Train_Model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "JoGt0Ma12qUB"
   },
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "UrMeQrDZg506"
   },
   "outputs": [],
   "source": [
    "import zipfile\n",
    "\n",
    "\n",
    "zip_ref = zipfile.ZipFile(\"/content/drive/My Drive/train.zip\", 'r')\n",
    "zip_ref.extractall(\"/content\")\n",
    "zip_ref.close()\n",
    "\n",
    "zip_ref = zipfile.ZipFile(\"/content/drive/My Drive/val.zip\", 'r')\n",
    "zip_ref.extractall(\"/content\")\n",
    "zip_ref.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "TbT_SOuA2yP4"
   },
   "outputs": [],
   "source": [
    "import shutil\n",
    "import os\n",
    "import numpy as np\n",
    "import json\n",
    "import pandas as pd\n",
    "import pprint\n",
    "import time\n",
    "import cv2\n",
    "import matplotlib.pyplot as plt\n",
    "import scipy.misc\n",
    "from skimage.io import imread, imshow, imread_collection, concatenate_images\n",
    "import tensorflow as tf\n",
    "from tensorflow import keras\n",
    "from tensorflow.keras.layers import *\n",
    "from keras.initializers import glorot_uniform\n",
    "from keras import backend as K\n",
    "K.clear_session()\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WEa5nrnv20tM"
   },
   "outputs": [],
   "source": [
    "def conv_2d( X, strides, kernel, name):\n",
    "        \"\"\"2D convolution\n",
    "        Args:\n",
    "            X: 4D tensor, image or feature map\n",
    "            strides: int, stride for convolution\n",
    "            kernel: list, 3 elements - size of kernel and number of filters\n",
    "            name: string, name of layer\n",
    "        Returns:\n",
    "            X: 4D tensor, output after convolution and bias operations\n",
    "        \"\"\"\n",
    "\n",
    "        with tf.compat.v1.variable_scope(name) as scope:\n",
    "            W = tf.compat.v1.get_variable(shape=[kernel[0], kernel[1], X.shape[3], kernel[2]], \n",
    "                                dtype=tf.float32, name=name + '_weights')\n",
    "            b = tf.compat.v1.get_variable(shape=[kernel[2]], dtype=tf.float32, \n",
    "                                name=name + '_bias')\n",
    "\n",
    "        X = tf.nn.conv2d(X, W, [1, strides, strides, 1], padding='SAME', name=None)\n",
    "        X = X + b\n",
    "\n",
    "        return X"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WOrZ_wNz21ll"
   },
   "outputs": [],
   "source": [
    "def model_DRN26(input):\n",
    "  \n",
    "\n",
    "  X=Conv2D(16, (7, 7), padding='same',name='layer1_conv')(input)\n",
    "\n",
    "  #short_link=conv_2d(X, 1,[1, 1, 16],'short')\n",
    "  short_link=Conv2D(16, (1, 1),padding='same')(X)\n",
    "\n",
    "  X = Conv2D(16, (3, 3),strides=(1, 1),padding='same',activation='relu' ,name='layer2_conv')(X)\n",
    "  X = Conv2D(16, (3, 3),strides=(1, 1),padding='same',activation='relu', name='layer3_conv')(X)\n",
    "  \n",
    "  X = Add()([short_link, X])\n",
    "  #X=  tf.add(short_link, X)\n",
    "  X = Activation('relu')(X)\n",
    "  #X =  tf.nn.relu(X)\n",
    "\n",
    "#*******************************************************************************\n",
    "\n",
    "  short_link=Conv2D(20, (1, 1),padding='same')(X)\n",
    "\n",
    "  X = Conv2D(20, (3, 3),padding='same',activation='relu' ,name='layer4_conv')(X)\n",
    "  X = Conv2D(20, (3, 3),padding='same',activation='relu', name='layer5_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "  X = MaxPooling2D(pool_size=(2, 2), padding=\"valid\",name='layer4_Max')(X)\n",
    "\n",
    "#*******************************************************************************\n",
    "  #short_link=Conv2D(64, (1, 1),padding='same')(X)\n",
    "  \n",
    "  #X = Conv2D(64, (3, 3),padding='same',activation='relu' ,name='layer4_conv')(X)\n",
    "  #X = Conv2D(64, (3, 3),padding='same',activation='relu', name='layer5_conv')(X)\n",
    "  #X = Add()([short_link, X])\n",
    "  #X = Activation('relu')(X)\n",
    "\n",
    "  short_link=Conv2D(32, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(32, (3, 3),strides=(1, 1),padding='same',activation='relu' ,name='layer6_conv')(X)\n",
    "  X = Conv2D(32, (3, 3),strides=(1, 1),padding='same',activation='relu', name='layer7_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "#*******************************************************************************\n",
    "  short_link=Conv2D(64, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(64, (3, 3),padding='same',activation='relu' ,name='layer8_conv')(X)\n",
    "  X = Conv2D(64, (3, 3),padding='same',activation='relu', name='layer9_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "  #short_link=Conv2D(128, (1, 1),padding='same')(X)\n",
    "  \n",
    "  #X = Conv2D(128, (3, 3),padding='same',activation='relu' ,name='layer10_disconv')(X)\n",
    "  #X = Conv2D(128, (3, 3),padding='same',activation='relu', name='layer11_disconv')(X)\n",
    "  #X = Add()([short_link, X])\n",
    "  #X = Activation('relu')(X)\n",
    "\n",
    "#*******************************************************************************\n",
    "  X = MaxPooling2D(pool_size=(2, 2), padding=\"valid\")(X)\n",
    "  \n",
    "  short_link=Conv2D(128, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(128, (3, 3),padding='same',dilation_rate=2,activation='relu' ,name='layer12_disconv')(X)\n",
    "  X = Conv2D(128, (3, 3),padding='same',dilation_rate=2,activation='relu', name='layer13_disconv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "  short_link=Conv2D(128, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(128, (3, 3),padding='same',dilation_rate=2,activation='relu' ,name='layer14_disconv')(X)\n",
    "  X = Conv2D(128, (3, 3),padding='same',dilation_rate=2,activation='relu', name='layer15_disconv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "#*******************************************************************************\n",
    "  \n",
    "\n",
    "  short_link=Conv2D(256, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(256, (3, 3),padding='same',dilation_rate=4,activation='relu' ,name='layer16_disconv')(X)\n",
    "  X = Conv2D(256, (3, 3),padding='same',dilation_rate=4,activation='relu', name='layer17_disconv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "  short_link=Conv2D(256, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(256, (3, 3),padding='same',dilation_rate=2,activation='relu' ,name='layer18_conv')(X)\n",
    "  X = Conv2D(256, (3, 3),padding='same',dilation_rate=2,activation='relu', name='layer19_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "#*******************************************************************************\n",
    "  X = MaxPooling2D(pool_size=(2, 2), padding=\"valid\")(X)\n",
    "\n",
    "  short_link=Conv2D(128, (1, 1),padding='same')(X)\n",
    "  \n",
    " \n",
    "  X = Conv2D(128, (3, 3),padding='same',activation='relu' ,name='layer20_conv')(X)\n",
    "  X = Conv2D(128, (3, 3),padding='same',activation='relu', name='layer21_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "  #X = MaxPooling2D(pool_size=(2, 2), padding=\"valid\")(X)\n",
    "\n",
    "  short_link=Conv2D(64, (1, 1),padding='same')(X)\n",
    "  \n",
    "  X = Conv2D(64, (2, 3),padding='same',activation='relu' ,name='layer22_conv')(X)\n",
    "  X = Conv2D(64, (2, 3),padding='same',activation='relu', name='layer23_conv')(X)\n",
    "  X = Add()([short_link, X])\n",
    "  X = Activation('relu')(X)\n",
    "\n",
    "\n",
    "  return X\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DmwEpRoj24J3"
   },
   "outputs": [],
   "source": [
    "#from: width x height x channels\n",
    "#to: width x height x groups x (channels/groups)\n",
    "#prermute the last two dimensions\n",
    "#reshape the tensor to the original shape\n",
    "#https://machinelearningtokyo.com/2020/04/27/cnn-architectures-shufflenet/\n",
    "\n",
    "def channel_shuffle(x, groups):  \n",
    "    _, width, height, channels = x.get_shape().as_list()\n",
    "    group_ch = channels // groups\n",
    "\n",
    "    x = Reshape([width, height, group_ch, groups])(x)\n",
    "    x = Permute([1, 2, 4, 3])(x)\n",
    "    x = Reshape([width, height, channels])(x)\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "3sUTBuFd27S9"
   },
   "outputs": [],
   "source": [
    "#https://machinelearningtokyo.com/2020/04/27/cnn-architectures-shufflenet/\n",
    "def gconv(tensor, channels, groups):\n",
    "    input_ch = tensor.get_shape().as_list()[-1]\n",
    "    group_ch = input_ch // groups\n",
    "    output_ch = channels // groups\n",
    "    groups_list = []\n",
    " \n",
    "    for i in range(groups):\n",
    "        # group_tensor = tensor[:, :, :, i * group_ch: (i+1) * group_ch]\n",
    "        group_tensor = Lambda(lambda x: x[:, :, :, i * group_ch: (i+1) * group_ch])(tensor)\n",
