project-short.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os \n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "from matplotlib import pyplot as plt\n",
    "import cv2\n",
    "import tensorflow as tf\n",
    "from keras.models import Sequential, Model\n",
    "from keras.engine.topology import Input\n",
    "from keras.layers.core import Dense, Dropout, Activation,Reshape\n",
    "from keras.layers import Conv1D, Conv2D, Flatten, MaxPooling2D, UpSampling2D, ZeroPadding2D\n",
    "from keras.layers.normalization import BatchNormalization\n",
    "from keras.utils import np_utils\n",
    "from keras.utils import plot_model\n",
    "from keras import backend as K\n",
    "import keras\n",
    "from PIL import Image\n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.layers.core import Lambda\n",
    "from IPython.display import SVG\n",
    "from keras.utils.vis_utils import model_to_dot\n",
    "from keras.callbacks import ModelCheckpoint, EarlyStopping,TensorBoard,CSVLogger,ReduceLROnPlateau"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "batch = 3\n",
    "epochs = 3;\n",
    "num_classes = 41;\n",
    "input_Shape = (150,150)\n",
    "num_train_samples = 800\n",
    "num_test_samples = 200\n",
    "model_name = 'short_skip'\n",
    "# creates folder\n",
    "if not os.path.exists('./'+model_name):\n",
    "    os.mkdir(model_name)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['gt', 'images', 'labels_categorical', 'label_colors.csv']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "os.listdir()\n",
    "os.chdir('./Dataset/Segmentation/')\n",
    "os.listdir()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Filenames"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dataset = []\n",
    "folder = \"./images/\"\n",
    "files = [f for f in os.listdir(folder) if os.path.isfile(os.path.join(folder, f))]\n",
    "for file in files:\n",
    "    dataset.append(file)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "dataset_labels = []\n",
    "folder = \"./labels_categorical/\"\n",
    "files = [f for f in os.listdir(folder) if os.path.isfile(os.path.join(folder, f))]\n",
    "for file in files:\n",
    "    dataset_labels.append(file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Split into Train and Test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x_train,x_test,y_train,y_test = train_test_split(dataset,dataset_labels,test_size = 0.2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "X_train = np.zeros((num_train_samples,input_Shape[0],input_Shape[1], 3 ))\n",
    "for index,item in enumerate(x_train): \n",
    "    path = \"./images/\" + item\n",
    "    img = plt.imread(path)\n",
    "    new_img = cv2.resize(img,(int(input_Shape[0]),int(input_Shape[1])))\n",
    "    array = np.array(new_img)\n",
    "    X_train[index] = array.reshape((input_Shape[0],input_Shape[1],3)) \n",
    "    \n",
    "    \n",
    "X_test = np.zeros((num_test_samples,input_Shape[0],input_Shape[1], 3 ))\n",
    "for index,item in enumerate(x_test):\n",
    "    path = \"./images/\" + item\n",
    "    img = plt.imread(path)\n",
    "    new_img = cv2.resize(img,(int(input_Shape[0]),int(input_Shape[1])))\n",
    "    array = np.array(new_img)\n",
    "    X_test[index] = array.reshape((input_Shape[0],input_Shape[1],3))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "Y_train = np.zeros((num_train_samples,input_Shape[0],input_Shape[1], num_classes ))\n",
    "for index,item in enumerate(y_train): \n",
    "    path = \"./labels_categorical/\" + item\n",
    "    Y_train[index] = np.load(path)\n",
    "    \n",
    "Y_test = np.zeros((num_test_samples,input_Shape[0],input_Shape[1], num_classes ))\n",
    "for index,item in enumerate(y_test): \n",
    "    path = \"./labels_categorical/\" + item\n",
    "    Y_test[index] = np.load(path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "gts = np.zeros((num_test_samples,input_Shape[0],input_Shape[1], 3 ))\n",
    "for index,item in enumerate(x_test): \n",
