRiMRERHccsst7RGaX63YU0G1S/f7oC7qozlg34/hz4+jBYf49dhCCCGE6Lg0TcPw+xvRZ96D+mIe2oXXBDokIUQn1S6JH8CwYcMYNmxYo2WXXHJJw+PLLruMyy67rL3CaRM5O8pIjjRzQkKo346pNv6MWvo12qSpaH0G+u2J1VOuAAAgAElEQVS4QgghhOgctB590EZPROUsQJ06CS3Jv4PHCSGOD9JL2E8KK+rYuK+aib2jmwxc01qquhL9recgOU2mbxBCCCGOY9oFV0JQEPprs1G1tYEORwjRCUni5yffbC/DoMGE3tF+O6b64FUoL8Fw7V1o5iC/HVcIIYQQnYsWFYvhuntgz3b0N55G6XqgQxJCdDKS+PmBR1cs2lHGScnhWMLMfjmmWvMD6vvFaGdfjNazr1+OKYQQQojOSxtyireP388rUJ++F+hwhBCdjCR+frCuqAp7jZvsdP/U9qmKcvS5L0D33mjnXOSXYwohhBCi89NOPx9t7CTUl/9G/35xoMMRQnQikvj5wf+2lxEVbOSUbsc+d59SCv3dF6GmCsM1d6GZ/FODKIQQQojOT9M0tMtugn6DUW8/h9r2a6BDEkJ0EpL4HaNyp5uVeRWM7xWF2Xjsg7qoVcvgpxVo512Gltrz2AMUQgghRJeimUwYbv4LxCWgv/A3VHFRoEMSQnQCkvgdo6W7ynHr+GXuPlXqQL37EvTuhzZpqh+iE0IIIURXpIVHYrj9ftB19Ocfk8ndhRBHJYnfMVBK8b/tZfS1hNAjJviYj6W//Ty467xNPI1GP0UphBBCiK5IS+qG4aZ7YV8++it/R3k8gQ5JCNGBSeJ3DLY5nOwurfXLoC7quxzYsBrtgqtlYlYhhBBC+EQbMMTb52/jT6gP3wh0OEKIDswU6AA6s2+2lxFk1BjbI+qYjqPs+1HzXoN+g9GyzvFTdEIIITqDtWvXMmfOHHRdZ+LEiUyZMqXJNitWrODDDz9E0zR69OjBnXfeGYBIRUdlGHcGemEeKudT9KRuGMafHeiQhBAdkCR+rVTr1vl2Vzmj0yIJD2p9s0yl6+hv/hMUGKbdgWaQSlghhDhe6LrO66+/zowZM7BYLEyfPp3MzExSU1MbtiksLGT+/Pk8+uijREREUFZWFsCIRUelXTQNtS8f9f4rqIRktIEnBTokIUQHI1lGK32/t4Iql052n2Nr5qmWfAmb16Ndch2aNdFP0QkhhOgMtm3bRlJSEomJiZhMJkaPHs2qVasabfPNN99wxhlnEBERAUB0tH/mjBVdi2YwYrjxT5Cchv7Sk6jCvECHJIToYCTxa6VvtpeRFGHmhISwVh9D7StAffwWDDoZ7dTT/RidEEKIzsDhcGCxWBqeWywWHA5Ho20KCgooLCzk/vvv569//Str165t7zBFJ6GFhHlH+jSZ0J97BFVZHuiQhBAdiDT1bIWSGjcb9lVz6WArBq11c/cp3YM+5xkwmTBcdRtaK48jhBCi81JKNVn22+uBrusUFhby4IMP4nA4eOCBB5g9ezbh4eFN9s3JySEnJweAWbNmYbVaATCZTA2PRVNdqnysVurue5KSB27H+Oo/iH3oWTSz+ZgO2aXKpw1I+TRPyqd57Vk+kvi1wqr8ShQwMi2i1cdQ//sUtm9Gu+6PaLGWo+8ghBCiy7FYLNjt9obndrud2NjYRtvExcWRkZGByWQiISGBlJQUCgsL6dOnT5PjZWdnk52d3fDcZrMBYLVaGx6Lprpc+ViT0a6+Hddrsyl+9hG0q+84phvMXa58/EzKp3lSPs3zR/mkpKT4tJ009WyFH/dWkBBubvXcfSp/D2r+OzBsFNqI0/wcnRBCiM4iPT2dwsJC9u/fj9vtZsWKFWRmZjbaZvjw4WzcuBGA8vJyCgsLSUyUPuGieYYRp6Gdeynqu29QC/8T6HCEEB2A1Pi1kNOts66omjP7xrTq7plyu9HfeBpCwzFcfrM08RRCiOOY0Wjk2muvZebM/2/v3uOjqu79/7/WzISQC4RkggmBcAsRuYohyEWsItHWK9RSrOegWDjUVisWH0WR01qqgnihWPvl/Gr5phbF1mgtcqRFEfBWgnIrYlXQAPIlEMEkEHKHZNbvj4GBmGRIQjI7Ce/n48Eje+/Ze/ZnPmyy+Mxae+35+Hw+xo0bR3JyMllZWaSkpJCens7FF1/MRx99xKxZs3C5XEyZMoVOnTo5Hbq0AebGH8BXudhXl2EvSMJcMsrpkETEQSr8GulfeaWc8Fku7dG0YZ52/evw/3bj+skcTOcuzRydiIi0NWlpaaSlpdXYdssttwSWjTFMnTqVqVOnhjo0aeOMywU/vBebfwjfHxfjWvAHTCfNCityvtJQz0balFtMVAcXA5s4m6fNXg/9BmDSxjRzZCIiIiI1mQ7huKb9DCorsX9/2elwRMRBKvwaodpn2XyglPSkaDyuJgzzzMuFA/sw6WNbIDoRERGR2ky3ZMxl47HvrsbmH3I6HBFxiAq/RtiZX05xZTUjmzrMc1s2AOaS0c0ZloiIiEhQ5sZbwbiwK//sdCgi4hAVfo2wKbcEjwsuSar97KSGsFs3QMpFmDg9y0RERERCx8TFY666HvvhO9jcvU6HIyIOCFnht337du69917uueceXnvttXr3++CDD5g8eTK7d+8OVWgNYq3lw9xihiREERnmbvzxhw/C/r2Y4Ze1QHQiIiIiwZlrJ0FEJL6/veB0KCLigJAUfj6fj8zMTObOncvixYvZsGEDubm5tfYrLy9n9erVpKamhiKsRsk9dpy84hNNH+a5dSOAJnURERERR5ioTpjvfA8+3oL9/BOnwxGREAtJ4ZeTk0NiYiIJCQl4PB7GjBnD5s2ba+2XlZXFTTfdRFhYWCjCapQPc0sAGNHkwm8D9LkQ4+3anGGJiIiINJi56kboEofvb8uw1jodjoiEUEgKv8LCQrxeb2Dd6/VSWFhYY5+9e/eSn5/P8OHDQxFSo23KLSElriPxkY0vSu3XX8G+HMxw9faJiIiIc0x4uP/B7rt3wkcfOh2OiIRQSB7gXtc3SsacfhyCz+dj2bJl3HXXXWd9r7Vr17J27VoAFi5cSHy8f6IUj8cTWG5uhaXH+Ty/nOmjejbpHKX/XEMJEDf+ejwtFOPZtGR+2gPlJzjlJzjlJzjlR6R1MZddjV2zEt/fXsA1dATG1fi5C0Sk7QlJ4ef1eikoKAisFxQUEBsbG1ivqKhg//79/PrXvwbg6NGjPPHEE9x///2kpKTUeK+MjAwyMjIC6/n5+QDEx8cHlpvbmpyjWGBwnKtJ56h+bw30TOGoJxxaKMazacn8tAfKT3DKT3DKT3DNlZ+kpKRmiEZEjNuN67tT8P3+cezGtzGXZZz9IBFp80JS+KWkpJCXl8fhw4eJi4sjOzubmTNnBl6PjIwkMzMzsD5v3jxuu+22WkWfUzbllnBBlIfeXcIbfawt+Br2fo757m0tEJmIiIhIE6SNgd6p2P/9M/bSb2HCOjgdkYi0sJDc4+d2u5k2bRrz589n1qxZjB49muTkZLKystiyZUsoQmiyiiofH31VyqU9OtUYntpQ9l8nH9quxziIiIhIK2GMwXXz7VCYj337706HIyIhEJIeP4C0tDTS0tJqbLvlllvq3HfevHkhiKhhtueVcrzansNjHLKhRx9MgoYoiYiISOthBlwMAy/B/uOv2LHXYCKjnA5JRFpQyB7g3lZtyi0hqoOLgRdENvpYe6QAcj7TbJ4iIiLSKrm+dzuUFmPf/JvToYhIC1PhF0S1z7LlQAnDk6LxuJoyzPPkQ9s1zFNERERaIdMzBTPicuzaldijhWc/QETaLBV+QezKL6eosvrchnkm9cR069HMkYmIiIg0DzPxP6G6GrvqJadDEZEWpMIviE25JXhckJbU+DHvtugIfPGJhnmKiIhIq2YuSMJc/m3s+2uwhw46HY6ItBAVfkF8mFvC4IQoIsMa/2BT+6+NYK2GeYqIiEirZ264BTxh2NeWOx2KiLQQFX71yC2q5GDx8XMb5pnYHZJ6NnNkIiIiIs3LxMRirp6A3fJP7L4cp8MRkRagwq8em3JLABjRvfGFny0ugl3/xqRd1qRn/4mIiIiEmvn2zRDdCd/fnnc6FBFpASr86vFhbgkpceF0jQpr9LH2Xx+A9WHSNcxTRERE2gYTEYm5bjJ8uh372UdOhyMizUyFXx2OllexK7+cS3t0atLxdms2XNANevRu3sBEREREWpC58lqI64rv1WVYa50OR0SakQq/Omw+UIIFLm3KMM+SY7DzI8zwMRrmKSIiIm2KCeuAmfAfsC8Htm5wOhwRaUYq/Oqw6UAJF0R56BMb3uhj7UebwOfTbJ4iItIg27dv59577+Wee+7htddeq3e/Dz74gMmTJ7N79+4QRifnIzPqSujeC9+K5diqKqfDEZFmosLvGyqrfGzPK2VEj05N6rGzW7PBewH0TGmB6EREpD3x+XxkZmYyd+5cFi9ezIYNG8jNza21X3l5OatXryY1NdWBKOV8Y1xuXN+9DQ4fxG5Y63Q4ItJMVPh9w/a8Uo5X26YN8ywrgU+3Y4ZrNk8RETm7nJwcEhMTSUhIwOPxMGbMGDZv3lxrv6ysLG666SbCwho/4ZhIkwwdAf0GYF9/CVtZ4XQ0ItIMVPh9w6YDJUSFuRicENnoY+32TVBdhRk+pgUiExGR9qawsBCv1xtY93q9FBYW1thn79695OfnM3z48FCHJ+cxYwyu702FokKOLpyDrShzOiQROUcepwNoTap9ls25JQxPisbjasIwz23ZEBcPfS5sgehERKS9qWvWxDNHjPh8PpYtW8Zdd93VoPdbu3Yta9f6h+YtXLiQ+Ph4ADweT2BZalN+6hH/LcrvfpBj/98TeBY/RJf/fgp3nPL0Tbp+glN+ggtlflT4neHz/HKKKqsZ0aMJwzzLy+CTbZgrr9cwTxERaRCv10tBQUFgvaCggNjY2MB6RUUF+/fv59e//jUAR48e5YknnuD+++8nJaX2veQZGRlkZGQE1vPz8wGIj48PLEttyk8Qw0bTZe4THH3yv8mfPR3Xz+ZhuiU7HVWrousnOOUnuObIT1JSUoP201DPM2w6UILHBcOTohp9rN2xGao0zFNERBouJSWFvLw8Dh8+TFVVFdnZ2aSnpwdej4yMJDMzkyVLlrBkyRJSU1PrLfpEWkr48NG4Zi+AqhP4Ft6P/fwTp0MSkSZQ4XeGD3NLGHxBJFEd3I0+1m7dAF3ioG//FohMRETaI7fbzbRp05g/fz6zZs1i9OjRJCcnk5WVxZYtW5wOTyTA9OqHa84T0LkLvsW/xG75p9MhiUgjaajnSbnHKjlw7DjXXxh79p2/wVaUw7+3YS6/BuNSLS0iIg2XlpZGWlpajW233HJLnfvOmzcvBBGJ1M10TcT1wOP4lszH9+wTmMJ8zNUTdIuLSBuhKuWkTbklAFzalPv7Pt4KJ45rmKeIiIi0aya6M65ZD0PaGOwrf8Rm/V+sr9rpsESkAVT4nbQpt4S+seF0jWrCM5K2boDOXaDfgOYPTERERKQVMR3Ccd05GzP+Ruy61/E9+yT2eKXTYYnIWajwA45WVLHz6/Km9fZVVmI/3oJJG41xNf7eQBEREZG2xrjcuH4wAzN5OmzLxrf4IWzJMafDEpEgQnaP3/bt23nuuefw+XyMHz+eiRMn1nh9zZo1vPnmm7hcLjp27Midd95Jjx49QhLblgMlWGBkj06NP/jfW+F4JWb4Zc0el4iIiEhr5rp6AjbWiy9zMb7HH8A181eYrolOhyUidQhJj5/P5yMzM5O5c+eyePFiNmzYQG5ubo19xo4dy6JFi3jyySeZMGECy5YtC0VogH+YZ3ykhz6x4Y0+1m7dAJ1iIHVQC0QmIiIi0rqZ9LH++/6OFfkf9/DlF06HJCJ1CEnhl5OTQ2JiIgkJCXg8HsaMGcPmzZtr7BMZGRlYrqioCNkMUZVVPv6VV8qlPaIbfU57vBK7YwvmklEYt4Z5ioiIyPnJXDgI15zHIawDvqf+G/uxHkci0tqEpPArLCzE6/UG1r1eL4WFhbX2e+ONN7jnnnt48cUX+eEPfxiK0Pjoq1KOV9umDfP89F9QWa7ZPEVEROS8Z7ol+5/1l5CE7/88iu+tldiqKqfDEpGTQnKPn7W21ra6ete+853v8J3vfId//vOfvPrqq/z0pz+ttc/atWtZu3YtAAsXLiQ+Ph4Aj8cTWG6Mj7YfIaqDmysGJhPmblwdXPTvrVRGdyZ+zDiMp3U/ErGp+TlfKD/BKT/BKT/BKT8i5w/TJQ7X7AX4/vAU9uVM7Ltv4PrubZA2Ws/7E3FYSKoVr9dLQUFBYL2goIDY2PoflD5mzBiWLl1a52sZGRlkZGQE1vPz8wGIj48PLDdUtc/y/u58LukWSdGR2j2QwdgTJ/Bteh+TNoaCo0cbdawTmpKf84nyE5zyE5zyE1xz5ScpKakZohGRlmY6RuK655fw0SZ8f3se3+8XQp8LcU26A3PhYKfDEzlvhWSoZ0pKCnl5eRw+fJiqqiqys7NJT0+vsU9eXl5gedu2bXTr1q3F4/qioIKiiuqmDfP8bDuUl2k2TxEREZFvMMZgho3E9atnMFPvgSMF+J6cS/UzD2Nzv3Q6PJHzUkh6/NxuN9OmTWP+/Pn4fD7GjRtHcnIyWVlZpKSkkJ6ezhtvvMHHH3+M2+0mOjqau+++u8XjOlRynE7hbtKSohp9rN2aDZFRMGBoC0QmIiIi0vYZtxsz9mrsiG9h16/Crv4rvofvxYy+CnPTf2C8XZ0OUeS8EbIb09LS0khLS6ux7ZZbbgksh2oylzNd0SeGsb0643Y1cjbPijLs9g8wF1+K8YS1UHQiIiIi7YMJD8dc+z3st67B/uMVfxG46T3M+Bsw107CRDVh9JWINEpIhnq2Zo0t+gDs2/+AslLMuOtbICIRERGR9slEdcL1/Wm4Hv09ZsTl2DWv4Zv7I3xvvIo9Xul0eCLtWuueirIVshVl2DdXwJB0TJ8LnQ5HRFoZay0VFRX4fD7NYHfSoUOHqKxs2H/orLW4XC46duyo/Im0Y8Z7AWbaz7DXTMT3t+exry7Drv87ZsJ/YEaPw7j0fGSR5qbCr5Hs+r9DaTGuG291OhQRaYUqKioICwvD08of8RJKHo8Ht7vh/4mrqqqioqKCiIiIFoxKRFoD06M37pkPYXd9jO/VZdg/PYNd8xqu70/DDE47+xuISIOd90M9G8NWlGHXvHayty/V6XBEpBXy+Xwq+s6Rx+PB5/M5HYaIhJDpPwTXg0/i+vEcqDqB77fz/DOA5uU6HZpIu6HCrxHU2yciZ6Phic1DeRQ5/xhjMMPH4Jr3fzDf/yHkfIrv1/fge2kptrTY6fBE2jx9Ld1Atvxkb9/QEertExEREWkhJiwMc813saPGYVf+Gbv+79gP3sHcdCvmW9/BaFSFSJOox6+B7PpVJ3v7fuB0KCIi9SoqKuJPf/pTo4+77bbbKCoqav6ARESayHTuguu2u3A9tBiS+2D/8gd8D9+L/fc2p0MTaZNU+DWALS/DvrXS39vXW719ItJ6HTt2jOeff77W9urq6qDHvfDCC8TExLRUWCIiTWZ69MF13yO47p6r+/9EzoH6yhtAvX0i0hS+l5Zi9+9t1vc0yX1w/WBGva8vWLCAffv2cfXVVxMWFkZkZCQJCQl88sknvPPOO0ybNo2DBw9SWVnJ9OnTmTJlCgAjR45k9erVlJaWMmXKFC699FK2bNlCYmIif/zjH+udYfPFF1/kxRdf5Pjx4/Tp04dnnnmGiIgIvv76a+bMmcO+ffswxrBgwQJGjBjBK6+8wrPPPgvAgAED+N3vftes+RGR9skYA8NG4Ro0HPv2KuyqLHy/vgdz5XWYG3+gB8CLNIAKv7MI3Nt38aWot09EWru5c+eya9cu3nrrLbKzs7n99ttZv349PXv2BGDRokXExsZSXl7O9ddfz3XXXUdcXFyN99i7dy9LlizhySef5M477+Qf//gH3/ve9+o837XXXst//ud/AvD444/zl7/8hWnTpvHLX/6SUaNGkZmZiTGGoqIidu3axTPPPMPKlSuJi4vjyJEjLZsMEWl3atz/97+6/0+kMfSv4yzs+lVQVqLePhFptGA9c6EybNiwQNEH8Mc//pHVq1cDcPDgQfbu3Vur8EtOTmbw4MEADB06lP3799f7/rt27eKJJ57g2LFjlJaWcsUVVwCwYcMGfvvb3wLgdrvp3Lkzf/3rX7n++usD54uNjW2+Dyoi5xXTuQtmyl3YK6/D93Im9i9/wL6zGnPV9ZjUQdAtGePSHU0iZ1LhF0SN3r5e/ZwOR0Sk0SIjIwPL2dnZvP/++7z++utEREQwadIkKisrax0THh4eWHa73VRUVNT7/rNmzSIzM5NBgwaRlZXFxo0b693XWqvHNNRh+/btPPfcc/h8PsaPH8/EiRNrvL5q1SrWrVsXKKB/8pOf0LVrV4eiFWldTI/euGY9DB9twvfqn7Av/h4LEBkN/QZg+g3EpA6AXqmYsDCnwxVxlAq/INTbJyJtTVRUFCUlJXW+VlxcTExMDBEREeTk5LBt27nPjFdSUkJCQgInTpxgxYoVJCYmAjB27Fief/55ZsyYQXV1NcXFxYwdO5bp06czY8aMwFDP873Xz+fzkZmZyS9+8Qu8Xi8PPvgg6enp9OjRI7BP7969WbhwIeHh4axZs4bly5cza9YsB6MWaV389/+NxHXxpZB/CPvFJ5DzGfaLT7E7NvsLQU8Y9Ek9WQgOhJSLMJHRTocuElIq/Ophy0rV2ycibU5cXBwjRozgqquuomPHjsTHxwdeu/LKK3nhhRfIyMigb9++pKWlnfP5Zs+ezQ033ECPHj246KKLAkXnww8/zP33389LL72E2+1mwYIFpKenM3PmTCZNmoTL5WLw4ME8/fTT5xxDW5aTk0NiYiIJCQkAjBkzhs2bN9co/E4NuwVITU3l/fffD3mcIm2BMQa6JmK6JsKY8QDY4iJ/EZjzqb8QXLMCu/qvYAx074XpN9DfM5g6CBMXf5YziLRtKvzqcbq371anQxERaZQlS5bUuT08PJzly5fX+dqHH34I+AvH9evXB7b/+Mc/DnquqVOnMnXq1Frbu3btynPPPQeAx+OhqqoKgMmTJzN58uSzf4jzRGFhIV6vN7Du9Xr54osv6t1//fr1DBs2LBShibQLplMMXDIKc8koAGxlJezddbIQ/Ay78W145x/+XsGuiZiLhsKFgzEXDcF08QZ9b5G2RoVfHWxZqf+5fRdfiumV4nQ4IiLSTllra22r7z7I9957jz179jBv3rx632/t2rWsXbsWgIULFwZ6fD0eT43eX6lJ+Qmu3eWne3cYexUAtrqKqi93c/zT7Zz4eCvHt23Evr8GC7i7JdNhSBodBqcRNjgNd2zdhWC7y08zU36CC2V+VPjVQb19IiI1zZ07l82bN9fY9l//9V/ccsstDkXUPni9XgoKCgLrBQUFdd73uGPHDlasWMG8efMICzJBRUZGBhkZGYH1/Px8AOLj4wPLUpvyE1y7z0+MF0aPh9HjMb5qzP4vsbs+pnrXx5S//xbla1b690vsgek/GPoPxfQfjOncBTgP8nOOlJ/gmiM/SUlJDdpPhd83+Hv7XoNhI9XbJyJy0oIFC5wOoV1KSUkhLy+Pw4cPExcXR3Z2NjNnzqyxz969e1m6dClz584lJibGoUhFzg/G5YZeKf7/A14zEVtdDfv3YHd9jN35MfaDd+HdN/xDQ7slYy4aQtlFQ7BRMZCQBDGxmr1YWi0Vft9g178OZaWayVNERFqc2+1m2rRpzJ8/H5/Px7hx40hOTiYrK4uUlBTS09NZvnw5FRUV/OY3vwH83w4/8MADDkcucn4wbjf0TsX0ToVv3+wvBPflYHf9G7trBzZ7PcVv/+P0AeERkJCESezuLwQTumNO/YyIrPc8IqGgwu8MgXv7ho3E9FRvn4iItLy0tLRaM6yeOYT2l7/8ZahDEpF6GLcb+vbH9O0P134P66smDh+Fn/0be+gAHDqIPXQAu2cXbH4frCVwJ2/nLpDYHZPgLwpNcl9IHaTnC0rIqPA7g12n3j4RERERaRjjcuOOT8C4wjCDLqnxmj1xHA5/BYcOYA8dDPy02z+E4iJ/QRjeES4aihmSjhk8HOPt6sjnkPODCr+TbFkJdq16+0RERETk3JmwDtC9J3TvyTfv+rOlJbBnJ/bjLdgdW7AfbfIXgt17+YvAIcOh70UYj/6rLs0nZFfT9u3bee655/D5fIwfP56JEyfWeH3VqlWsW7cOt9tN586d+clPfkLXrqH71sOuW6XePhE576SmpgZ9bpyIiDQ/ExUNQ9IxQ9Kxt1r4KtdfBH68FfvWa9g3XoWIKMzAYf79BqdhYmrP+CvSGCEp/Hw+H5mZmfziF7/A6/Xy4IMPkp6eTo8ePQL79O7dm4ULFxIeHs6aNWtYvnw5s2bNCkV4Z/T2jVJvn4iIiIiEjDHGP0Not2S45rvY8jL47CPsv7diP94CWzf4ewN79cMMGY4ZcDF0jDx1MKe7E41//dTyqR9nbusQDlFREB7R4rOP2upqqCjDajbiViMkhV9OTg6JiYkkJCQAMGbMGDZv3lyj8Bs8eHBgOTU1lffffz8UoQFg1566t0/PoxKR5vN/txxi75GKZn3PPrEd+a/0hHpfnz9/Pt27d+eOO+4AYNGiRRhj+OCDDygqKqKqqor777+fb3/722c9V2lpKT/84Q/rPO6VV17h2WefBWDAgAH87ne/4+uvv2bOnDns27cPgMcee4wRI0ac4ycWETm/mIhISBuNSRuNtRZyvzzdG/j3V7Crss79JC4XREZBZHTgj4mKPr0tKtrf4xh18vUO4VBehi0vhbJSqPHz1PYSKC87vb3S3/4dBojuBDFx/sddxMQGlomJw8R0Of1ax4hz/2yNYH3VcKQQ8g9hCw5hwjtCnwshNr5dPpYjJIVfYWEhXq83sO71eoMOLVq/fj3Dhg0LRWgne/v+V719ItIuTJgwgV/96leBwu/111/nxRdfZMaMGXTq1InCwkJuvPFGrrnmmrM2auHh4WRmZtY67vPPP+eZZ6D6hBoAAA9MSURBVJ5h5cqVxMXFceTIEcA/++SoUaPIzMykurqa0tLSlv64IiLtmjEGkvtgkvvAdd/HlhbDnl1QVQXWwqk5Qy1nLNszfpyxzVo4XukvzMpK/H9KS7BlJVBWis0/dHq7z8cZR9ftVPEYcbJYjIj0F28RUae3R0QQaaAs7wC26AgUHcF+lQtFR6G6qvY5wiP8BWGXWEynLtCpM0Sf/mOiOtXc1iE8aFtmrYXio/D1If/nyz8EBYdPLxfmB+KoEUtMHPS9ENPn5AyuvVKapSi1x47CgX3YA/sCP8uuug5GXXXO790QISn8rK192dT3l/Tee++xZ88e5s2bV+fra9euZe3atQAsXLiQ+Ph4ADweT2C5MUqyVlJaXkrc7T8mrAnHtxVNzc/5QvkJTvkJ7sz8HDp0CM/Jm/F/PKp7yGMZNmwYBQUF5OfnU1BQQJcuXUhKSuKhhx5i48aNuFwuvvrqK44cOcIFF1wQiL8u1lqeeOKJWsdt3LiRG2+8MXD8qfuxs7OzWbJkCR6PB4/HQ3h4eOC96jtHfcLDw3XNiYh8g4nqBEPSG7ZvE89hrfX31gWKw1J/wRgRARGnegWjzlp0nRIdH09Ffn7Nc/h8UFoCRYVw7Aj2qL8opKjQXxwWFWJzv4TSY/79ThWz33zzsA7+AvBkQWiiO0PHCOzRwkCRx/HKmsd0ioH4BEzvVEi/zL8cnwDeBCgrwe75HPbswu7dhf3XB/5zGpd/4p2+F0KfC/3FYGIPjMtVdw4ryuHg/6tR4HFgHxQXnZGYzv73jO501hw2l5AUfl6vl4KCgsB6QUEBsbG1b1DdsWMHK1asYN68eYTV80yTjIwMMjIyAuv5Jy+k+Pj4wHJD2bISfCtfgktGUdQpDhp5fFvSlPycT5Sf4JSf4M7MT2VlJW6329F4rrvuOlauXMnhw4e56aabePnll/n6669ZvXo1YWFhjBw5ktLSUqqq/N9ynvr5TfUdV11djbW21nGntn3z83s8nnrPUZ/Kyspa11xSUlKj3kNERBrPGAMdI/x/4lpmokXjcvl77jp1BnoHLVKtr9pffJYegxL/H1t8DEqK/eulx7Anl+2+3f5hpjGx/mclDrrkdGEXnwDeC87ac2f6XAjjb/Cfu7gI9n6O3fs5ds8u7OZ/wntv+ovBiEjonYrp0x8SusGhPOyBL/0FXv6h02/YIRySemKGjoAevTBJvaBHL+jUBWMMEfHxlIbo/1ghKfxSUlLIy8vj8OHDxMXFkZ2dzcyZM2vss3fvXpYuXcrcuXOJCdFNoDZ7HZSX4rpBM3mKSPsxYcIEZs+eTWFhIa+++iqvv/468fHxhIWFsWHDBnJzcxv0PsXFxXUeN3bsWKZPn86MGTMCQz1jY2MZO3Yszz//PDNmzKC6upqysjI6dQrdN5kiItL+GJf7jCLx5LZQnbtTDAwd4S/aONlTeehAzV7BN/7qHxrrckFCd0zvVLgsA9O9F3Tv5S886+kZDLWQFH5ut5tp06Yxf/58fD4f48aNIzk5maysLFJSUkhPT2f58uVUVFTwm9/8BvB/g/7AAw+0aFzmqhswyX0xPfu26HlEREKpf//+lJaWBibVuvnmm5k6dSrXXnstgwYNol+/fg16n/qO69+/PzNnzmTSpEm4XC4GDx7M008/zcMPP8z999/PSy+9hMvl4rHHHiM9vWFDkkRERFo743KdnoH1svEA2MoK/72C8QmYekYsthbG1nUDXhty8OBBQEPRzkb5CU75CU75Ce7M/JSVlREZGelwRK1LU4Z61pVHDfVsPLWRDaP8BKf8BKf8BKf8BNcc+Wlo+9g6+h1FRERERESkxYRkqKeIiLRen332Wa37rsPDw1m1apVDEYmIiEhzU+EnInKeGzBgAG+99ZbTYYiIiEgL0lBPEZFm1MZvm241lEcREZHmpcJPRKQZuVyuRk9kIjVVVVXhaiVTX4uIiLQXGuopItKMOnbsSEVFBZWVlf6H4Arh4eFUVlY2aF9rLS6Xi44dO7ZwVCIiIucXFX4iIs3IGENERITTYbQqmspbRETEeRpLIyIiIiIi0s6p8BMREREREWnnVPiJiIiIiIi0c8ZqzmwREREREZF2rd30+M2ZM8fpEFo15Sc45Sc45Sc45Sc45cd5+jsITvkJTvkJTvkJTvkJLpT5aTeFn4iIiIiIiNRNhZ+IiIiIiEg75543b948p4NoLn379nU6hFZN+QlO+QlO+QlO+QlO+XGe/g6CU36CU36CU36CU36CC1V+NLmLiIiIiIhIO6ehniIiIiIiIu2cx+kAztX27dt57rnn8Pl8jB8/nokTJzodUqtz991307FjR1wuF263m4ULFzodkqP+53/+h23bthETE8OiRYsAKCkpYfHixXz99dd07dqVWbNmER0d7XCkzqgrPy+//DLr1q2jc+fOANx6662kpaU5GaZj8vPzWbJkCUePHsUYQ0ZGBtddd52uoZPqy4+uIWeojQxO7WNNah+DU/sYnNrH4FpF+2jbsOrqavvTn/7UfvXVV/bEiRP25z//ud2/f7/TYbU6d911ly0qKnI6jFbjk08+sbt377b33XdfYNsLL7xgV6xYYa21dsWKFfaFF15wKjzH1ZWfrKwsu3LlSgejaj0KCwvt7t27rbXWlpWV2ZkzZ9r9+/frGjqpvvzoGgo9tZFnp/axJrWPwal9DE7tY3CtoX1s00M9c3JySExMJCEhAY/Hw5gxY9i8ebPTYUkrN3DgwFrfNG3evJkrrrgCgCuuuOK8vo7qyo+cFhsbG7gJOyIigu7du1NYWKhr6KT68iOhpzZSGkvtY3BqH4NT+xhca2gf2/RQz8LCQrxeb2Dd6/XyxRdfOBhR6zV//nwArr76ajIyMhyOpvUpKioiNjYW8P/DPHbsmMMRtT5vvvkm7733Hn379uX2229X4wccPnyYvXv30q9fP11DdTgzPzt37tQ1FGJqIxtG7WNw+t12dvrdVpvax+Ccah/bdOFn65iQ1BjjQCSt2yOPPEJcXBxFRUU8+uijJCUlMXDgQKfDkjbkmmuuYdKkSQBkZWXx/PPPc9dddzkclbMqKipYtGgRd9xxB5GRkU6H0+p8Mz+6hkJPbeTZqX2Uc6XfbbWpfQzOyfaxTQ/19Hq9FBQUBNYLCgoC3yjIaXFxcQDExMQwYsQIcnJyHI6o9YmJieHIkSMAHDlyJHCDrfh16dIFl8uFy+Vi/Pjx7N692+mQHFVVVcWiRYu4/PLLGTlyJKBr6Ex15UfXUOipjTw7tY9np99twel3W01qH4Nzun1s04VfSkoKeXl5HD58mKqqKrKzs0lPT3c6rFaloqKC8vLywPKOHTvo2bOnw1G1Punp6bz77rsAvPvuu4wYMcLhiFqXU7+wATZt2kRycrKD0TjLWsvvf/97unfvzg033BDYrmvIr7786BoKPbWRwal9bBj9bgtOv9tOU/sYXGtoH9v8A9y3bdvGsmXL8Pl8jBs3jptvvtnpkFqVQ4cO8dRTTwFQXV3N2LFjz/scPf3003z66acUFxcTExPD5MmTGTFiBIsXLyY/P5/4+Hjuu+++83aMfl35+eSTT/jyyy8xxtC1a1d+9KMfnbc9Bzt37uShhx6iZ8+egWFzt956K6mpqbqGqD8/GzZs0DXkALWR9VP7WJvax+DUPgan9jG41tA+tvnCT0RERERERIJr00M9RURERERE5OxU+ImIiIiIiLRzKvxERERERETaORV+IiIiIiIi7ZwKPxERERERkXZOhZ9IOzB58mS++uorp8MQERFpddRGivh5nA5ApL25++67OXr0KC7X6e9VrrzySqZPn+5gVHV78803KSws5NZbb+VXv/oV06ZNo1evXk6HJSIi7ZTaSBHnqPATaQEPPPAAQ4cOdTqMs9qzZw9paWn4fD5yc3Pp0aOH0yGJiEg7pzZSxBkq/ERC6J133mHdunX06dOHd999l9jYWKZPn86QIUMAKCwsZOnSpezcuZPo6GgmTJhARkYGAD6fj9dee423336boqIiunXrxuzZs4mPjwdgx44dLFiwgOLiYi677DKmT5+OMSZoPHv27GHSpEkcPHiQCy64ALfb3bIJEBERqYfaSJGWpcJPJMS++OILRo4cSWZmJps2beKpp55iyZIlREdH89vf/pbk5GSeffZZDh48yCOPPEJCQgJDhgxh1apVbNiwgQcffJBu3bqxb98+wsPDA++7bds2HnvsMcrLy3nggQdIT09n2LBhtc5/4sQJZsyYgbWWiooKZs+eTVVVFT6fjzvuuIObbrqJm2++OZQpERERAdRGirQkFX4iLeDJJ5+s8c3glClTAt9KxsTEcP3112OMYcyYMbz++uts27aNgQMHsnPnTubMmUOHDh3o3bs348eP57333mPIkCGsW7eOKVOmkJSUBEDv3r1rnHPixIlERUURFRXFoEGD+PLLL+ts1MLCwvjTn/7EunXr2L9/P3fccQePPvooP/jBD+jXr1/LJUVERAS1kSJOUeEn0gJmz55d7/0LcXFxNYaXdO3alcLCQo4cOUJ0dDQRERGB1+Lj49m9ezcABQUFJCQk1HvOLl26BJbDw8OpqKioc7+nn36a7du3U1lZSVhYGG+//TYVFRXk5OTQrVs3HnvssUZ9VhERkcZQGyniDBV+IiFWWFiItTbQsOXn55Oenk5sbCwlJSWUl5cHGrb8/Hzi4uIA8Hq9HDp0iJ49e57T+X/2s5/h8/n40Y9+xB/+8Ae2bt3Kxo0bmTlz5rl9MBERkXOkNlKk5eg5fiIhVlRUxOrVq6mqqmLjxo0cOHCASy65hPj4ePr378+f//xnjh8/zr59+3j77be5/PLLARg/fjxZWVnk5eVhrWXfvn0UFxc3KYYDBw6QkJCAy+Vi7969pKSkNOdHFBERaRK1kSItRz1+Ii3g8ccfr/GMoqFDhzJ79mwAUlNTycvLY/r06XTp0oX77ruPTp06AXDvvfeydOlS7rzzTqKjo/n+978fGA5zww03cOLECR599FGKi4vp3r07P//5z5sU3549e+jTp09gecKECefycUVERBpMbaSIM4y11jodhMj54tRU1Y888ojToYiIiLQqaiNFWpaGeoqIiIiIiLRzKvxERERERETaOQ31FBERERERaefU4yciIiIiItLOqfATERERERFp51T4iYiIiIiItHMq/ERERERERNo5FX4iIiIiIiLtnAo/ERERERGRdu7/B7Yp69pjxXo1AAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the training loss and accuracy\n",
    "N = num_epochs\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(15,5))\n",
    "\n",
    "# summarize history for Accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(np.arange(1, N+1), H1.history[\"categorical_accuracy\"], label=\"train_acc\")\n",
    "plt.plot(np.arange(1, N+1), H1.history[\"val_categorical_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"modelC2D_1:Training and Validation Accuracy on Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(np.arange(1, N+1), H1.history['loss'], label=\"train_loss\")\n",
    "plt.plot(np.arange(1, N+1), H1.history['val_loss'], label=\"validation_loss\")\n",
    "plt.title(\"modelC2D_1:Training and Validation Loss on Dataset\")\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model 2: TimeDistributed Conv2d + RNN Model (CNN + RNN) : Increased layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_10 (TimeDis (None, 10, 60, 60, 32)    4736      \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 10, 60, 60, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_11 (TimeDis (None, 10, 58, 58, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 10, 58, 58, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_12 (TimeDis (None, 10, 29, 29, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_13 (TimeDis (None, 10, 29, 29, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_7 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_14 (TimeDis (None, 10, 29, 29, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_8 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_15 (TimeDis (None, 10, 14, 14, 64)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_16 (TimeDis (None, 10, 14, 14, 128)   73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_9 (Batch (None, 10, 14, 14, 128)   512       \n",
      "_________________________________________________________________\n",
      "time_distributed_17 (TimeDis (None, 10, 14, 14, 128)   147584    \n",
      "_________________________________________________________________\n",
      "batch_normalization_10 (Batc (None, 10, 14, 14, 128)   512       \n",
      "_________________________________________________________________\n",
      "time_distributed_18 (TimeDis (None, 10, 7, 7, 128)     0         \n",
      "_________________________________________________________________\n",
      "time_distributed_19 (TimeDis (None, 10, 6272)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_20 (TimeDis (None, 10, 64)            401472    \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_21 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_2 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 1,097,957\n",
      "Trainable params: 1,097,061\n",
      "Non-trainable params: 896\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Model 2 : Conv2D + RNN with LSTM\n",
    "\n",
    "# Initial layers take the image nD array as input\n",
    "modelC2D_2 = Sequential()\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(32, (7, 7), strides=(2, 2), activation='relu', padding='same'), input_shape=(10, 120, 120, 3)))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(32, (3,3), kernel_initializer=\"he_normal\", activation='relu')))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_2.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_2.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 128 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(128, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# 128 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_2.add(TimeDistributed(Conv2D(128, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_2.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_2.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# Flatten the feature map into Vector\n",
    "modelC2D_2.add(TimeDistributed(Flatten()))\n",
    "\n",
    "# Add a fully connected layer with 50% connections dropped out\n",
    "modelC2D_2.add(TimeDistributed(Dense(64, activation='relu')))\n",
    "modelC2D_2.add(Dropout(0.5))\n",
    "\n",
    "# Add second fully connected layer with 50% connections dropped out\n",
    "modelC2D_2.add(TimeDistributed(Dense(128, activation='relu')))\n",
    "modelC2D_2.add(Dropout(0.5))\n",
    "\n",
    "# Add a LSTM layer\n",
    "modelC2D_2.add(LSTM(256, return_sequences=False, dropout=0.5))\n",
    "\n",
    "# Softmax layer for final 5-class output\n",
    "modelC2D_2.add(Dense(5, activation='softmax'))\n",
    "modelC2D_2.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_10 (TimeDis (None, 10, 60, 60, 32)    4736      \n",
      "_________________________________________________________________\n",