    "        group_tensor = Conv2D(output_ch, 1)(group_tensor)\n",
    "        groups_list.append(group_tensor)\n",
    " \n",
    "    output = Concatenate()(groups_list)\n",
    "    return output"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_-Yqntz03ARH"
   },
   "outputs": [],
   "source": [
    "def context_flow(input_1,input_2,name_):\n",
    "  X = Concatenate()([input_1,input_2])\n",
    "  X = AveragePooling2D((2, 2),  (2, 2),padding='same')(X)\n",
    "  X = gconv(X, 64, groups=16)\n",
    "  X = channel_shuffle(X, 16)\n",
    "  X = gconv(X, 64, groups=16)\n",
    "\n",
    "  short_link=X\n",
    "\n",
    "  X = Conv2D(64, (1, 1),activation='relu',padding = 'same' ,name='context_flow_conv1_'+name_)(X)\n",
    "  X = Activation('relu')(X)\n",
    "  X = Conv2D(64, (1, 1),activation='relu' ,padding = 'same' ,name='context_flow_conv2_'+name_)(X)\n",
    "  X = Activation('sigmoid')(X)\n",
    "  \n",
    "  X = Multiply()([short_link , X])\n",
    "  X = Add()([short_link,X])\n",
    "\n",
    "  return X\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "u05wEVTB3CES"
   },
   "outputs": [],
   "source": [
    "def AGCN(input):\n",
    "  X = Conv2D(64, (1, 7), activation = 'relu', padding = 'same', name = 'AGCN_conv1' )(input)\n",
    "  X = Conv2D(64, (7, 1), activation = 'relu', padding = 'same', name = 'AGCN_conv2' )(X)\n",
    "\n",
    "  X1 = Conv2D(64, (7, 1), activation = 'relu', padding = 'same', name = 'AGCN_conv3' )(input)\n",
    "  X1 = Conv2D(64, (1, 7), activation = 'relu', padding = 'same', name = 'AGCN_conv4' )(X1)\n",
    "\n",
    "  O  = Add(name = 'add_AGCN1')([X, X1])\n",
    "\n",
    "  short_link=O\n",
    "  X = Conv2D(64, (3, 3), activation = 'relu', padding = 'same', name = 'AGCN_conv5' )(O)\n",
    "\n",
    "  O  = Add(name = 'add_AGCN2')([X, short_link])\n",
    "\n",
    "  return O"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "YQZxOtE63EXr"
   },
   "outputs": [],
   "source": [
    "def model_CAM(input):\n",
    "  X = AveragePooling2D(1,(16,16))(input)                      \n",
    "  X = BatchNormalization()(X)                               \n",
    "  X = Activation('relu')(X)    \n",
    "  X = Conv2D(64, (1, 1), activation = 'relu', padding = 'same', name = 'CAM_conv1' )(X)\n",
    "\n",
    "  X_in  = Conv2DTranspose(64 ,(71,85), use_bias = False)(X) #Conv2D(32, (1, 1),strides=(2, 2), activation = 'relu', name = 'CAM_conv2' )(input)\n",
    " \n",
    "  X1  = context_flow(X_in,input,'1')\n",
    "\n",
    "  X1  = Conv2DTranspose(64 ,(38,46), use_bias = False)(X1)\n",
    "\n",
    "  X2 = context_flow(X1,input,'2')\n",
    "\n",
    "  X2  = Conv2DTranspose(64 ,(38,46), use_bias = False)(X2)\n",
    "\n",
    "  X3 = context_flow(X2,input,'3')\n",
    "\n",
    "  X3  = Conv2DTranspose(64 ,(38,46), use_bias = False)(X3)\n",
    "\n",
    "  O  = Add(name = 'add_FE')([X_in, X1, X2, X3])\n",
    "\n",
    "  X = Conv2D(64, (3, 3), activation = 'relu', padding = 'same', name = 'CAM_conv3' )(O)\n",
    "\n",
    "  short_link=X\n",
    "\n",
    "  X = AveragePooling2D(1,(2, 2),padding = 'same', name = 'FS_pool')(X)\n",
    "  X = Conv2D(64, (1, 1), activation = 'relu', padding = 'same', name = 'CAM_conv4')(X)\n",
    "  X = BatchNormalization()(X)\n",
    "  X = Activation('sigmoid')(X)\n",
    "\n",
    "  X  = Conv2DTranspose(64 ,(38,46), use_bias = False)(X)\n",
    "  X = Multiply()([X,short_link])\n",
    "\n",
    "  X = Conv2D(64, (3, 3), activation = 'relu', padding = 'same', name = 'CAM_conv5')(X)\n",
    "\n",
    "  AGCN_out=AGCN(input)\n",
    "\n",
    "  Conc_A =  (UpSampling2D((2,2), interpolation='nearest'))(X)\n",
    "  Conc_B =  (UpSampling2D((2,2), interpolation='nearest'))(AGCN_out)\n",
    "  X = concatenate([Conc_A,Conc_B])\n",
    "  \n",
    "  X = Conv2D(40, (3, 3), activation = 'relu', padding = 'same', name = 'FinalBlock1')(X)\n",
    "      \n",
    "  X  = UpSampling2D((4,4), interpolation = 'nearest')(X)\n",
    "  output = (Activation('softmax'))(X)\n",
    "\n",
    "  return output\n",
    "  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 1000
    },
    "colab_type": "code",
    "id": "U3j5rflf3Gc-",
    "outputId": "93594a24-c96f-4366-eace-7ea0bb7974cc"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"DRN26_CAM_Model\"\n",
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_1 (InputLayer)            [(None, 600, 720, 3) 0                                            \n",
      "__________________________________________________________________________________________________\n",
      "layer1_conv (Conv2D)            (None, 600, 720, 16) 2368        input_1[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "layer2_conv (Conv2D)            (None, 600, 720, 16) 2320        layer1_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "conv2d (Conv2D)                 (None, 600, 720, 16) 272         layer1_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "layer3_conv (Conv2D)            (None, 600, 720, 16) 2320        layer2_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "add (Add)                       (None, 600, 720, 16) 0           conv2d[0][0]                     \n",
      "                                                                 layer3_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation (Activation)         (None, 600, 720, 16) 0           add[0][0]                        \n",
      "__________________________________________________________________________________________________\n",
      "layer4_conv (Conv2D)            (None, 600, 720, 20) 2900        activation[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_1 (Conv2D)               (None, 600, 720, 20) 340         activation[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "layer5_conv (Conv2D)            (None, 600, 720, 20) 3620        layer4_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "add_1 (Add)                     (None, 600, 720, 20) 0           conv2d_1[0][0]                   \n",
      "                                                                 layer5_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_1 (Activation)       (None, 600, 720, 20) 0           add_1[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer4_Max (MaxPooling2D)       (None, 300, 360, 20) 0           activation_1[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer6_conv (Conv2D)            (None, 300, 360, 32) 5792        layer4_Max[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_2 (Conv2D)               (None, 300, 360, 32) 672         layer4_Max[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "layer7_conv (Conv2D)            (None, 300, 360, 32) 9248        layer6_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "add_2 (Add)                     (None, 300, 360, 32) 0           conv2d_2[0][0]                   \n",
      "                                                                 layer7_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_2 (Activation)       (None, 300, 360, 32) 0           add_2[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer8_conv (Conv2D)            (None, 300, 360, 64) 18496       activation_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_3 (Conv2D)               (None, 300, 360, 64) 2112        activation_2[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer9_conv (Conv2D)            (None, 300, 360, 64) 36928       layer8_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "add_3 (Add)                     (None, 300, 360, 64) 0           conv2d_3[0][0]                   \n",
      "                                                                 layer9_conv[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_3 (Activation)       (None, 300, 360, 64) 0           add_3[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D)    (None, 150, 180, 64) 0           activation_3[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer12_disconv (Conv2D)        (None, 150, 180, 128 73856       max_pooling2d[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_4 (Conv2D)               (None, 150, 180, 128 8320        max_pooling2d[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "layer13_disconv (Conv2D)        (None, 150, 180, 128 147584      layer12_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_4 (Add)                     (None, 150, 180, 128 0           conv2d_4[0][0]                   \n",
      "                                                                 layer13_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "activation_4 (Activation)       (None, 150, 180, 128 0           add_4[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer14_disconv (Conv2D)        (None, 150, 180, 128 147584      activation_4[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_5 (Conv2D)               (None, 150, 180, 128 16512       activation_4[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer15_disconv (Conv2D)        (None, 150, 180, 128 147584      layer14_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_5 (Add)                     (None, 150, 180, 128 0           conv2d_5[0][0]                   \n",