    "    path = \"./gt/\" + item\n",
    "    img = plt.imread(path,format = 'RGB')\n",
    "    print(img.shape)\n",
    "    new_img = cv2.resize(img,(int(input_Shape[0]),int(input_Shape[1])))\n",
    "    array = np.array(new_img)\n",
    "    print (array.shape)\n",
    "    break\n",
    "    gts[index] = array.reshape((input_Shape[0],input_Shape[1],3))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "classes = pd.read_csv('./label_colors.csv',sep='\\t',header=None)\n",
    "classes['class'] = classes.index.values.astype(int)\n",
    "classes.columns = ['rgb','name','class']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def image_map(im) :\n",
    "    \n",
    "    t = np.zeros((input_Shape[0],input_Shape[1],num_classes),dtype = int)\n",
    "    for n in range(num_classes) :\n",
    "        for i in range(im.shape[0]):\n",
    "            for j in range(im.shape[1]):\n",
    "            \n",
    "                color = classes['rgb'][n].split(' ')\n",
    "                color = [int(i) for i in color]\n",
    "                \n",
    "                if ( im[i,j,0] == color[0] and im[i,j,1] == color[1] and im[i,j,2] == color[2] ):\n",
    "                    t[i,j,n] = 1\n",
    "\n",
    "    return t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def map_img(segmap) :\n",
    "    t = np.zeros((input_Shape[0],input_Shape[1],3),dtype = int)\n",
    "    for i in range(input_Shape[0]) :\n",
    "        for j in range(input_Shape[1]):\n",
    "            a = segmap[i,j,:].argmax()\n",
    "            color = classes['rgb'][a].split(' ')\n",
    "            t[i,j,0] = int(color[0]) \n",
    "            t[i,j,1] = int(color[1])\n",
    "            t[i,j,2] = int(color[2])\n",
    "\n",
    "    return t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# for index,item in enumerate(dataset):\n",
    "#   im = labels[index]\n",
    "#   np.save(item,image_map(im))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def jaccard_distance(y_true, y_pred, smooth=100):\n",
    "    intersection = K.sum(K.abs(y_true * y_pred), axis=-1)\n",
    "    sum_ = K.sum(K.abs(y_true) + K.abs(y_pred), axis=-1)\n",
    "    jac = (intersection + smooth) / (sum_ - intersection + smooth)\n",
    "    return (1 - jac) * smooth"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# # Conv, BN, activation\n",
    "# # ENCODER\n",
    "\n",
    "# # Channels last \n",
    "# input_img = Input(shape=(input_Shape[0],input_Shape[1],3))\n",
    "\n",
    "# layer_1 = Conv2D(32,(3,3),padding = 'same', name = 'Conv_1',strides = 1)(input_img)\n",
    "# layer_1 = BatchNormalization(name = 'BN_1')(layer_1)\n",
    "# layer_1 = Activation('relu')(layer_1)\n",
    "# layer_1 = MaxPooling2D((2, 2), padding='same', name = 'pool_1', strides = 2)(layer_1)\n",
    "# # 75,75,32\n",
    "\n",
    "# layer_2 = Conv2D(32,(3,3),padding = 'same', name = 'Conv_2',strides = 1)(layer_1)\n",
    "# layer_2 = BatchNormalization(name = 'BN_2')(layer_2)\n",
    "# layer_2 = Activation('relu')(layer_2)\n",
    "# layer_2 = MaxPooling2D((2, 2), padding='same', name = 'pool_2', strides = 1)(layer_2)\n",
    "# # 75,75,32\n",
    "\n",
    "# added_1 = keras.layers.Add()([layer_1, layer_2]) \n",
    "# # 75,75,32\n",
    "# added1_conv =  Conv2D(64,(3,3),padding = 'same', name = 'Conv_3',strides = 1)(added_1)\n",
    "# # 75,75,64\n",
    "# added1_pool = MaxPooling2D((2, 2), padding='same', name = 'pool_3', strides = 2)(added1_conv)\n",
    "# # 38,38,64\n",
    "\n",
    "# layer_3 = Conv2D(64,(3,3),padding = 'same', name = 'Conv_4',strides = 1)(added_1)\n",
    "# layer_3 = BatchNormalization(name = 'BN_3')(layer_3)\n",
    "# layer_3 = Activation('relu')(layer_3)\n",
    "# layer_3 = MaxPooling2D((2, 2), padding='same', name = 'pool_4', strides = 2)(layer_3)\n",
    "# # 38,38,64\n",
    "\n",
    "# added_2 = keras.layers.Add()([added1_pool, layer_3]) \n",
    "# # 38,38,64\n",
    "# added2_conv =  Conv2D(128,(3,3),padding = 'same', name = 'Conv_5',strides = 1)(added_2)\n",