      "batch_normalization_5 (Batch (None, 10, 60, 60, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_11 (TimeDis (None, 10, 58, 58, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_6 (Batch (None, 10, 58, 58, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_12 (TimeDis (None, 10, 29, 29, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_13 (TimeDis (None, 10, 29, 29, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_7 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_14 (TimeDis (None, 10, 29, 29, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_8 (Batch (None, 10, 29, 29, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_15 (TimeDis (None, 10, 14, 14, 64)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_16 (TimeDis (None, 10, 14, 14, 128)   73856     \n",
      "_________________________________________________________________\n",
      "batch_normalization_9 (Batch (None, 10, 14, 14, 128)   512       \n",
      "_________________________________________________________________\n",
      "time_distributed_17 (TimeDis (None, 10, 14, 14, 128)   147584    \n",
      "_________________________________________________________________\n",
      "batch_normalization_10 (Batc (None, 10, 14, 14, 128)   512       \n",
      "_________________________________________________________________\n",
      "time_distributed_18 (TimeDis (None, 10, 7, 7, 128)     0         \n",
      "_________________________________________________________________\n",
      "time_distributed_19 (TimeDis (None, 10, 6272)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_20 (TimeDis (None, 10, 64)            401472    \n",
      "_________________________________________________________________\n",
      "dropout_3 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_21 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_4 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_2 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_6 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 1,097,957\n",
      "Trainable params: 1,097,061\n",
      "Non-trainable params: 896\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "optimiser = Adam(lr = 0.0001) #write your optimizer\n",
    "modelC2D_2.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['categorical_accuracy'])\n",
    "print (modelC2D_2.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "45/45 [==============================] - 23s 502ms/step - loss: 1.6579 - categorical_accuracy: 0.2371 - val_loss: 1.4669 - val_categorical_accuracy: 0.3700\n",
      "\n",
      "Epoch 00001: saving model to model_init_2020-03-0819_30_13.737684/model-00001-1.64995-0.24133-1.46685-0.37000.h5\n",
      "Epoch 2/25\n",
      "45/45 [==============================] - 18s 410ms/step - loss: 1.5113 - categorical_accuracy: 0.3172 - val_loss: 1.2766 - val_categorical_accuracy: 0.4600\n",
      "\n",
      "Epoch 00002: saving model to model_init_2020-03-0819_30_13.737684/model-00002-1.51031-0.32278-1.27662-0.46000.h5\n",
      "Epoch 3/25\n",
      "45/45 [==============================] - 20s 451ms/step - loss: 1.3726 - categorical_accuracy: 0.4223 - val_loss: 1.2416 - val_categorical_accuracy: 0.5500\n",
      "\n",
      "Epoch 00003: saving model to model_init_2020-03-0819_30_13.737684/model-00003-1.37162-0.42383-1.24165-0.55000.h5\n",
      "Epoch 4/25\n",
      "45/45 [==============================] - 20s 443ms/step - loss: 1.2210 - categorical_accuracy: 0.5199 - val_loss: 1.2180 - val_categorical_accuracy: 0.5400\n",
      "\n",
      "Epoch 00004: saving model to model_init_2020-03-0819_30_13.737684/model-00004-1.22404-0.51735-1.21799-0.54000.h5\n",
      "Epoch 5/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 1.1142 - categorical_accuracy: 0.5779 - val_loss: 0.9565 - val_categorical_accuracy: 0.6000\n",
      "\n",
      "Epoch 00005: saving model to model_init_2020-03-0819_30_13.737684/model-00005-1.10262-0.58220-0.95647-0.60000.h5\n",
      "Epoch 6/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.9190 - categorical_accuracy: 0.6816 - val_loss: 0.7883 - val_categorical_accuracy: 0.6500\n",
      "\n",
      "Epoch 00006: saving model to model_init_2020-03-0819_30_13.737684/model-00006-0.91784-0.68778-0.78829-0.65000.h5\n",
      "Epoch 7/25\n",
      "45/45 [==============================] - 20s 445ms/step - loss: 0.7788 - categorical_accuracy: 0.7498 - val_loss: 1.0615 - val_categorical_accuracy: 0.6000\n",
      "\n",
      "Epoch 00007: saving model to model_init_2020-03-0819_30_13.737684/model-00007-0.77088-0.75716-1.06151-0.60000.h5\n",
      "Epoch 8/25\n",
      "45/45 [==============================] - 20s 434ms/step - loss: 0.6206 - categorical_accuracy: 0.8134 - val_loss: 1.1528 - val_categorical_accuracy: 0.5700\n",
      "\n",
      "Epoch 00008: saving model to model_init_2020-03-0819_30_13.737684/model-00008-0.62083-0.81599-1.15284-0.57000.h5\n",
      "\n",
      "Epoch 00008: ReduceLROnPlateau reducing learning rate to 4.999999873689376e-05.\n",
      "Epoch 9/25\n",
      "45/45 [==============================] - 20s 447ms/step - loss: 0.4884 - categorical_accuracy: 0.8504 - val_loss: 0.7096 - val_categorical_accuracy: 0.7400\n",
      "\n",
      "Epoch 00009: saving model to model_init_2020-03-0819_30_13.737684/model-00009-0.48628-0.85370-0.70964-0.74000.h5\n",
      "Epoch 10/25\n",
      "45/45 [==============================] - 20s 442ms/step - loss: 0.3869 - categorical_accuracy: 0.9051 - val_loss: 0.7407 - val_categorical_accuracy: 0.7600\n",
      "\n",
      "Epoch 00010: saving model to model_init_2020-03-0819_30_13.737684/model-00010-0.39098-0.90347-0.74066-0.76000.h5\n",
      "Epoch 11/25\n",
      "45/45 [==============================] - 20s 448ms/step - loss: 0.3397 - categorical_accuracy: 0.9068 - val_loss: 0.7515 - val_categorical_accuracy: 0.7400\n",
      "\n",
      "Epoch 00011: saving model to model_init_2020-03-0819_30_13.737684/model-00011-0.33747-0.91101-0.75146-0.74000.h5\n",
      "\n",
      "Epoch 00011: ReduceLROnPlateau reducing learning rate to 2.499999936844688e-05.\n",
      "Epoch 12/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.2715 - categorical_accuracy: 0.9422 - val_loss: 0.6271 - val_categorical_accuracy: 0.7600\n",
      "\n",
      "Epoch 00012: saving model to model_init_2020-03-0819_30_13.737684/model-00012-0.27469-0.94118-0.62714-0.76000.h5\n",
      "Epoch 13/25\n",
      "45/45 [==============================] - 20s 451ms/step - loss: 0.2184 - categorical_accuracy: 0.9600 - val_loss: 0.8554 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00013: saving model to model_init_2020-03-0819_30_13.737684/model-00013-0.22049-0.95928-0.85540-0.72000.h5\n",
      "Epoch 14/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.2101 - categorical_accuracy: 0.9585 - val_loss: 0.6449 - val_categorical_accuracy: 0.7900\n",
      "\n",
      "Epoch 00014: saving model to model_init_2020-03-0819_30_13.737684/model-00014-0.21127-0.95777-0.64492-0.79000.h5\n",
      "\n",
      "Epoch 00014: ReduceLROnPlateau reducing learning rate to 1.249999968422344e-05.\n",
      "Epoch 15/25\n",
      "45/45 [==============================] - 20s 441ms/step - loss: 0.2100 - categorical_accuracy: 0.9555 - val_loss: 0.8109 - val_categorical_accuracy: 0.7300\n",
      "\n",
      "Epoch 00015: saving model to model_init_2020-03-0819_30_13.737684/model-00015-0.20962-0.95475-0.81088-0.73000.h5\n",
      "Epoch 16/25\n",
      "45/45 [==============================] - 20s 436ms/step - loss: 0.1867 - categorical_accuracy: 0.9557 - val_loss: 0.7124 - val_categorical_accuracy: 0.7300\n",
      "\n",
      "Epoch 00016: saving model to model_init_2020-03-0819_30_13.737684/model-00016-0.17956-0.96078-0.71239-0.73000.h5\n",
      "\n",
      "Epoch 00016: ReduceLROnPlateau reducing learning rate to 6.24999984211172e-06.\n",
      "Epoch 17/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.1603 - categorical_accuracy: 0.9718 - val_loss: 0.8503 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00017: saving model to model_init_2020-03-0819_30_13.737684/model-00017-0.16251-0.97134-0.85034-0.70000.h5\n",
      "Epoch 18/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.1500 - categorical_accuracy: 0.9748 - val_loss: 0.8620 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00018: saving model to model_init_2020-03-0819_30_13.737684/model-00018-0.14992-0.97436-0.86198-0.70000.h5\n",
      "\n",
      "Epoch 00018: ReduceLROnPlateau reducing learning rate to 3.12499992105586e-06.\n",
      "Epoch 19/25\n",
      "45/45 [==============================] - 20s 446ms/step - loss: 0.1646 - categorical_accuracy: 0.9631 - val_loss: 0.6053 - val_categorical_accuracy: 0.7800\n",
      "\n",
      "Epoch 00019: saving model to model_init_2020-03-0819_30_13.737684/model-00019-0.15639-0.96833-0.60531-0.78000.h5\n",
      "Epoch 20/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.1417 - categorical_accuracy: 0.9807 - val_loss: 0.8084 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00020: saving model to model_init_2020-03-0819_30_13.737684/model-00020-0.14044-0.98039-0.80838-0.71000.h5\n",
      "Epoch 21/25\n",
      "45/45 [==============================] - 20s 450ms/step - loss: 0.1480 - categorical_accuracy: 0.9778 - val_loss: 0.7513 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00021: saving model to model_init_2020-03-0819_30_13.737684/model-00021-0.14811-0.97738-0.75132-0.71000.h5\n",
      "\n",
      "Epoch 00021: ReduceLROnPlateau reducing learning rate to 1.56249996052793e-06.\n",
      "Epoch 22/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.1436 - categorical_accuracy: 0.9733 - val_loss: 0.8036 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00022: saving model to model_init_2020-03-0819_30_13.737684/model-00022-0.14412-0.97285-0.80361-0.68000.h5\n",
      "Epoch 23/25\n",
      "45/45 [==============================] - 20s 443ms/step - loss: 0.1576 - categorical_accuracy: 0.9659 - val_loss: 0.9285 - val_categorical_accuracy: 0.6700\n",
      "\n",
      "Epoch 00023: saving model to model_init_2020-03-0819_30_13.737684/model-00023-0.15884-0.96531-0.92852-0.67000.h5\n",
      "\n",
      "Epoch 00023: ReduceLROnPlateau reducing learning rate to 7.81249980263965e-07.\n",
      "Epoch 24/25\n",
      "45/45 [==============================] - 19s 427ms/step - loss: 0.1530 - categorical_accuracy: 0.9704 - val_loss: 0.7022 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00024: saving model to model_init_2020-03-0819_30_13.737684/model-00024-0.15387-0.96983-0.70217-0.71000.h5\n",
      "Epoch 25/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.1499 - categorical_accuracy: 0.9674 - val_loss: 0.8406 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00025: saving model to model_init_2020-03-0819_30_13.737684/model-00025-0.14959-0.96682-0.84055-0.72000.h5\n",