      "                                                                 layer15_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "activation_5 (Activation)       (None, 150, 180, 128 0           add_5[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer16_disconv (Conv2D)        (None, 150, 180, 256 295168      activation_5[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_6 (Conv2D)               (None, 150, 180, 256 33024       activation_5[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer17_disconv (Conv2D)        (None, 150, 180, 256 590080      layer16_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "add_6 (Add)                     (None, 150, 180, 256 0           conv2d_6[0][0]                   \n",
      "                                                                 layer17_disconv[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "activation_6 (Activation)       (None, 150, 180, 256 0           add_6[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer18_conv (Conv2D)           (None, 150, 180, 256 590080      activation_6[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_7 (Conv2D)               (None, 150, 180, 256 65792       activation_6[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer19_conv (Conv2D)           (None, 150, 180, 256 590080      layer18_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "add_7 (Add)                     (None, 150, 180, 256 0           conv2d_7[0][0]                   \n",
      "                                                                 layer19_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_7 (Activation)       (None, 150, 180, 256 0           add_7[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2D)  (None, 75, 90, 256)  0           activation_7[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer20_conv (Conv2D)           (None, 75, 90, 128)  295040      max_pooling2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_8 (Conv2D)               (None, 75, 90, 128)  32896       max_pooling2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "layer21_conv (Conv2D)           (None, 75, 90, 128)  147584      layer20_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "add_8 (Add)                     (None, 75, 90, 128)  0           conv2d_8[0][0]                   \n",
      "                                                                 layer21_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_8 (Activation)       (None, 75, 90, 128)  0           add_8[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "layer22_conv (Conv2D)           (None, 75, 90, 64)   49216       activation_8[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_9 (Conv2D)               (None, 75, 90, 64)   8256        activation_8[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "layer23_conv (Conv2D)           (None, 75, 90, 64)   24640       layer22_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "add_9 (Add)                     (None, 75, 90, 64)   0           conv2d_9[0][0]                   \n",
      "                                                                 layer23_conv[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_9 (Activation)       (None, 75, 90, 64)   0           add_9[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "average_pooling2d (AveragePooli (None, 5, 6, 64)     0           activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization (BatchNorma (None, 5, 6, 64)     256         average_pooling2d[0][0]          \n",
      "__________________________________________________________________________________________________\n",
      "activation_10 (Activation)      (None, 5, 6, 64)     0           batch_normalization[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "CAM_conv1 (Conv2D)              (None, 5, 6, 64)     4160        activation_10[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_transpose (Conv2DTranspo (None, 75, 90, 64)   24719360    CAM_conv1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate (Concatenate)       (None, 75, 90, 128)  0           conv2d_transpose[0][0]           \n",
      "                                                                 activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "average_pooling2d_1 (AveragePoo (None, 38, 45, 128)  0           concatenate[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "lambda (Lambda)                 (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_1 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_2 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_3 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_4 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_5 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_6 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_7 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_8 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_9 (Lambda)               (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_10 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_11 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_12 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_13 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_14 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_15 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_1[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_10 (Conv2D)              (None, 38, 45, 4)    36          lambda[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_11 (Conv2D)              (None, 38, 45, 4)    36          lambda_1[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_12 (Conv2D)              (None, 38, 45, 4)    36          lambda_2[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_13 (Conv2D)              (None, 38, 45, 4)    36          lambda_3[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_14 (Conv2D)              (None, 38, 45, 4)    36          lambda_4[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_15 (Conv2D)              (None, 38, 45, 4)    36          lambda_5[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_16 (Conv2D)              (None, 38, 45, 4)    36          lambda_6[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_17 (Conv2D)              (None, 38, 45, 4)    36          lambda_7[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_18 (Conv2D)              (None, 38, 45, 4)    36          lambda_8[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_19 (Conv2D)              (None, 38, 45, 4)    36          lambda_9[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_20 (Conv2D)              (None, 38, 45, 4)    36          lambda_10[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_21 (Conv2D)              (None, 38, 45, 4)    36          lambda_11[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_22 (Conv2D)              (None, 38, 45, 4)    36          lambda_12[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_23 (Conv2D)              (None, 38, 45, 4)    36          lambda_13[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_24 (Conv2D)              (None, 38, 45, 4)    36          lambda_14[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_25 (Conv2D)              (None, 38, 45, 4)    36          lambda_15[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_1 (Concatenate)     (None, 38, 45, 64)   0           conv2d_10[0][0]                  \n",
      "                                                                 conv2d_11[0][0]                  \n",
      "                                                                 conv2d_12[0][0]                  \n",
      "                                                                 conv2d_13[0][0]                  \n",
      "                                                                 conv2d_14[0][0]                  \n",
      "                                                                 conv2d_15[0][0]                  \n",
      "                                                                 conv2d_16[0][0]                  \n",
      "                                                                 conv2d_17[0][0]                  \n",
      "                                                                 conv2d_18[0][0]                  \n",
      "                                                                 conv2d_19[0][0]                  \n",
      "                                                                 conv2d_20[0][0]                  \n",
      "                                                                 conv2d_21[0][0]                  \n",