    "# # 38,38,64\n",
    "# added2_pool = MaxPooling2D((2, 2), padding='same', name = 'pool_5', strides = 2)(added2_conv)\n",
    "# # 19,19,128\n",
    "\n",
    "# layer_4 = Conv2D(128,(3,3),padding = 'same', name = 'Conv_6',strides = 1)(added_2)\n",
    "# layer_4 = BatchNormalization(name = 'BN_4')(layer_4)\n",
    "# layer_4 = Activation('relu')(layer_4)\n",
    "# layer_4 = MaxPooling2D((2, 2), padding='same', name = 'pool_6', strides = 2)(layer_4)\n",
    "# # 19,19,128\n",
    "\n",
    "# added_3 = keras.layers.Add()([added2_pool, layer_4]) \n",
    "# # 19,19,128\n",
    "\n",
    "# added3_upsample = UpSampling2D((2, 2),name = 'upsample_1')(added_3)\n",
    "# # 38,38,128\n",
    "\n",
    "# # DECODER\n",
    "# layer_5 = Conv2D(128,(3,3),padding = 'same', name = 'Conv_7',strides = 1)(added_3)\n",
    "# layer_5 = BatchNormalization(name = 'BN_5')(layer_5)\n",
    "# layer_5 = Activation('relu')(layer_5)\n",
    "# layer_5 = UpSampling2D((2, 2),name = 'upsample_2')(layer_5)\n",
    "# # 38,38,128\n",
    "\n",
    "# added_4 = keras.layers.Add()([added3_upsample, layer_5])\n",
    "# # 38,38,128\n",
    "# added4_upsample = UpSampling2D((2, 2),name = 'upsample_3')(added_4)\n",
    "# # 76,76,128\n",
    "# added4_conv = Conv2D(64,(3,3),padding = 'same', name = 'Conv_8',strides = 1)(added4_upsample)\n",
    "# # 76,76,64\n",
    "\n",
    "\n",
    "# layer_5 = Conv2D(64,(3,3),padding = 'same', name = 'Conv_9',strides = 1)(added_4)\n",
    "# layer_5 = BatchNormalization(name = 'BN_6')(layer_5)\n",
    "# layer_5 = Activation('relu')(layer_5)\n",
    "# layer_5 = UpSampling2D((2, 2),name = 'upsample_4')(layer_5)\n",
    "# # 76,76,64\n",
    "\n",
    "# added_5 = keras.layers.Add()([added4_conv, layer_5])\n",
    "\n",
    "# # 76,76,64\n",
    "# added5_conv = Conv2D(32,(3,3),padding = 'same', name = 'Conv_10',strides = 1)(added_5)\n",
    "# added5_upsample = UpSampling2D((2, 2),name = 'upsample_5')(added5_conv) \n",
    "# # 152,152,32\n",
    "\n",
    "\n",
    "# layer_6 = Conv2D(32,(3,3),padding = 'same', name = 'Conv_11',strides = 1)(added_5)\n",
    "# layer_6 = BatchNormalization(name = 'BN_7')(layer_6)\n",
    "# layer_6 = Activation('relu')(layer_6)\n",
    "# layer_6 = UpSampling2D((2, 2),name = 'upsample_6')(layer_6)\n",
    "# # 152,152,32\n",
    "\n",
    "# added_6 = keras.layers.Add()([added5_upsample, layer_6])\n",
    "# # 152,152,32\n",
    "\n",
    "# layer_6 = Conv2D(num_classes,(3,3),padding ='valid',name = 'Conv_12')(added_6)\n",
    "# layer_6  = BatchNormalization(name = 'BN_8')(layer_6)\n",
    "# decoded= Activation('softmax')(layer_6)\n",
    "\n",
    "# autoencoder = Model(input_img, decoded)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Conv, BN, activation\n",
    "# ENCODER\n",
    "\n",
    "# Channels last \n",
    "input_img = Input(shape=(input_Shape[0],input_Shape[1],3))\n",
    "\n",
    "layer_1 = Conv2D(32,(1,1),padding = 'same', name = 'Conv_1',strides = 1)(input_img)\n",
    "layer_1 = BatchNormalization(name = 'BN_1')(layer_1)\n",
    "layer_1 = Activation('relu')(layer_1)\n",
    "layer_1 = MaxPooling2D((2, 2), padding='same', name = 'pool_1', strides = 2)(layer_1)\n",
    "# 75,75,32\n",
    "\n",
    "layer_2 = Conv2D(32,(1,1),padding = 'same', name = 'Conv_2',strides = 1)(layer_1)\n",
    "layer_2 = BatchNormalization(name = 'BN_2')(layer_2)\n",
    "layer_2 = Activation('relu')(layer_2)\n",
    "layer_2 = MaxPooling2D((2, 2), padding='same', name = 'pool_2', strides = 1)(layer_2)\n",
    "# 75,75,32\n",
    "\n",
    "added_1 = keras.layers.Add()([layer_1, layer_2]) \n",
    "# 75,75,32\n",
    "added1_conv =  Conv2D(64,(1,1),padding = 'same', name = 'Conv_3',strides = 1)(added_1)\n",
    "# 75,75,64\n",
    "added1_pool = MaxPooling2D((2, 2), padding='same', name = 'pool_3', strides = 2)(added1_conv)\n",
    "# 38,38,64\n",
    "\n",
    "layer_3 = Conv2D(64,(1,1),padding = 'same', name = 'Conv_4',strides = 1)(added_1)\n",