      "\n",
      "Epoch 00025: ReduceLROnPlateau reducing learning rate to 3.906249901319825e-07.\n"
     ]
    }
   ],
   "source": [
    "H2 = modelC2D_2.fit_generator(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                    callbacks=callbacks_list, validation_data=val_generator, \n",
    "                    validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the training loss and accuracy\n",
    "N = num_epochs\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(15,5))\n",
    "\n",
    "# summarize history for Accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(np.arange(1, N+1), H2.history[\"categorical_accuracy\"], label=\"train_acc\")\n",
    "plt.plot(np.arange(1, N+1), H2.history[\"val_categorical_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"modelC2D_2:Training and Validation Accuracy on Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(np.arange(1, N+1), H2.history['loss'], label=\"train_loss\")\n",
    "plt.plot(np.arange(1, N+1), H2.history['val_loss'], label=\"validation_loss\")\n",
    "plt.title(\"modelC2D_2:Training and Validation Loss on Dataset\")\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model 3: TimeDistributed Conv2d + RNN Model (CNN + RNN) : Reduced Filter size, same no. of layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_22 (TimeDis (None, 10, 60, 60, 16)    2368      \n",
      "_________________________________________________________________\n",
      "batch_normalization_11 (Batc (None, 10, 60, 60, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_23 (TimeDis (None, 10, 58, 58, 16)    2320      \n",
      "_________________________________________________________________\n",
      "batch_normalization_12 (Batc (None, 10, 58, 58, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_24 (TimeDis (None, 10, 29, 29, 16)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_25 (TimeDis (None, 10, 29, 29, 32)    4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_13 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_26 (TimeDis (None, 10, 29, 29, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_14 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_27 (TimeDis (None, 10, 14, 14, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_28 (TimeDis (None, 10, 14, 14, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_15 (Batc (None, 10, 14, 14, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_29 (TimeDis (None, 10, 14, 14, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_16 (Batc (None, 10, 14, 14, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_30 (TimeDis (None, 10, 7, 7, 64)      0         \n",
      "_________________________________________________________________\n",
      "time_distributed_31 (TimeDis (None, 10, 3136)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_32 (TimeDis (None, 10, 64)            200768    \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_33 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_3 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 679,509\n",
      "Trainable params: 679,061\n",
      "Non-trainable params: 448\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Model 3 : Conv2D + RNN with LSTM\n",
    "\n",
    "# Initial layers take the image nD array as input\n",
    "modelC2D_3 = Sequential()\n",
    "# 16 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(16, (7, 7), strides=(2, 2), activation='relu', padding='same'), input_shape=(10, 120, 120, 3)))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# 16 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(16, (3,3), kernel_initializer=\"he_normal\", activation='relu')))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_3.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(32, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(32, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_3.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# 64 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_3.add(TimeDistributed(Conv2D(64, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_3.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_3.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# Flatten the feature map into Vector\n",
    "modelC2D_3.add(TimeDistributed(Flatten()))\n",
    "\n",
    "# Add a fully connected layer with 50% connections dropped out\n",
    "modelC2D_3.add(TimeDistributed(Dense(64, activation='relu')))\n",
    "modelC2D_3.add(Dropout(0.5))\n",
    "\n",
    "# Add second fully connected layer with 50% connections dropped out\n",
    "modelC2D_3.add(TimeDistributed(Dense(128, activation='relu')))\n",
    "modelC2D_3.add(Dropout(0.5))\n",
    "\n",
    "# Add a LSTM layer\n",
    "modelC2D_3.add(LSTM(256, return_sequences=False, dropout=0.5))\n",
    "\n",
    "# Softmax layer for final 5-class output\n",
    "modelC2D_3.add(Dense(5, activation='softmax'))\n",
    "modelC2D_3.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_22 (TimeDis (None, 10, 60, 60, 16)    2368      \n",
      "_________________________________________________________________\n",
      "batch_normalization_11 (Batc (None, 10, 60, 60, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_23 (TimeDis (None, 10, 58, 58, 16)    2320      \n",
      "_________________________________________________________________\n",
      "batch_normalization_12 (Batc (None, 10, 58, 58, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_24 (TimeDis (None, 10, 29, 29, 16)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_25 (TimeDis (None, 10, 29, 29, 32)    4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_13 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_26 (TimeDis (None, 10, 29, 29, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_14 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_27 (TimeDis (None, 10, 14, 14, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_28 (TimeDis (None, 10, 14, 14, 64)    18496     \n",
      "_________________________________________________________________\n",
      "batch_normalization_15 (Batc (None, 10, 14, 14, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_29 (TimeDis (None, 10, 14, 14, 64)    36928     \n",
      "_________________________________________________________________\n",
      "batch_normalization_16 (Batc (None, 10, 14, 14, 64)    256       \n",
      "_________________________________________________________________\n",
      "time_distributed_30 (TimeDis (None, 10, 7, 7, 64)      0         \n",
      "_________________________________________________________________\n",
      "time_distributed_31 (TimeDis (None, 10, 3136)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_32 (TimeDis (None, 10, 64)            200768    \n",
      "_________________________________________________________________\n",
      "dropout_5 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_33 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_6 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_3 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_9 (Dense)              (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 679,509\n",
      "Trainable params: 679,061\n",
      "Non-trainable params: 448\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "optimiser = Adam(lr = 0.0001) #write your optimizer\n",
    "modelC2D_3.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['categorical_accuracy'])\n",
    "print (modelC2D_3.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "45/45 [==============================] - 22s 490ms/step - loss: 1.6597 - categorical_accuracy: 0.2266 - val_loss: 1.5029 - val_categorical_accuracy: 0.3900\n",
      "\n",
      "Epoch 00001: saving model to model_init_2020-03-0819_30_13.737684/model-00001-1.65925-0.22474-1.50293-0.39000.h5\n",
      "Epoch 2/25\n",
      "45/45 [==============================] - 17s 375ms/step - loss: 1.5603 - categorical_accuracy: 0.2844 - val_loss: 1.4611 - val_categorical_accuracy: 0.4500\n",
      "\n",
      "Epoch 00002: saving model to model_init_2020-03-0819_30_13.737684/model-00002-1.56152-0.28356-1.46106-0.45000.h5\n",
      "Epoch 3/25\n",
      "45/45 [==============================] - 19s 430ms/step - loss: 1.4809 - categorical_accuracy: 0.3763 - val_loss: 1.3323 - val_categorical_accuracy: 0.4200\n",
      "\n",
      "Epoch 00003: saving model to model_init_2020-03-0819_30_13.737684/model-00003-1.48204-0.37707-1.33234-0.42000.h5\n",
      "Epoch 4/25\n",
      "45/45 [==============================] - 19s 432ms/step - loss: 1.3647 - categorical_accuracy: 0.4415 - val_loss: 1.1582 - val_categorical_accuracy: 0.5100\n",
      "\n",
      "Epoch 00004: saving model to model_init_2020-03-0819_30_13.737684/model-00004-1.36457-0.44344-1.15820-0.51000.h5\n",
      "Epoch 5/25\n",
      "45/45 [==============================] - 19s 430ms/step - loss: 1.2258 - categorical_accuracy: 0.5187 - val_loss: 1.1886 - val_categorical_accuracy: 0.5300\n",
      "\n",
      "Epoch 00005: saving model to model_init_2020-03-0819_30_13.737684/model-00005-1.22038-0.52790-1.18859-0.53000.h5\n",
      "Epoch 6/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 1.1290 - categorical_accuracy: 0.5437 - val_loss: 1.1984 - val_categorical_accuracy: 0.5300\n",
      "\n",
      "Epoch 00006: saving model to model_init_2020-03-0819_30_13.737684/model-00006-1.12587-0.54148-1.19837-0.53000.h5\n",
      "\n",
      "Epoch 00006: ReduceLROnPlateau reducing learning rate to 4.999999873689376e-05.\n",
      "Epoch 7/25\n",
      "45/45 [==============================] - 18s 410ms/step - loss: 1.0136 - categorical_accuracy: 0.6133 - val_loss: 0.9824 - val_categorical_accuracy: 0.6400\n",
      "\n",
      "Epoch 00007: saving model to model_init_2020-03-0819_30_13.737684/model-00007-1.01494-0.61237-0.98242-0.64000.h5\n",
      "Epoch 8/25\n",
      "45/45 [==============================] - 19s 412ms/step - loss: 0.9242 - categorical_accuracy: 0.6756 - val_loss: 0.9652 - val_categorical_accuracy: 0.6300\n",
      "\n",
      "Epoch 00008: saving model to model_init_2020-03-0819_30_13.737684/model-00008-0.91736-0.67572-0.96523-0.63000.h5\n",
      "Epoch 9/25\n",
      "45/45 [==============================] - 19s 416ms/step - loss: 0.8539 - categorical_accuracy: 0.6815 - val_loss: 0.8542 - val_categorical_accuracy: 0.6500\n",
      "\n",
      "Epoch 00009: saving model to model_init_2020-03-0819_30_13.737684/model-00009-0.85207-0.68175-0.85422-0.65000.h5\n",
      "Epoch 10/25\n",
      "45/45 [==============================] - 19s 412ms/step - loss: 0.7977 - categorical_accuracy: 0.7229 - val_loss: 0.9350 - val_categorical_accuracy: 0.6400\n",