      "                                                                 conv2d_22[0][0]                  \n",
      "                                                                 conv2d_23[0][0]                  \n",
      "                                                                 conv2d_24[0][0]                  \n",
      "                                                                 conv2d_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "reshape (Reshape)               (None, 38, 45, 4, 16 0           concatenate_1[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "permute (Permute)               (None, 38, 45, 16, 4 0           reshape[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "reshape_1 (Reshape)             (None, 38, 45, 64)   0           permute[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "lambda_16 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_17 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_18 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_19 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_20 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_21 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_22 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_23 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_24 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_25 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_26 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_27 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_28 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_29 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_30 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_31 (Lambda)              (None, 38, 45, 4)    0           reshape_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_26 (Conv2D)              (None, 38, 45, 4)    20          lambda_16[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_27 (Conv2D)              (None, 38, 45, 4)    20          lambda_17[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_28 (Conv2D)              (None, 38, 45, 4)    20          lambda_18[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_29 (Conv2D)              (None, 38, 45, 4)    20          lambda_19[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_30 (Conv2D)              (None, 38, 45, 4)    20          lambda_20[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_31 (Conv2D)              (None, 38, 45, 4)    20          lambda_21[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_32 (Conv2D)              (None, 38, 45, 4)    20          lambda_22[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_33 (Conv2D)              (None, 38, 45, 4)    20          lambda_23[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_34 (Conv2D)              (None, 38, 45, 4)    20          lambda_24[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_35 (Conv2D)              (None, 38, 45, 4)    20          lambda_25[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_36 (Conv2D)              (None, 38, 45, 4)    20          lambda_26[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_37 (Conv2D)              (None, 38, 45, 4)    20          lambda_27[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_38 (Conv2D)              (None, 38, 45, 4)    20          lambda_28[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_39 (Conv2D)              (None, 38, 45, 4)    20          lambda_29[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_40 (Conv2D)              (None, 38, 45, 4)    20          lambda_30[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_41 (Conv2D)              (None, 38, 45, 4)    20          lambda_31[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_2 (Concatenate)     (None, 38, 45, 64)   0           conv2d_26[0][0]                  \n",
      "                                                                 conv2d_27[0][0]                  \n",
      "                                                                 conv2d_28[0][0]                  \n",
      "                                                                 conv2d_29[0][0]                  \n",
      "                                                                 conv2d_30[0][0]                  \n",
      "                                                                 conv2d_31[0][0]                  \n",
      "                                                                 conv2d_32[0][0]                  \n",
      "                                                                 conv2d_33[0][0]                  \n",
      "                                                                 conv2d_34[0][0]                  \n",
      "                                                                 conv2d_35[0][0]                  \n",
      "                                                                 conv2d_36[0][0]                  \n",
      "                                                                 conv2d_37[0][0]                  \n",
      "                                                                 conv2d_38[0][0]                  \n",
      "                                                                 conv2d_39[0][0]                  \n",
      "                                                                 conv2d_40[0][0]                  \n",
      "                                                                 conv2d_41[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv1_1 (Conv2D)   (None, 38, 45, 64)   4160        concatenate_2[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_11 (Activation)      (None, 38, 45, 64)   0           context_flow_conv1_1[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv2_1 (Conv2D)   (None, 38, 45, 64)   4160        activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_12 (Activation)      (None, 38, 45, 64)   0           context_flow_conv2_1[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "multiply (Multiply)             (None, 38, 45, 64)   0           concatenate_2[0][0]              \n",
      "                                                                 activation_12[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_10 (Add)                    (None, 38, 45, 64)   0           concatenate_2[0][0]              \n",
      "                                                                 multiply[0][0]                   \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_transpose_1 (Conv2DTrans (None, 75, 90, 64)   7159808     add_10[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_3 (Concatenate)     (None, 75, 90, 128)  0           conv2d_transpose_1[0][0]         \n",
      "                                                                 activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "average_pooling2d_2 (AveragePoo (None, 38, 45, 128)  0           concatenate_3[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "lambda_32 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_33 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_34 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_35 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_36 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_37 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_38 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_39 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_40 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_41 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_42 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_43 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_44 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_45 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_46 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_47 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_2[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_42 (Conv2D)              (None, 38, 45, 4)    36          lambda_32[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_43 (Conv2D)              (None, 38, 45, 4)    36          lambda_33[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_44 (Conv2D)              (None, 38, 45, 4)    36          lambda_34[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_45 (Conv2D)              (None, 38, 45, 4)    36          lambda_35[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_46 (Conv2D)              (None, 38, 45, 4)    36          lambda_36[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_47 (Conv2D)              (None, 38, 45, 4)    36          lambda_37[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_48 (Conv2D)              (None, 38, 45, 4)    36          lambda_38[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_49 (Conv2D)              (None, 38, 45, 4)    36          lambda_39[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_50 (Conv2D)              (None, 38, 45, 4)    36          lambda_40[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_51 (Conv2D)              (None, 38, 45, 4)    36          lambda_41[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_52 (Conv2D)              (None, 38, 45, 4)    36          lambda_42[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_53 (Conv2D)              (None, 38, 45, 4)    36          lambda_43[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_54 (Conv2D)              (None, 38, 45, 4)    36          lambda_44[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_55 (Conv2D)              (None, 38, 45, 4)    36          lambda_45[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_56 (Conv2D)              (None, 38, 45, 4)    36          lambda_46[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_57 (Conv2D)              (None, 38, 45, 4)    36          lambda_47[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_4 (Concatenate)     (None, 38, 45, 64)   0           conv2d_42[0][0]                  \n",