    "layer_3 = BatchNormalization(name = 'BN_3')(layer_3)\n",
    "layer_3 = Activation('relu')(layer_3)\n",
    "layer_3 = MaxPooling2D((2, 2), padding='same', name = 'pool_4', strides = 2)(layer_3)\n",
    "# 38,38,64\n",
    "\n",
    "added_2 = keras.layers.Add()([added1_pool, layer_3]) \n",
    "# 38,38,64\n",
    "added2_conv =  Conv2D(128,(1,1),padding = 'same', name = 'Conv_5',strides = 1)(added_2)\n",
    "# 38,38,64\n",
    "added2_pool = MaxPooling2D((2, 2), padding='same', name = 'pool_5', strides = 2)(added2_conv)\n",
    "# 19,19,128\n",
    "\n",
    "layer_4 = Conv2D(128,(1,1),padding = 'same', name = 'Conv_6',strides = 1)(added_2)\n",
    "layer_4 = BatchNormalization(name = 'BN_4')(layer_4)\n",
    "layer_4 = Activation('relu')(layer_4)\n",
    "layer_4 = MaxPooling2D((2, 2), padding='same', name = 'pool_6', strides = 2)(layer_4)\n",
    "# 19,19,128\n",
    "\n",
    "added_3 = keras.layers.Add()([added2_pool, layer_4]) \n",
    "# 19,19,128\n",
    "\n",
    "added3_upsample = UpSampling2D((2, 2),name = 'upsample_1')(added_3)\n",
    "# 38,38,128\n",
    "\n",
    "# DECODER\n",
    "layer_5 = Conv2D(128,(1,1),padding = 'same', name = 'Conv_7',strides = 1)(added_3)\n",
    "layer_5 = BatchNormalization(name = 'BN_5')(layer_5)\n",
    "layer_5 = Activation('relu')(layer_5)\n",
    "layer_5 = UpSampling2D((2, 2),name = 'upsample_2')(layer_5)\n",
    "# 38,38,128\n",
    "\n",
    "added_4 = keras.layers.Add()([added3_upsample, layer_5])\n",
    "# 38,38,128\n",
    "added4_upsample = UpSampling2D((2, 2),name = 'upsample_3')(added_4)\n",
    "# 76,76,128\n",
    "added4_conv = Conv2D(64,(1,1),padding = 'same', name = 'Conv_8',strides = 1)(added4_upsample)\n",
    "# 76,76,64\n",
    "\n",
    "\n",
    "layer_5 = Conv2D(64,(1,1),padding = 'same', name = 'Conv_9',strides = 1)(added_4)\n",
    "layer_5 = BatchNormalization(name = 'BN_6')(layer_5)\n",
    "layer_5 = Activation('relu')(layer_5)\n",
    "layer_5 = UpSampling2D((2, 2),name = 'upsample_4')(layer_5)\n",
    "# 76,76,64\n",
    "\n",
    "added_5 = keras.layers.Add()([added4_conv, layer_5])\n",
    "\n",
    "# 76,76,64\n",
    "added5_conv = Conv2D(32,(1,1),padding = 'same', name = 'Conv_10',strides = 1)(added_5)\n",
    "added5_upsample = UpSampling2D((2, 2),name = 'upsample_5')(added5_conv) \n",
    "# 152,152,32\n",
    "\n",
    "\n",
    "layer_6 = Conv2D(32,(1,1),padding = 'same', name = 'Conv_11',strides = 1)(added_5)\n",
    "layer_6 = BatchNormalization(name = 'BN_7')(layer_6)\n",
    "layer_6 = Activation('relu')(layer_6)\n",
    "layer_6 = UpSampling2D((2, 2),name = 'upsample_6')(layer_6)\n",
    "# 152,152,32\n",
    "\n",
    "added_6 = keras.layers.Add()([added5_upsample, layer_6])\n",
    "# 152,152,32\n",
    "\n",
    "layer_6 = Conv2D(num_classes,(3,3),padding ='valid',name = 'Conv_12')(added_6)\n",
    "layer_6  = BatchNormalization(name = 'BN_8')(layer_6)\n",
    "decoded= Activation('softmax')(layer_6)\n",
    "\n",
    "autoencoder = Model(input_img, decoded)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Plot model\n",
    "plot_model(autoencoder, to_file= model_name + '_model.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "__________________________________________________________________________________________________\n",
      "Layer (type)                    Output Shape         Param #     Connected to                     \n",
      "==================================================================================================\n",
      "input_2 (InputLayer)            (None, 150, 150, 3)  0                                            \n",
      "__________________________________________________________________________________________________\n",
      "Conv_1 (Conv2D)                 (None, 150, 150, 32) 128         input_2[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "BN_1 (BatchNormalization)       (None, 150, 150, 32) 128         Conv_1[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_9 (Activation)       (None, 150, 150, 32) 0           BN_1[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "pool_1 (MaxPooling2D)           (None, 75, 75, 32)   0           activation_9[0][0]               \n",