      "\n",
      "Epoch 00010: saving model to model_init_2020-03-0819_30_13.737684/model-00010-0.80391-0.71795-0.93498-0.64000.h5\n",
      "Epoch 11/25\n",
      "45/45 [==============================] - 19s 414ms/step - loss: 0.7237 - categorical_accuracy: 0.7466 - val_loss: 0.8492 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00011: saving model to model_init_2020-03-0819_30_13.737684/model-00011-0.72903-0.74208-0.84921-0.68000.h5\n",
      "Epoch 12/25\n",
      "45/45 [==============================] - 18s 408ms/step - loss: 0.6562 - categorical_accuracy: 0.7793 - val_loss: 0.7802 - val_categorical_accuracy: 0.6900\n",
      "\n",
      "Epoch 00012: saving model to model_init_2020-03-0819_30_13.737684/model-00012-0.65917-0.78130-0.78018-0.69000.h5\n",
      "Epoch 13/25\n",
      "45/45 [==============================] - 19s 417ms/step - loss: 0.6253 - categorical_accuracy: 0.7794 - val_loss: 0.9149 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00013: saving model to model_init_2020-03-0819_30_13.737684/model-00013-0.61367-0.78733-0.91488-0.68000.h5\n",
      "Epoch 14/25\n",
      "45/45 [==============================] - 19s 422ms/step - loss: 0.5710 - categorical_accuracy: 0.8015 - val_loss: 0.8499 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00014: saving model to model_init_2020-03-0819_30_13.737684/model-00014-0.56699-0.80392-0.84989-0.75000.h5\n",
      "\n",
      "Epoch 00014: ReduceLROnPlateau reducing learning rate to 2.499999936844688e-05.\n",
      "Epoch 15/25\n",
      "45/45 [==============================] - 19s 417ms/step - loss: 0.4807 - categorical_accuracy: 0.8488 - val_loss: 1.0401 - val_categorical_accuracy: 0.6000\n",
      "\n",
      "Epoch 00015: saving model to model_init_2020-03-0819_30_13.737684/model-00015-0.48103-0.84615-1.04014-0.60000.h5\n",
      "Epoch 16/25\n",
      "45/45 [==============================] - 19s 415ms/step - loss: 0.4692 - categorical_accuracy: 0.8475 - val_loss: 0.7797 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00016: saving model to model_init_2020-03-0819_30_13.737684/model-00016-0.46670-0.85068-0.77974-0.75000.h5\n",
      "Epoch 17/25\n",
      "45/45 [==============================] - 19s 420ms/step - loss: 0.4474 - categorical_accuracy: 0.8562 - val_loss: 0.7682 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00017: saving model to model_init_2020-03-0819_30_13.737684/model-00017-0.45018-0.85370-0.76820-0.75000.h5\n",
      "Epoch 18/25\n",
      "45/45 [==============================] - 19s 413ms/step - loss: 0.4113 - categorical_accuracy: 0.8696 - val_loss: 0.9426 - val_categorical_accuracy: 0.6500\n",
      "\n",
      "Epoch 00018: saving model to model_init_2020-03-0819_30_13.737684/model-00018-0.41497-0.86727-0.94261-0.65000.h5\n",
      "Epoch 19/25\n",
      "45/45 [==============================] - 19s 421ms/step - loss: 0.3892 - categorical_accuracy: 0.8785 - val_loss: 1.0332 - val_categorical_accuracy: 0.6400\n",
      "\n",
      "Epoch 00019: saving model to model_init_2020-03-0819_30_13.737684/model-00019-0.39327-0.87632-1.03318-0.64000.h5\n",
      "\n",
      "Epoch 00019: ReduceLROnPlateau reducing learning rate to 1.249999968422344e-05.\n",
      "Epoch 20/25\n",
      "45/45 [==============================] - 19s 411ms/step - loss: 0.3646 - categorical_accuracy: 0.8770 - val_loss: 0.7693 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00020: saving model to model_init_2020-03-0819_30_13.737684/model-00020-0.36546-0.87481-0.76930-0.72000.h5\n",
      "Epoch 21/25\n",
      "45/45 [==============================] - 19s 415ms/step - loss: 0.3553 - categorical_accuracy: 0.8830 - val_loss: 1.0087 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00021: saving model to model_init_2020-03-0819_30_13.737684/model-00021-0.35503-0.88688-1.00866-0.70000.h5\n",
      "\n",
      "Epoch 00021: ReduceLROnPlateau reducing learning rate to 6.24999984211172e-06.\n",
      "Epoch 22/25\n",
      "45/45 [==============================] - 19s 413ms/step - loss: 0.3336 - categorical_accuracy: 0.9081 - val_loss: 0.6816 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00022: saving model to model_init_2020-03-0819_30_13.737684/model-00022-0.33775-0.90649-0.68164-0.75000.h5\n",
      "Epoch 23/25\n",
      "45/45 [==============================] - 19s 425ms/step - loss: 0.3190 - categorical_accuracy: 0.9081 - val_loss: 0.9760 - val_categorical_accuracy: 0.6200\n",
      "\n",
      "Epoch 00023: saving model to model_init_2020-03-0819_30_13.737684/model-00023-0.31870-0.90649-0.97595-0.62000.h5\n",
      "Epoch 24/25\n",
      "45/45 [==============================] - 19s 419ms/step - loss: 0.3288 - categorical_accuracy: 0.8992 - val_loss: 0.7759 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00024: saving model to model_init_2020-03-0819_30_13.737684/model-00024-0.32898-0.89744-0.77592-0.72000.h5\n",
      "\n",
      "Epoch 00024: ReduceLROnPlateau reducing learning rate to 3.12499992105586e-06.\n",
      "Epoch 25/25\n",
      "45/45 [==============================] - 19s 422ms/step - loss: 0.2991 - categorical_accuracy: 0.9201 - val_loss: 0.9030 - val_categorical_accuracy: 0.7400\n",
      "\n",
      "Epoch 00025: saving model to model_init_2020-03-0819_30_13.737684/model-00025-0.29575-0.92459-0.90296-0.74000.h5\n"
     ]
    }
   ],
   "source": [
    "H3 = modelC2D_3.fit_generator(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                    callbacks=callbacks_list, validation_data=val_generator, \n",
    "                    validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the training loss and accuracy\n",
    "N = num_epochs\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(15,5))\n",
    "\n",
    "# summarize history for Accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(np.arange(1, N+1), H3.history[\"categorical_accuracy\"], label=\"train_acc\")\n",
    "plt.plot(np.arange(1, N+1), H3.history[\"val_categorical_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"modelC2D_3:Training and Validation Accuracy on Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(np.arange(1, N+1), H3.history['loss'], label=\"train_loss\")\n",
    "plt.plot(np.arange(1, N+1), H3.history['val_loss'], label=\"validation_loss\")\n",
    "plt.title(\"modelC2D_3:Training and Validation Loss on Dataset\")\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model 4: TimeDistributed Conv2d + RNN Model (CNN + RNN) : Reduced Filter size, reduced layers"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_34 (TimeDis (None, 10, 60, 60, 16)    2368      \n",
      "_________________________________________________________________\n",
      "batch_normalization_17 (Batc (None, 10, 60, 60, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_35 (TimeDis (None, 10, 58, 58, 16)    2320      \n",
      "_________________________________________________________________\n",
      "batch_normalization_18 (Batc (None, 10, 58, 58, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_36 (TimeDis (None, 10, 29, 29, 16)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_37 (TimeDis (None, 10, 29, 29, 32)    4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_19 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_38 (TimeDis (None, 10, 29, 29, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_20 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_39 (TimeDis (None, 10, 14, 14, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_40 (TimeDis (None, 10, 6272)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_41 (TimeDis (None, 10, 64)            401472    \n",
      "_________________________________________________________________\n",
      "dropout_7 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_42 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_4 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 824,277\n",
      "Trainable params: 824,085\n",
      "Non-trainable params: 192\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Model 4 : Conv2D + RNN with LSTM\n",
    "\n",
    "# Initial layers take the image nD array as input\n",
    "modelC2D_4 = Sequential()\n",
    "# 16 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_4.add(TimeDistributed(Conv2D(16, (7, 7), strides=(2, 2), activation='relu', padding='same'), input_shape=(10, 120, 120, 3)))\n",
    "modelC2D_4.add(BatchNormalization())\n",
    "# 16 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_4.add(TimeDistributed(Conv2D(16, (3,3), kernel_initializer=\"he_normal\", activation='relu')))\n",
    "modelC2D_4.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_4.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_4.add(TimeDistributed(Conv2D(32, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_4.add(BatchNormalization())\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC2D_4.add(TimeDistributed(Conv2D(32, (3,3), padding='same', activation='relu')))\n",
    "modelC2D_4.add(BatchNormalization())\n",
    "# Max Pooling Layer\n",
    "modelC2D_4.add(TimeDistributed(MaxPooling2D((2, 2), strides=(2, 2))))\n",
    "\n",
    "# Flatten the feature map into Vector\n",
    "modelC2D_4.add(TimeDistributed(Flatten()))\n",
    "\n",
    "# Add a fully connected layer with 50% connections dropped out\n",
    "modelC2D_4.add(TimeDistributed(Dense(64, activation='relu')))\n",
    "modelC2D_4.add(Dropout(0.5))\n",
    "\n",
    "# Add second fully connected layer with 50% connections dropped out\n",
    "modelC2D_4.add(TimeDistributed(Dense(128, activation='relu')))\n",
    "modelC2D_4.add(Dropout(0.5))\n",
    "\n",
    "# Add a LSTM layer\n",
    "modelC2D_4.add(LSTM(256, return_sequences=False, dropout=0.5))\n",
    "\n",
    "# Softmax layer for final 5-class output\n",
    "modelC2D_4.add(Dense(5, activation='softmax'))\n",
    "modelC2D_4.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "time_distributed_34 (TimeDis (None, 10, 60, 60, 16)    2368      \n",
      "_________________________________________________________________\n",
      "batch_normalization_17 (Batc (None, 10, 60, 60, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_35 (TimeDis (None, 10, 58, 58, 16)    2320      \n",
      "_________________________________________________________________\n",
      "batch_normalization_18 (Batc (None, 10, 58, 58, 16)    64        \n",
      "_________________________________________________________________\n",
      "time_distributed_36 (TimeDis (None, 10, 29, 29, 16)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_37 (TimeDis (None, 10, 29, 29, 32)    4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_19 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_38 (TimeDis (None, 10, 29, 29, 32)    9248      \n",