      "                                                                 conv2d_43[0][0]                  \n",
      "                                                                 conv2d_44[0][0]                  \n",
      "                                                                 conv2d_45[0][0]                  \n",
      "                                                                 conv2d_46[0][0]                  \n",
      "                                                                 conv2d_47[0][0]                  \n",
      "                                                                 conv2d_48[0][0]                  \n",
      "                                                                 conv2d_49[0][0]                  \n",
      "                                                                 conv2d_50[0][0]                  \n",
      "                                                                 conv2d_51[0][0]                  \n",
      "                                                                 conv2d_52[0][0]                  \n",
      "                                                                 conv2d_53[0][0]                  \n",
      "                                                                 conv2d_54[0][0]                  \n",
      "                                                                 conv2d_55[0][0]                  \n",
      "                                                                 conv2d_56[0][0]                  \n",
      "                                                                 conv2d_57[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "reshape_2 (Reshape)             (None, 38, 45, 4, 16 0           concatenate_4[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "permute_1 (Permute)             (None, 38, 45, 16, 4 0           reshape_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "reshape_3 (Reshape)             (None, 38, 45, 64)   0           permute_1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_48 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_49 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_50 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_51 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_52 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_53 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_54 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_55 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_56 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_57 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_58 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_59 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_60 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_61 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_62 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_63 (Lambda)              (None, 38, 45, 4)    0           reshape_3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_58 (Conv2D)              (None, 38, 45, 4)    20          lambda_48[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_59 (Conv2D)              (None, 38, 45, 4)    20          lambda_49[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_60 (Conv2D)              (None, 38, 45, 4)    20          lambda_50[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_61 (Conv2D)              (None, 38, 45, 4)    20          lambda_51[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_62 (Conv2D)              (None, 38, 45, 4)    20          lambda_52[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_63 (Conv2D)              (None, 38, 45, 4)    20          lambda_53[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_64 (Conv2D)              (None, 38, 45, 4)    20          lambda_54[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_65 (Conv2D)              (None, 38, 45, 4)    20          lambda_55[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_66 (Conv2D)              (None, 38, 45, 4)    20          lambda_56[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_67 (Conv2D)              (None, 38, 45, 4)    20          lambda_57[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_68 (Conv2D)              (None, 38, 45, 4)    20          lambda_58[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_69 (Conv2D)              (None, 38, 45, 4)    20          lambda_59[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_70 (Conv2D)              (None, 38, 45, 4)    20          lambda_60[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_71 (Conv2D)              (None, 38, 45, 4)    20          lambda_61[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_72 (Conv2D)              (None, 38, 45, 4)    20          lambda_62[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_73 (Conv2D)              (None, 38, 45, 4)    20          lambda_63[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_5 (Concatenate)     (None, 38, 45, 64)   0           conv2d_58[0][0]                  \n",
      "                                                                 conv2d_59[0][0]                  \n",
      "                                                                 conv2d_60[0][0]                  \n",
      "                                                                 conv2d_61[0][0]                  \n",
      "                                                                 conv2d_62[0][0]                  \n",
      "                                                                 conv2d_63[0][0]                  \n",
      "                                                                 conv2d_64[0][0]                  \n",
      "                                                                 conv2d_65[0][0]                  \n",
      "                                                                 conv2d_66[0][0]                  \n",
      "                                                                 conv2d_67[0][0]                  \n",
      "                                                                 conv2d_68[0][0]                  \n",
      "                                                                 conv2d_69[0][0]                  \n",
      "                                                                 conv2d_70[0][0]                  \n",
      "                                                                 conv2d_71[0][0]                  \n",
      "                                                                 conv2d_72[0][0]                  \n",
      "                                                                 conv2d_73[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv1_2 (Conv2D)   (None, 38, 45, 64)   4160        concatenate_5[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_13 (Activation)      (None, 38, 45, 64)   0           context_flow_conv1_2[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv2_2 (Conv2D)   (None, 38, 45, 64)   4160        activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_14 (Activation)      (None, 38, 45, 64)   0           context_flow_conv2_2[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "multiply_1 (Multiply)           (None, 38, 45, 64)   0           concatenate_5[0][0]              \n",
      "                                                                 activation_14[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_11 (Add)                    (None, 38, 45, 64)   0           concatenate_5[0][0]              \n",
      "                                                                 multiply_1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_transpose_2 (Conv2DTrans (None, 75, 90, 64)   7159808     add_11[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_6 (Concatenate)     (None, 75, 90, 128)  0           conv2d_transpose_2[0][0]         \n",