      "_________________________________________________________________\n",
      "batch_normalization_20 (Batc (None, 10, 29, 29, 32)    128       \n",
      "_________________________________________________________________\n",
      "time_distributed_39 (TimeDis (None, 10, 14, 14, 32)    0         \n",
      "_________________________________________________________________\n",
      "time_distributed_40 (TimeDis (None, 10, 6272)          0         \n",
      "_________________________________________________________________\n",
      "time_distributed_41 (TimeDis (None, 10, 64)            401472    \n",
      "_________________________________________________________________\n",
      "dropout_7 (Dropout)          (None, 10, 64)            0         \n",
      "_________________________________________________________________\n",
      "time_distributed_42 (TimeDis (None, 10, 128)           8320      \n",
      "_________________________________________________________________\n",
      "dropout_8 (Dropout)          (None, 10, 128)           0         \n",
      "_________________________________________________________________\n",
      "lstm_4 (LSTM)                (None, 256)               394240    \n",
      "_________________________________________________________________\n",
      "dense_12 (Dense)             (None, 5)                 1285      \n",
      "=================================================================\n",
      "Total params: 824,277\n",
      "Trainable params: 824,085\n",
      "Non-trainable params: 192\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "optimiser = Adam(lr = 0.0001) #write your optimizer\n",
    "modelC2D_4.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['categorical_accuracy'])\n",
    "print (modelC2D_4.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "45/45 [==============================] - 22s 485ms/step - loss: 1.6288 - categorical_accuracy: 0.2265 - val_loss: 1.5332 - val_categorical_accuracy: 0.3500\n",
      "\n",
      "Epoch 00001: saving model to model_init_2020-03-0819_30_13.737684/model-00001-1.63330-0.21870-1.53320-0.35000.h5\n",
      "Epoch 2/25\n",
      "45/45 [==============================] - 17s 372ms/step - loss: 1.5795 - categorical_accuracy: 0.2623 - val_loss: 1.3845 - val_categorical_accuracy: 0.4800\n",
      "\n",
      "Epoch 00002: saving model to model_init_2020-03-0819_30_13.737684/model-00002-1.57773-0.26697-1.38454-0.48000.h5\n",
      "Epoch 3/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 1.4679 - categorical_accuracy: 0.3689 - val_loss: 1.3211 - val_categorical_accuracy: 0.5200\n",
      "\n",
      "Epoch 00003: saving model to model_init_2020-03-0819_30_13.737684/model-00003-1.46785-0.36953-1.32107-0.52000.h5\n",
      "Epoch 4/25\n",
      "45/45 [==============================] - 19s 425ms/step - loss: 1.3588 - categorical_accuracy: 0.4209 - val_loss: 1.5195 - val_categorical_accuracy: 0.4500\n",
      "\n",
      "Epoch 00004: saving model to model_init_2020-03-0819_30_13.737684/model-00004-1.35835-0.42836-1.51946-0.45000.h5\n",
      "Epoch 5/25\n",
      "45/45 [==============================] - 19s 428ms/step - loss: 1.2149 - categorical_accuracy: 0.5109 - val_loss: 1.1947 - val_categorical_accuracy: 0.6100\n",
      "\n",
      "Epoch 00005: saving model to model_init_2020-03-0819_30_13.737684/model-00005-1.22104-0.50226-1.19469-0.61000.h5\n",
      "Epoch 6/25\n",
      "45/45 [==============================] - 20s 433ms/step - loss: 1.1547 - categorical_accuracy: 0.5660 - val_loss: 1.1817 - val_categorical_accuracy: 0.5400\n",
      "\n",
      "Epoch 00006: saving model to model_init_2020-03-0819_30_13.737684/model-00006-1.13802-0.57014-1.18168-0.54000.h5\n",
      "Epoch 7/25\n",
      "45/45 [==============================] - 20s 436ms/step - loss: 1.0048 - categorical_accuracy: 0.6326 - val_loss: 1.1319 - val_categorical_accuracy: 0.5800\n",
      "\n",
      "Epoch 00007: saving model to model_init_2020-03-0819_30_13.737684/model-00007-1.00365-0.63198-1.13191-0.58000.h5\n",
      "Epoch 8/25\n",
      "45/45 [==============================] - 19s 432ms/step - loss: 0.8840 - categorical_accuracy: 0.6978 - val_loss: 0.9777 - val_categorical_accuracy: 0.6000\n",
      "\n",
      "Epoch 00008: saving model to model_init_2020-03-0819_30_13.737684/model-00008-0.88533-0.69834-0.97767-0.60000.h5\n",
      "Epoch 9/25\n",
      "45/45 [==============================] - 19s 432ms/step - loss: 0.7546 - categorical_accuracy: 0.7392 - val_loss: 0.9987 - val_categorical_accuracy: 0.6300\n",
      "\n",
      "Epoch 00009: saving model to model_init_2020-03-0819_30_13.737684/model-00009-0.76022-0.73454-0.99865-0.63000.h5\n",
      "Epoch 10/25\n",
      "45/45 [==============================] - 20s 444ms/step - loss: 0.6423 - categorical_accuracy: 0.7956 - val_loss: 0.9730 - val_categorical_accuracy: 0.6300\n",
      "\n",
      "Epoch 00010: saving model to model_init_2020-03-0819_30_13.737684/model-00010-0.63870-0.79789-0.97299-0.63000.h5\n",
      "Epoch 11/25\n",
      "45/45 [==============================] - 19s 425ms/step - loss: 0.5202 - categorical_accuracy: 0.8325 - val_loss: 0.9880 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00011: saving model to model_init_2020-03-0819_30_13.737684/model-00011-0.52758-0.82956-0.98796-0.70000.h5\n",
      "Epoch 12/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.4763 - categorical_accuracy: 0.8579 - val_loss: 0.9981 - val_categorical_accuracy: 0.6700\n",
      "\n",
      "Epoch 00012: saving model to model_init_2020-03-0819_30_13.737684/model-00012-0.44528-0.86124-0.99811-0.67000.h5\n",
      "\n",
      "Epoch 00012: ReduceLROnPlateau reducing learning rate to 4.999999873689376e-05.\n",
      "Epoch 13/25\n",
      "45/45 [==============================] - 19s 432ms/step - loss: 0.3581 - categorical_accuracy: 0.8830 - val_loss: 1.0366 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00013: saving model to model_init_2020-03-0819_30_13.737684/model-00013-0.35539-0.88688-1.03662-0.71000.h5\n",
      "Epoch 14/25\n",
      "45/45 [==============================] - 19s 427ms/step - loss: 0.3052 - categorical_accuracy: 0.9111 - val_loss: 1.0399 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00014: saving model to model_init_2020-03-0819_30_13.737684/model-00014-0.30812-0.90950-1.03992-0.71000.h5\n",
      "\n",
      "Epoch 00014: ReduceLROnPlateau reducing learning rate to 2.499999936844688e-05.\n",
      "Epoch 15/25\n",
      "45/45 [==============================] - 19s 432ms/step - loss: 0.2511 - categorical_accuracy: 0.9259 - val_loss: 0.9783 - val_categorical_accuracy: 0.6700\n",
      "\n",
      "Epoch 00015: saving model to model_init_2020-03-0819_30_13.737684/model-00015-0.25233-0.92459-0.97830-0.67000.h5\n",
      "Epoch 16/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 0.2203 - categorical_accuracy: 0.9481 - val_loss: 0.9972 - val_categorical_accuracy: 0.7300\n",
      "\n",
      "Epoch 00016: saving model to model_init_2020-03-0819_30_13.737684/model-00016-0.21705-0.94721-0.99718-0.73000.h5\n",
      "\n",
      "Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.249999968422344e-05.\n",
      "Epoch 17/25\n",
      "45/45 [==============================] - 20s 435ms/step - loss: 0.2236 - categorical_accuracy: 0.9437 - val_loss: 0.9405 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00017: saving model to model_init_2020-03-0819_30_13.737684/model-00017-0.22476-0.94268-0.94055-0.72000.h5\n",
      "Epoch 18/25\n",
      "45/45 [==============================] - 20s 434ms/step - loss: 0.2390 - categorical_accuracy: 0.9289 - val_loss: 0.9877 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00018: saving model to model_init_2020-03-0819_30_13.737684/model-00018-0.23903-0.92760-0.98767-0.71000.h5\n",
      "Epoch 19/25\n",
      "45/45 [==============================] - 20s 434ms/step - loss: 0.2182 - categorical_accuracy: 0.9481 - val_loss: 0.9832 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00019: saving model to model_init_2020-03-0819_30_13.737684/model-00019-0.21895-0.94721-0.98320-0.72000.h5\n",
      "\n",
      "Epoch 00019: ReduceLROnPlateau reducing learning rate to 6.24999984211172e-06.\n",
      "Epoch 20/25\n",
      "45/45 [==============================] - 19s 426ms/step - loss: 0.2029 - categorical_accuracy: 0.9496 - val_loss: 0.9921 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00020: saving model to model_init_2020-03-0819_30_13.737684/model-00020-0.20342-0.94872-0.99211-0.75000.h5\n",
      "Epoch 21/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 0.2029 - categorical_accuracy: 0.9466 - val_loss: 1.0060 - val_categorical_accuracy: 0.7100\n",
      "\n",
      "Epoch 00021: saving model to model_init_2020-03-0819_30_13.737684/model-00021-0.19925-0.94570-1.00604-0.71000.h5\n",
      "\n",
      "Epoch 00021: ReduceLROnPlateau reducing learning rate to 3.12499992105586e-06.\n",
      "Epoch 22/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.1842 - categorical_accuracy: 0.9600 - val_loss: 0.9613 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00022: saving model to model_init_2020-03-0819_30_13.737684/model-00022-0.18519-0.95928-0.96129-0.72000.h5\n",
      "Epoch 23/25\n",
      "45/45 [==============================] - 20s 440ms/step - loss: 0.2002 - categorical_accuracy: 0.9466 - val_loss: 0.9879 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00023: saving model to model_init_2020-03-0819_30_13.737684/model-00023-0.20305-0.94570-0.98789-0.70000.h5\n",
      "\n",
      "Epoch 00023: ReduceLROnPlateau reducing learning rate to 1.56249996052793e-06.\n",
      "Epoch 24/25\n",
      "45/45 [==============================] - 20s 439ms/step - loss: 0.1832 - categorical_accuracy: 0.9526 - val_loss: 0.9950 - val_categorical_accuracy: 0.7200\n",
      "\n",
      "Epoch 00024: saving model to model_init_2020-03-0819_30_13.737684/model-00024-0.18467-0.95173-0.99504-0.72000.h5\n",
      "Epoch 25/25\n",
      "45/45 [==============================] - 20s 440ms/step - loss: 0.1621 - categorical_accuracy: 0.9600 - val_loss: 1.0165 - val_categorical_accuracy: 0.6900\n",
      "\n",
      "Epoch 00025: saving model to model_init_2020-03-0819_30_13.737684/model-00025-0.16337-0.95928-1.01648-0.69000.h5\n",
      "\n",
      "Epoch 00025: ReduceLROnPlateau reducing learning rate to 7.81249980263965e-07.\n"
     ]
    }
   ],
   "source": [
    "H4 = modelC2D_4.fit_generator(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                    callbacks=callbacks_list, validation_data=val_generator, \n",
    "                    validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the training loss and accuracy\n",
    "N = num_epochs\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(15,5))\n",
    "\n",
    "# summarize history for Accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(np.arange(1, N+1), H4.history[\"categorical_accuracy\"], label=\"train_acc\")\n",
    "plt.plot(np.arange(1, N+1), H4.history[\"val_categorical_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"modelC2D_4:Training and Validation Accuracy on Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(np.arange(1, N+1), H4.history['loss'], label=\"train_loss\")\n",
    "plt.plot(np.arange(1, N+1), H4.history['val_loss'], label=\"validation_loss\")\n",
    "plt.title(\"modelC2D_4:Training and Validation Loss on Dataset\")\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Model 5: Conv3D Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv3d_1 (Conv3D)            (None, 10, 120, 120, 8)   656       \n",
      "_________________________________________________________________\n",
      "batch_normalization_21 (Batc (None, 10, 120, 120, 8)   32        \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 10, 120, 120, 8)   0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_1 (MaxPooling3 (None, 5, 60, 60, 8)      0         \n",
      "_________________________________________________________________\n",
      "conv3d_2 (Conv3D)            (None, 5, 60, 60, 16)     3472      \n",
      "_________________________________________________________________\n",
      "batch_normalization_22 (Batc (None, 5, 60, 60, 16)     64        \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 5, 60, 60, 16)     0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_2 (MaxPooling3 (None, 2, 30, 30, 16)     0         \n",
      "_________________________________________________________________\n",
      "conv3d_3 (Conv3D)            (None, 2, 30, 30, 32)     4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_23 (Batc (None, 2, 30, 30, 32)     128       \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 2, 30, 30, 32)     0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_3 (MaxPooling3 (None, 1, 15, 15, 32)     0         \n",
      "_________________________________________________________________\n",
      "flatten_5 (Flatten)          (None, 7200)              0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 256)               1843456   \n",
      "_________________________________________________________________\n",
      "dropout_9 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 512)               131584    \n",
      "_________________________________________________________________\n",
      "dropout_10 (Dropout)         (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 5)                 2565      \n",
      "=================================================================\n",
      "Total params: 1,986,597\n",
      "Trainable params: 1,986,485\n",
      "Non-trainable params: 112\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Model 5: Conv3d with 3-Conv layers + 2 Fully Connected\n",
    "\n",
    "modelC3D_1 = Sequential()\n",
    "\n",
    "# Initial layers take the image nD array as input\n",
    "# 8 Filters Conv3D layer with Batch Norm\n",
    "modelC3D_1.add(Conv3D(8, kernel_size=(3,3,3), input_shape=(10, 120, 120, 3), padding='same'))\n",
    "modelC3D_1.add(BatchNormalization())\n",
    "modelC3D_1.add(Activation('relu'))\n",
    "\n",
    "modelC3D_1.add(MaxPooling3D(pool_size=(2,2,2)))\n",
    "\n",
    "# 16 Filters Conv3D layer with Batch Norm\n",
    "modelC3D_1.add(Conv3D(16, kernel_size=(3,3,3), padding='same'))\n",
    "modelC3D_1.add(BatchNormalization())\n",
    "modelC3D_1.add(Activation('relu'))\n",
    "# Max Pooling Layer\n",
    "modelC3D_1.add(MaxPooling3D(pool_size=(2,2,2)))\n",
    "\n",
    "# 32 Filters Conv2D layer with Batch Norm\n",
    "modelC3D_1.add(Conv3D(32, kernel_size=(1,3,3), padding='same'))\n",
    "modelC3D_1.add(BatchNormalization())\n",
    "modelC3D_1.add(Activation('relu'))\n",
    "# Max Pooling Layer\n",
    "modelC3D_1.add(MaxPooling3D(pool_size=(2,2,2)))\n",
    "\n",
    "# Flatten the feature map into Vector\n",
    "modelC3D_1.add(Flatten())\n",
    "\n",
    "# Add a fully connected layer with 25% connections dropped out\n",
    "modelC3D_1.add(Dense(256, activation='relu'))\n",
    "modelC3D_1.add(Dropout(0.25))\n",
    "\n",
    "# Add second fully connected layer with 25% connections dropped out\n",
    "modelC3D_1.add(Dense(512, activation='relu'))\n",
    "modelC3D_1.add(Dropout(0.25))\n",
    "\n",
    "# Softmax layer for final 5-class output\n",
    "modelC3D_1.add(Dense(5, activation='softmax'))\n",
    "\n",
    "modelC3D_1.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv3d_1 (Conv3D)            (None, 10, 120, 120, 8)   656       \n",
      "_________________________________________________________________\n",
      "batch_normalization_21 (Batc (None, 10, 120, 120, 8)   32        \n",
      "_________________________________________________________________\n",
      "activation_1 (Activation)    (None, 10, 120, 120, 8)   0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_1 (MaxPooling3 (None, 5, 60, 60, 8)      0         \n",
      "_________________________________________________________________\n",
      "conv3d_2 (Conv3D)            (None, 5, 60, 60, 16)     3472      \n",
      "_________________________________________________________________\n",
      "batch_normalization_22 (Batc (None, 5, 60, 60, 16)     64        \n",
      "_________________________________________________________________\n",
      "activation_2 (Activation)    (None, 5, 60, 60, 16)     0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_2 (MaxPooling3 (None, 2, 30, 30, 16)     0         \n",
      "_________________________________________________________________\n",
      "conv3d_3 (Conv3D)            (None, 2, 30, 30, 32)     4640      \n",
      "_________________________________________________________________\n",
      "batch_normalization_23 (Batc (None, 2, 30, 30, 32)     128       \n",
      "_________________________________________________________________\n",
      "activation_3 (Activation)    (None, 2, 30, 30, 32)     0         \n",
      "_________________________________________________________________\n",
      "max_pooling3d_3 (MaxPooling3 (None, 1, 15, 15, 32)     0         \n",
      "_________________________________________________________________\n",
      "flatten_5 (Flatten)          (None, 7200)              0         \n",
      "_________________________________________________________________\n",
      "dense_13 (Dense)             (None, 256)               1843456   \n",
      "_________________________________________________________________\n",
      "dropout_9 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_14 (Dense)             (None, 512)               131584    \n",
      "_________________________________________________________________\n",
      "dropout_10 (Dropout)         (None, 512)               0         \n",
      "_________________________________________________________________\n",
      "dense_15 (Dense)             (None, 5)                 2565      \n",
      "=================================================================\n",
      "Total params: 1,986,597\n",
      "Trainable params: 1,986,485\n",
      "Non-trainable params: 112\n",
      "_________________________________________________________________\n",
      "None\n"
     ]
    }
   ],
   "source": [
    "optimiser = Adam() #write your optimizer\n",
    "modelC3D_1.compile(optimizer=optimiser, loss='categorical_crossentropy', metrics=['categorical_accuracy'])\n",
    "print (modelC3D_1.summary())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/25\n",
      "45/45 [==============================] - 21s 470ms/step - loss: 4.3797 - categorical_accuracy: 0.3111 - val_loss: 6.2599 - val_categorical_accuracy: 0.2400\n",
      "\n",
      "Epoch 00001: saving model to model_init_2020-03-0819_30_13.737684/model-00001-4.41033-0.31071-6.25991-0.24000.h5\n",
      "Epoch 2/25\n",
      "45/45 [==============================] - 17s 382ms/step - loss: 1.2645 - categorical_accuracy: 0.4725 - val_loss: 1.4469 - val_categorical_accuracy: 0.4400\n",
      "\n",
      "Epoch 00002: saving model to model_init_2020-03-0819_30_13.737684/model-00002-1.27342-0.46908-1.44689-0.44000.h5\n",
      "Epoch 3/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 1.1615 - categorical_accuracy: 0.5187 - val_loss: 1.7559 - val_categorical_accuracy: 0.3900\n",
      "\n",
      "Epoch 00003: saving model to model_init_2020-03-0819_30_13.737684/model-00003-1.14739-0.52790-1.75590-0.39000.h5\n",
      "Epoch 4/25\n",
      "45/45 [==============================] - 19s 427ms/step - loss: 0.9778 - categorical_accuracy: 0.6000 - val_loss: 1.1367 - val_categorical_accuracy: 0.5600\n",
      "\n",
      "Epoch 00004: saving model to model_init_2020-03-0819_30_13.737684/model-00004-0.97910-0.59879-1.13670-0.56000.h5\n",
      "Epoch 5/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 0.9887 - categorical_accuracy: 0.6252 - val_loss: 0.8269 - val_categorical_accuracy: 0.6900\n",
      "\n",
      "Epoch 00005: saving model to model_init_2020-03-0819_30_13.737684/model-00005-0.98787-0.62443-0.82694-0.69000.h5\n",
      "Epoch 6/25\n",
      "45/45 [==============================] - 19s 426ms/step - loss: 0.8510 - categorical_accuracy: 0.6699 - val_loss: 1.4219 - val_categorical_accuracy: 0.4900\n",
      "\n",
      "Epoch 00006: saving model to model_init_2020-03-0819_30_13.737684/model-00006-0.83355-0.68175-1.42187-0.49000.h5\n",
      "Epoch 7/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 0.8439 - categorical_accuracy: 0.6490 - val_loss: 0.9670 - val_categorical_accuracy: 0.6300\n",
      "\n",
      "Epoch 00007: saving model to model_init_2020-03-0819_30_13.737684/model-00007-0.84104-0.65460-0.96702-0.63000.h5\n",
      "\n",
      "Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.0005000000237487257.\n",
      "Epoch 8/25\n",
      "45/45 [==============================] - 19s 426ms/step - loss: 0.6370 - categorical_accuracy: 0.7440 - val_loss: 0.8084 - val_categorical_accuracy: 0.7300\n",
      "\n",
      "Epoch 00008: saving model to model_init_2020-03-0819_30_13.737684/model-00008-0.62550-0.75716-0.80845-0.73000.h5\n",
      "Epoch 9/25\n",