      "                                                                 activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "average_pooling2d_3 (AveragePoo (None, 38, 45, 128)  0           concatenate_6[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "lambda_64 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_65 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_66 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_67 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_68 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_69 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_70 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_71 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_72 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_73 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_74 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_75 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_76 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_77 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_78 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "lambda_79 (Lambda)              (None, 38, 45, 8)    0           average_pooling2d_3[0][0]        \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_74 (Conv2D)              (None, 38, 45, 4)    36          lambda_64[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_75 (Conv2D)              (None, 38, 45, 4)    36          lambda_65[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_76 (Conv2D)              (None, 38, 45, 4)    36          lambda_66[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_77 (Conv2D)              (None, 38, 45, 4)    36          lambda_67[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_78 (Conv2D)              (None, 38, 45, 4)    36          lambda_68[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_79 (Conv2D)              (None, 38, 45, 4)    36          lambda_69[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_80 (Conv2D)              (None, 38, 45, 4)    36          lambda_70[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_81 (Conv2D)              (None, 38, 45, 4)    36          lambda_71[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_82 (Conv2D)              (None, 38, 45, 4)    36          lambda_72[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_83 (Conv2D)              (None, 38, 45, 4)    36          lambda_73[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_84 (Conv2D)              (None, 38, 45, 4)    36          lambda_74[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_85 (Conv2D)              (None, 38, 45, 4)    36          lambda_75[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_86 (Conv2D)              (None, 38, 45, 4)    36          lambda_76[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_87 (Conv2D)              (None, 38, 45, 4)    36          lambda_77[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_88 (Conv2D)              (None, 38, 45, 4)    36          lambda_78[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_89 (Conv2D)              (None, 38, 45, 4)    36          lambda_79[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_7 (Concatenate)     (None, 38, 45, 64)   0           conv2d_74[0][0]                  \n",
      "                                                                 conv2d_75[0][0]                  \n",
      "                                                                 conv2d_76[0][0]                  \n",
      "                                                                 conv2d_77[0][0]                  \n",
      "                                                                 conv2d_78[0][0]                  \n",
      "                                                                 conv2d_79[0][0]                  \n",
      "                                                                 conv2d_80[0][0]                  \n",
      "                                                                 conv2d_81[0][0]                  \n",
      "                                                                 conv2d_82[0][0]                  \n",
      "                                                                 conv2d_83[0][0]                  \n",
      "                                                                 conv2d_84[0][0]                  \n",
      "                                                                 conv2d_85[0][0]                  \n",
      "                                                                 conv2d_86[0][0]                  \n",
      "                                                                 conv2d_87[0][0]                  \n",
      "                                                                 conv2d_88[0][0]                  \n",
      "                                                                 conv2d_89[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "reshape_4 (Reshape)             (None, 38, 45, 4, 16 0           concatenate_7[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "permute_2 (Permute)             (None, 38, 45, 16, 4 0           reshape_4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "reshape_5 (Reshape)             (None, 38, 45, 64)   0           permute_2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_80 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_81 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_82 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_83 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_84 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_85 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_86 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_87 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_88 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_89 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_90 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_91 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_92 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_93 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_94 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "lambda_95 (Lambda)              (None, 38, 45, 4)    0           reshape_5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_90 (Conv2D)              (None, 38, 45, 4)    20          lambda_80[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_91 (Conv2D)              (None, 38, 45, 4)    20          lambda_81[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_92 (Conv2D)              (None, 38, 45, 4)    20          lambda_82[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_93 (Conv2D)              (None, 38, 45, 4)    20          lambda_83[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_94 (Conv2D)              (None, 38, 45, 4)    20          lambda_84[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_95 (Conv2D)              (None, 38, 45, 4)    20          lambda_85[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_96 (Conv2D)              (None, 38, 45, 4)    20          lambda_86[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_97 (Conv2D)              (None, 38, 45, 4)    20          lambda_87[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_98 (Conv2D)              (None, 38, 45, 4)    20          lambda_88[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_99 (Conv2D)              (None, 38, 45, 4)    20          lambda_89[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_100 (Conv2D)             (None, 38, 45, 4)    20          lambda_90[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_101 (Conv2D)             (None, 38, 45, 4)    20          lambda_91[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_102 (Conv2D)             (None, 38, 45, 4)    20          lambda_92[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_103 (Conv2D)             (None, 38, 45, 4)    20          lambda_93[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_104 (Conv2D)             (None, 38, 45, 4)    20          lambda_94[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_105 (Conv2D)             (None, 38, 45, 4)    20          lambda_95[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_8 (Concatenate)     (None, 38, 45, 64)   0           conv2d_90[0][0]                  \n",
      "                                                                 conv2d_91[0][0]                  \n",
      "                                                                 conv2d_92[0][0]                  \n",
      "                                                                 conv2d_93[0][0]                  \n",
      "                                                                 conv2d_94[0][0]                  \n",