      "45/45 [==============================] - 19s 428ms/step - loss: 0.6061 - categorical_accuracy: 0.7719 - val_loss: 0.9571 - val_categorical_accuracy: 0.6500\n",
      "\n",
      "Epoch 00009: saving model to model_init_2020-03-0819_30_13.737684/model-00009-0.59980-0.77376-0.95711-0.65000.h5\n",
      "Epoch 10/25\n",
      "45/45 [==============================] - 19s 422ms/step - loss: 0.5299 - categorical_accuracy: 0.7868 - val_loss: 0.7079 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00010: saving model to model_init_2020-03-0819_30_13.737684/model-00010-0.52102-0.79487-0.70786-0.75000.h5\n",
      "Epoch 11/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 0.4734 - categorical_accuracy: 0.8091 - val_loss: 0.7883 - val_categorical_accuracy: 0.7000\n",
      "\n",
      "Epoch 00011: saving model to model_init_2020-03-0819_30_13.737684/model-00011-0.45678-0.81750-0.78827-0.70000.h5\n",
      "Epoch 12/25\n",
      "45/45 [==============================] - 19s 430ms/step - loss: 0.4841 - categorical_accuracy: 0.8076 - val_loss: 0.9584 - val_categorical_accuracy: 0.6800\n",
      "\n",
      "Epoch 00012: saving model to model_init_2020-03-0819_30_13.737684/model-00012-0.46991-0.81599-0.95839-0.68000.h5\n",
      "\n",
      "Epoch 00012: ReduceLROnPlateau reducing learning rate to 0.0002500000118743628.\n",
      "Epoch 13/25\n",
      "45/45 [==============================] - 19s 421ms/step - loss: 0.4374 - categorical_accuracy: 0.8313 - val_loss: 0.8370 - val_categorical_accuracy: 0.6600\n",
      "\n",
      "Epoch 00013: saving model to model_init_2020-03-0819_30_13.737684/model-00013-0.42719-0.84012-0.83699-0.66000.h5\n",
      "Epoch 14/25\n",
      "45/45 [==============================] - 20s 435ms/step - loss: 0.3636 - categorical_accuracy: 0.8562 - val_loss: 0.7904 - val_categorical_accuracy: 0.7300\n",
      "\n",
      "Epoch 00014: saving model to model_init_2020-03-0819_30_13.737684/model-00014-0.36963-0.85370-0.79040-0.73000.h5\n",
      "\n",
      "Epoch 00014: ReduceLROnPlateau reducing learning rate to 0.0001250000059371814.\n",
      "Epoch 15/25\n",
      "45/45 [==============================] - 19s 428ms/step - loss: 0.2716 - categorical_accuracy: 0.8934 - val_loss: 0.6270 - val_categorical_accuracy: 0.8200\n",
      "\n",
      "Epoch 00015: saving model to model_init_2020-03-0819_30_13.737684/model-00015-0.26885-0.89744-0.62697-0.82000.h5\n",
      "Epoch 16/25\n",
      "45/45 [==============================] - 20s 435ms/step - loss: 0.2572 - categorical_accuracy: 0.9111 - val_loss: 0.7273 - val_categorical_accuracy: 0.8200\n",
      "\n",
      "Epoch 00016: saving model to model_init_2020-03-0819_30_13.737684/model-00016-0.26134-0.90950-0.72735-0.82000.h5\n",
      "Epoch 17/25\n",
      "45/45 [==============================] - 20s 438ms/step - loss: 0.3026 - categorical_accuracy: 0.8859 - val_loss: 0.7324 - val_categorical_accuracy: 0.8100\n",
      "\n",
      "Epoch 00017: saving model to model_init_2020-03-0819_30_13.737684/model-00017-0.30595-0.88386-0.73239-0.81000.h5\n",
      "\n",
      "Epoch 00017: ReduceLROnPlateau reducing learning rate to 6.25000029685907e-05.\n",
      "Epoch 18/25\n",
      "45/45 [==============================] - 19s 421ms/step - loss: 0.2612 - categorical_accuracy: 0.8992 - val_loss: 0.8850 - val_categorical_accuracy: 0.7500\n",
      "\n",
      "Epoch 00018: saving model to model_init_2020-03-0819_30_13.737684/model-00018-0.26572-0.89744-0.88503-0.75000.h5\n",
      "Epoch 19/25\n",
      "45/45 [==============================] - 19s 431ms/step - loss: 0.2473 - categorical_accuracy: 0.9126 - val_loss: 0.6002 - val_categorical_accuracy: 0.8200\n",
      "\n",
      "Epoch 00019: saving model to model_init_2020-03-0819_30_13.737684/model-00019-0.25171-0.91101-0.60021-0.82000.h5\n",
      "Epoch 20/25\n",
      "45/45 [==============================] - 19s 427ms/step - loss: 0.2316 - categorical_accuracy: 0.9082 - val_loss: 0.8161 - val_categorical_accuracy: 0.7800\n",
      "\n",
      "Epoch 00020: saving model to model_init_2020-03-0819_30_13.737684/model-00020-0.22467-0.91252-0.81609-0.78000.h5\n",
      "Epoch 21/25\n",
      "45/45 [==============================] - 19s 430ms/step - loss: 0.2409 - categorical_accuracy: 0.9053 - val_loss: 0.7144 - val_categorical_accuracy: 0.7700\n",
      "\n",
      "Epoch 00021: saving model to model_init_2020-03-0819_30_13.737684/model-00021-0.22915-0.90950-0.71441-0.77000.h5\n",
      "\n",
      "Epoch 00021: ReduceLROnPlateau reducing learning rate to 3.125000148429535e-05.\n",
      "Epoch 22/25\n",
      "45/45 [==============================] - 19s 429ms/step - loss: 0.1887 - categorical_accuracy: 0.9303 - val_loss: 0.6330 - val_categorical_accuracy: 0.8200\n",
      "\n",
      "Epoch 00022: saving model to model_init_2020-03-0819_30_13.737684/model-00022-0.19162-0.92911-0.63296-0.82000.h5\n",
      "Epoch 23/25\n",
      "45/45 [==============================] - 19s 426ms/step - loss: 0.2269 - categorical_accuracy: 0.9170 - val_loss: 0.6506 - val_categorical_accuracy: 0.8200\n",
      "\n",
      "Epoch 00023: saving model to model_init_2020-03-0819_30_13.737684/model-00023-0.23020-0.91554-0.65061-0.82000.h5\n",
      "\n",
      "Epoch 00023: ReduceLROnPlateau reducing learning rate to 1.5625000742147677e-05.\n",
      "Epoch 24/25\n",
      "45/45 [==============================] - 19s 419ms/step - loss: 0.2434 - categorical_accuracy: 0.9068 - val_loss: 0.8036 - val_categorical_accuracy: 0.7700\n",
      "\n",
      "Epoch 00024: saving model to model_init_2020-03-0819_30_13.737684/model-00024-0.24035-0.91101-0.80364-0.77000.h5\n",
      "Epoch 25/25\n",
      "45/45 [==============================] - 19s 423ms/step - loss: 0.2140 - categorical_accuracy: 0.9259 - val_loss: 0.7048 - val_categorical_accuracy: 0.8000\n",
      "\n",
      "Epoch 00025: saving model to model_init_2020-03-0819_30_13.737684/model-00025-0.21736-0.92459-0.70481-0.80000.h5\n",
      "\n",
      "Epoch 00025: ReduceLROnPlateau reducing learning rate to 7.812500371073838e-06.\n"
     ]
    }
   ],
   "source": [
    "H5 = modelC3D_1.fit_generator(train_generator, steps_per_epoch=steps_per_epoch, epochs=num_epochs, verbose=1, \n",
    "                    callbacks=callbacks_list, validation_data=val_generator, \n",
    "                    validation_steps=validation_steps, class_weight=None, workers=1, initial_epoch=0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot the training loss and accuracy\n",
    "N = num_epochs\n",
    "plt.style.use(\"ggplot\")\n",
    "plt.figure(figsize=(15,5))\n",
    "\n",
    "# summarize history for Accuracy\n",
    "plt.subplot(121)\n",
    "plt.plot(np.arange(1, N+1), H5.history[\"categorical_accuracy\"], label=\"train_acc\")\n",
    "plt.plot(np.arange(1, N+1), H5.history[\"val_categorical_accuracy\"], label=\"val_acc\")\n",
    "plt.title(\"modelC3D_1:Training and Validation Accuracy on Dataset\")\n",
    "plt.xlabel(\"Epoch #\")\n",
    "plt.ylabel(\"Accuracy\")\n",
    "plt.legend(loc=\"lower right\")\n",
    "\n",
    "# summarize history for loss\n",
    "plt.subplot(122)\n",
    "plt.plot(np.arange(1, N+1), H5.history['loss'], label=\"train_loss\")\n",
    "plt.plot(np.arange(1, N+1), H5.history['val_loss'], label=\"validation_loss\")\n",
    "plt.title(\"modelC3D_1:Training and Validation Loss on Dataset\")\n",
    "plt.xlabel('Epoch #')\n",
    "plt.ylabel('Loss')\n",
    "plt.legend(loc='upper right')\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Conclusion"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## <font color = blue> A total of 5 different models have been tried, 4 CNN+RNN (Conv2D + LSTM)  & 1 Conv3D </font>\n",
    "### Below is the best performance accuracy achieved by each of these model at different epochs"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color = blue> 1. Model 1 : Conv2D + RNN with LSTM with 2 groups of 32 filters & 64 filters followed by two fully connected, LSTM & Softmax </font>\n",
    "##### This model has 1,276,901 trainable parameters. It achieved best performance at epoch 25\n",
    "##### Training Accuracy: 97.63% against Validation Accuracy of 71%. Training loss: 0.1297 & Validation Loss: 1.0842"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color = blue> 2. Model 2 : Conv2D + RNN with LSTM with 3 groups of 32 , 64 & 128 filters followed by two fully connected, LSTM & Softmax </font>\n",
    "##### This model has 1,097,957 trainable parameters. It achieved best performance at epoch 14\n",
    "##### Training Accuracy: 95.85% against Validation Accuracy of 79%. Training loss: 0.2101 & Validation Loss: 0.6449"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color = blue> 3. Model 3 : Conv2D + RNN with LSTM with 3 groups of 16, 32 & 64 filters followed by two fully connected, LSTM & Softmax </font>\n",
    "##### This model has 679,509 trainable parameters. It achieved best performance at epoch 22\n",
    "##### Training Accuracy: 90.81% against Validation Accuracy of 75%. Training loss: 0.3336 & Validation Loss: 0.6816"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color = blue> 4. Model 4 : Conv2D + RNN with LSTM with 2 groups of 16 & 32 filters followed by two fully connected, LSTM & Softmax </font>\n",
    "##### This model has 824,277 trainable parameters. It achieved best performance at epoch 20\n",
    "##### Training Accuracy: 94.96% against Validation Accuracy of 75%. Training loss: 0.2029 & Validation Loss: 0.9921"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### <font color = blue> 5. Model 5 : Conv3D with 2 layers of 8, 16 & 32 filters followed by two fully connected with dropouts & Softmax </font>\n",
    "##### This model has 1,986,597 trainable parameters. It achieved best performance at epoch 15\n",
    "##### Training Accuracy: 89.34% against Validation Accuracy of 82%. Training loss: 0.2716 & Validation Loss: 0.6270"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Considering the crucial factor of significant validation accuracy with least difference between the Validation & Training Accuracy (so as to avoid overfitting), Model # 5 i.e. the Conv3D model at epoch 15 has given the best result of the lot. So, we will be considering the file <font color=blue> \"model-00015-0.26885-0.89744-0.62697-0.82000.h5\" </font>  generated as a part of this epoch for the test validations as well. Although this Model # 5 has higher parameter count, but because of its more accurate results with considerably less loss values, we consider it as the best fit of all the models."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.8.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}