      "                                                                 conv2d_95[0][0]                  \n",
      "                                                                 conv2d_96[0][0]                  \n",
      "                                                                 conv2d_97[0][0]                  \n",
      "                                                                 conv2d_98[0][0]                  \n",
      "                                                                 conv2d_99[0][0]                  \n",
      "                                                                 conv2d_100[0][0]                 \n",
      "                                                                 conv2d_101[0][0]                 \n",
      "                                                                 conv2d_102[0][0]                 \n",
      "                                                                 conv2d_103[0][0]                 \n",
      "                                                                 conv2d_104[0][0]                 \n",
      "                                                                 conv2d_105[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv1_3 (Conv2D)   (None, 38, 45, 64)   4160        concatenate_8[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_15 (Activation)      (None, 38, 45, 64)   0           context_flow_conv1_3[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "context_flow_conv2_3 (Conv2D)   (None, 38, 45, 64)   4160        activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "activation_16 (Activation)      (None, 38, 45, 64)   0           context_flow_conv2_3[0][0]       \n",
      "__________________________________________________________________________________________________\n",
      "multiply_2 (Multiply)           (None, 38, 45, 64)   0           concatenate_8[0][0]              \n",
      "                                                                 activation_16[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_12 (Add)                    (None, 38, 45, 64)   0           concatenate_8[0][0]              \n",
      "                                                                 multiply_2[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_transpose_3 (Conv2DTrans (None, 75, 90, 64)   7159808     add_12[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "add_FE (Add)                    (None, 75, 90, 64)   0           conv2d_transpose[0][0]           \n",
      "                                                                 conv2d_transpose_1[0][0]         \n",
      "                                                                 conv2d_transpose_2[0][0]         \n",
      "                                                                 conv2d_transpose_3[0][0]         \n",
      "__________________________________________________________________________________________________\n",
      "CAM_conv3 (Conv2D)              (None, 75, 90, 64)   36928       add_FE[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "FS_pool (AveragePooling2D)      (None, 38, 45, 64)   0           CAM_conv3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "CAM_conv4 (Conv2D)              (None, 38, 45, 64)   4160        FS_pool[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "batch_normalization_1 (BatchNor (None, 38, 45, 64)   256         CAM_conv4[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "AGCN_conv1 (Conv2D)             (None, 75, 90, 64)   28736       activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "AGCN_conv3 (Conv2D)             (None, 75, 90, 64)   28736       activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "activation_17 (Activation)      (None, 38, 45, 64)   0           batch_normalization_1[0][0]      \n",
      "__________________________________________________________________________________________________\n",
      "AGCN_conv2 (Conv2D)             (None, 75, 90, 64)   28736       AGCN_conv1[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "AGCN_conv4 (Conv2D)             (None, 75, 90, 64)   28736       AGCN_conv3[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "conv2d_transpose_4 (Conv2DTrans (None, 75, 90, 64)   7159808     activation_17[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_AGCN1 (Add)                 (None, 75, 90, 64)   0           AGCN_conv2[0][0]                 \n",
      "                                                                 AGCN_conv4[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "multiply_3 (Multiply)           (None, 75, 90, 64)   0           conv2d_transpose_4[0][0]         \n",
      "                                                                 CAM_conv3[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "AGCN_conv5 (Conv2D)             (None, 75, 90, 64)   36928       add_AGCN1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "CAM_conv5 (Conv2D)              (None, 75, 90, 64)   36928       multiply_3[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "add_AGCN2 (Add)                 (None, 75, 90, 64)   0           AGCN_conv5[0][0]                 \n",
      "                                                                 add_AGCN1[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d (UpSampling2D)    (None, 150, 180, 64) 0           CAM_conv5[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_1 (UpSampling2D)  (None, 150, 180, 64) 0           add_AGCN2[0][0]                  \n",
      "__________________________________________________________________________________________________\n",
      "concatenate_9 (Concatenate)     (None, 150, 180, 128 0           up_sampling2d[0][0]              \n",
      "                                                                 up_sampling2d_1[0][0]            \n",
      "__________________________________________________________________________________________________\n",
      "FinalBlock1 (Conv2D)            (None, 150, 180, 40) 46120       concatenate_9[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "up_sampling2d_2 (UpSampling2D)  (None, 600, 720, 40) 0           FinalBlock1[0][0]                \n",
      "__________________________________________________________________________________________________\n",
      "activation_18 (Activation)      (None, 600, 720, 40) 0           up_sampling2d_2[0][0]            \n",
      "==================================================================================================\n",
      "Total params: 57,017,604\n",
      "Trainable params: 57,017,348\n",
      "Non-trainable params: 256\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "from keras.models import Model, load_model\n",
    "from keras import backend as K\n",
    "K.clear_session()\n",
    "\n",
    "\n",
    "X_input = Input(shape=(600,720,3))\n",
    "\n",
    "X= model_DRN26(X_input)\n",
    "\n",
    "output=model_CAM(X)\n",
    "\n",
    "model = Model(inputs = X_input, outputs = output, name='DRN26_CAM_Model')\n",
    "\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "soGXXuyW9ytC"
   },
   "outputs": [],
   "source": [
    "\n",
    "def compute_mean_iou(y_true, y_pred, smooth=1):\n",
    "\n",
    "    intersection = K.sum(K.abs(y_true * y_pred), axis=[1,2,3])\n",
    "    union = K.sum(y_true,[1,2,3])+K.sum(y_pred,[1,2,3])-intersection\n",
    "    iou = K.mean((intersection + smooth) / (union + smooth), axis=0)\n",
    "    return iou\n",
    "\n",
    "def compute_mean_diec(y_true, y_pred, smooth=1):\n",
    "\n",
    "    intersection = K.sum(y_true * y_pred, axis=[1,2,3])\n",
    "    union = K.sum(y_true, axis=[1,2,3]) + K.sum(y_pred, axis=[1,2,3])\n",
    "    dice = K.mean((2. * intersection + smooth)/(union + smooth), axis=0)\n",
    "    return dice\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BzpxSqHC9ulX"
   },
   "outputs": [],
   "source": [
    "import datetime\n",
    "from keras import optimizers\n",
    "from keras.callbacks import ModelCheckpoint, EarlyStopping, ReduceLROnPlateau, TensorBoard\n",
    "\n",
    "\n",
    "Adam = optimizers.Adam()\n",
    "model.compile(loss = 'categorical_crossentropy', optimizer = Adam, metrics = [compute_mean_iou,compute_mean_diec,'accuracy'])\n",
    "log_dir=\"logs\\\\fit\\\\\" + datetime.datetime.now().strftime(\"%Y%m%d-%H%M%S\")\n",
    "tensorboard_callback = TensorBoard(log_dir=log_dir, histogram_freq=1)\n",
    "\n",
    "mcp_save = ModelCheckpoint('weights-improvement.rescan_wts.hdf5', save_best_only=True, monitor='val_loss', mode='min')\n",
    "reduce_lr_loss = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience=7, verbose=1, min_delta=1e-4, mode='min')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "BxLHeuWq_AZ6"
   },
   "outputs": [],
   "source": [
    "class DataGenerator(keras.utils.Sequence):\n",
    "    'Generates data for Keras'\n",
    "    def __init__(self, input_image,input_height,input_width,label_values, batch_size,shuffle=True):\n",
    "        'Initialization'\n",
    "        \n",
    "        self.batch_size = batch_size\n",
    "\n",
    "        self.input_image = input_image\n",
    "        self.input_height=input_height\n",
    "        self.input_width=input_width\n",
    "        self.label_values=label_values\n",
    "        self.shuffle = shuffle\n",
    "        self.on_epoch_end()\n",
    "\n",
    "    def __len__(self):\n",
    "        'Denotes the number of batches per epoch'\n",
    "        return int(np.floor(len(self.input_image) / self.batch_size))\n",
    "\n",
    "    def __getitem__(self, index):\n",
    "        'Generate one batch of data'\n",
    "        \n",
    "        # Generate indexes of the batch\n",
    "        indexes = self.indexes[index*self.batch_size:(index+1)*self.batch_size]\n",
    "\n",
    "        # Find list of IDs\n",
    "        self.list_IDs_temp_y = [self.input_image[k] for k in indexes]\n",
    "        self.list_IDs_temp_x = [p.replace('/content/drive/My Drive/datasets_all/output', '/content') for p in self.list_IDs_temp_y]\n",
    "\n",
    "        # Generate data\n",
    "        X, Y = self.__data_generation(self.list_IDs_temp_x,self.list_IDs_temp_y)\n",
    "        X=np.squeeze(X, axis = 1) \n",
    "        return X, Y\n",
    "\n",
    "    def on_epoch_end(self):\n",
    "        'Updates indexes after each epoch'\n",
    "        self.indexes = np.arange(len(self.input_image))\n",
    "        if self.shuffle == True:\n",
    "            np.random.shuffle(self.indexes)\n",
    "\n",
    "    def __data_generation(self, in_X,in_Y):\n",
    "        'Generates data containing batch_size samples' # X : (n_samples, *dim, n_channels)\n",
    "        return_X=[]\n",
    "        return_Y=[]\n",
    "        # Generate data\n",
    "        for X_Image,Y_Image in  zip(in_X,in_Y):\n",
    "            # Store sample\n",
    "            return_X.append(image_array(X_Image,self.input_height,self.input_width))\n",
    "            return_Y.append(np.array(one_hot_it(Y_Image,self.input_height, self.input_width, self.label_values), dtype=np.uint8))\n",
    "\n",
    "        return np.array(return_X), np.array(return_Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jHuSBE_lMfQG"
   },
   "outputs": [],
   "source": [
    "with open('/content/drive/My Drive/label_txt.txt') as f:\n",
    "    label_to_color_map = json.load(f)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "bMNCFnXfCrNX"
   },
   "outputs": [],
   "source": [
    "def load_image(path):\n",
    "    image = cv2.imread(path,-1)\n",
    "    return image\n",
    "    \n",
    "def image_array(path,height,width):\n",
    "  img=np.expand_dims(np.float32(load_image(path)[:height, :width]),axis=0)/255.0\n",
    "  return img\n",
    "\n",
    "def one_hot_it(path,height,width, label_values):\n",
    "\n",
    "    label=load_image(path)[:height, :width]\n",
    "    semantic_map = []\n",
    "\n",
    "    for key in list(label_to_color_map.keys()):\n",
    "        colour = label_to_color_map[key]\n",
    "        equality = np.equal(label, colour)\n",
    "        class_map = np.all(equality, axis = -1)\n",
    "        semantic_map.append(class_map)\n",
    "        \n",
    "    semantic_map = np.stack(semantic_map, axis=-1)\n",
    "    # print(\"Time 2 = \", time.time() - st)\n",
    "\n",
    "    return semantic_map"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "eIXKuOAaMuW0"
   },
   "outputs": [],
   "source": [
    "from glob import glob\n",
    "\n",
    "train_image_y = glob('/content/drive/My Drive/datasets_all/output/train/*_leftImg8bit.png')\n",
    "val_image_y = glob('/content/drive/My Drive/datasets_all/output/val/*_leftImg8bit.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jeruJd4z291P"
   },
   "outputs": [],
   "source": [
    "# Generators\n",
    "training_generator = DataGenerator(train_image_y,600,720,' ', batch_size=6,shuffle=True)\n",
    "validation_generator = DataGenerator(val_image_y,600,720,' ', batch_size=6,shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "0mbrPwM4N8NN",
    "outputId": "dfe7faa7-010f-46fd-e56a-3872581fc441"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0.5527749061584473\n"
     ]
    }
   ],
   "source": [
    "import time\n",
    "\n",
    "start_time = time.time()\n",
    "one_hot_it('/content/train/000128_leftImg8bit.png',600,720,' ')\n",
    "end_time = time.time()\n",
    "print (end_time-start_time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "U-Ctq3ImvRdP"
   },
   "outputs": [],
   "source": [
    "a = [12]\n",
    "while(1):\n",
    "    a.append(a)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 171
    },
    "colab_type": "code",
    "id": "MvYszxPOO05M",
    "outputId": "332b21a2-7234-49ba-cde3-33e2270de460"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/2\n",
      "  1/807 [..............................] - ETA: 0s - loss: 1.0111 - compute_mean_iou: 0.5405 - compute_mean_diec: 0.6620 - accuracy: 0.7292WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/summary_ops_v2.py:1277: stop (from tensorflow.python.eager.profiler) is deprecated and will be removed after 2020-07-01.\n",
      "Instructions for updating:\n",
      "use `tf.profiler.experimental.stop` instead.\n",
      "807/807 [==============================] - 3038s 4s/step - loss: 0.8373 - compute_mean_iou: 0.5696 - compute_mean_diec: 0.7118 - accuracy: 0.7878 - val_loss: 0.9724 - val_compute_mean_iou: 0.5503 - val_compute_mean_diec: 0.6915 - val_accuracy: 0.7683\n",
      "Epoch 2/2\n",
      "807/807 [==============================] - 3031s 4s/step - loss: 0.8227 - compute_mean_iou: 0.5746 - compute_mean_diec: 0.7161 - accuracy: 0.7908 - val_loss: 0.9570 - val_compute_mean_iou: 0.5483 - val_compute_mean_diec: 0.6891 - val_accuracy: 0.7667\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x7ff32c4d52b0>"
      ]
     },
     "execution_count": 17,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "import gc\n",
    "gc.collect()\n",
    "\n",
    "model.load_weights('/content/drive/My Drive/weights-improvement.rescan_wts.hdf5')\n",
    "# Train model on dataset\n",
    "model.fit(training_generator, epochs=2, verbose=1,   \n",
    "                    validation_data=validation_generator, \n",
    "                callbacks = [tensorboard_callback,mcp_save,reduce_lr_loss])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "OPVk2KRQGc-4"
   },
   "outputs": [],
   "source": [
    "\n",
    "XX=image_array('/content/train/001133_leftImg8bit.png',600,720)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "oJmtXskkF-rn"
   },
   "outputs": [],
   "source": [
    "hist = model.predict(XX)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CwJp1SKWHAbO"
   },
   "outputs": [],
   "source": [
    "def reverse_one_hot(image):\n",
    "\n",
    "    x = np.argmax(image, axis = -1)\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "CnLXHKaHHBdf"
   },
   "outputs": [],
   "source": [
    "hist_tt=reverse_one_hot(hist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "CwjGnCtHHJjk",
    "outputId": "9b955fb3-fea1-401d-c10c-5fdae7368a32"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(600, 720)"
      ]
     },
     "execution_count": 22,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "hist_mm=np.squeeze(hist_tt, axis = 0) \n",
    "hist_mm.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "MqLDqAR2KRJZ"
   },
   "outputs": [],
   "source": [
    "label_values=[]\n",
    "for key in list(label_to_color_map.keys()):\n",
    "  label_values.append(label_to_color_map[key])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "DnIf83lLHxtF"
   },
   "outputs": [],
   "source": [
    "def colour_code_segmentation(image, label_values):\n",
    "    \n",
    "    colour_codes = np.array(label_values)\n",
    "    x = colour_codes[image.astype(int)]\n",
    "\n",
    "    return x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "WvOSolDhH7Y3"
   },
   "outputs": [],
   "source": [
    "img=colour_code_segmentation(hist_mm, label_values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "id": "vd1impF_Ir3b",
    "outputId": "6200923b-a42c-4c08-9dec-e4fca646e4e8"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.imshow(img)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "id": "ZqZkg6fOI52u",
    "outputId": "14265537-7052-4454-ed29-d8e673b5d221"
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light",
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "YY=image_array('/content/drive/My Drive/datasets_all/output/train/001133_leftImg8bit.png',600,720)\n",
    "plt.imshow(np.squeeze(YY, axis = 0) )\n",
    "plt.show()"
   ]
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Untitled1.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.9"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}