      "__________________________________________________________________________________________________\n",
      "Conv_2 (Conv2D)                 (None, 75, 75, 32)   1056        pool_1[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "BN_2 (BatchNormalization)       (None, 75, 75, 32)   128         Conv_2[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_10 (Activation)      (None, 75, 75, 32)   0           BN_2[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "pool_2 (MaxPooling2D)           (None, 75, 75, 32)   0           activation_10[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_7 (Add)                     (None, 75, 75, 32)   0           pool_1[0][0]                     \n",
      "                                                                 pool_2[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "Conv_4 (Conv2D)                 (None, 75, 75, 64)   2112        add_7[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "BN_3 (BatchNormalization)       (None, 75, 75, 64)   256         Conv_4[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "Conv_3 (Conv2D)                 (None, 75, 75, 64)   2112        add_7[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_11 (Activation)      (None, 75, 75, 64)   0           BN_3[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "pool_3 (MaxPooling2D)           (None, 38, 38, 64)   0           Conv_3[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "pool_4 (MaxPooling2D)           (None, 38, 38, 64)   0           activation_11[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_8 (Add)                     (None, 38, 38, 64)   0           pool_3[0][0]                     \n",
      "                                                                 pool_4[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "Conv_6 (Conv2D)                 (None, 38, 38, 128)  8320        add_8[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "BN_4 (BatchNormalization)       (None, 38, 38, 128)  512         Conv_6[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "Conv_5 (Conv2D)                 (None, 38, 38, 128)  8320        add_8[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "activation_12 (Activation)      (None, 38, 38, 128)  0           BN_4[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "pool_5 (MaxPooling2D)           (None, 19, 19, 128)  0           Conv_5[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "pool_6 (MaxPooling2D)           (None, 19, 19, 128)  0           activation_12[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_9 (Add)                     (None, 19, 19, 128)  0           pool_5[0][0]                     \n",
      "                                                                 pool_6[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "Conv_7 (Conv2D)                 (None, 19, 19, 128)  16512       add_9[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "BN_5 (BatchNormalization)       (None, 19, 19, 128)  512         Conv_7[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_13 (Activation)      (None, 19, 19, 128)  0           BN_5[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "upsample_1 (UpSampling2D)       (None, 38, 38, 128)  0           add_9[0][0]                      \n",
      "__________________________________________________________________________________________________\n",
      "upsample_2 (UpSampling2D)       (None, 38, 38, 128)  0           activation_13[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_10 (Add)                    (None, 38, 38, 128)  0           upsample_1[0][0]                 \n",
      "                                                                 upsample_2[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "Conv_9 (Conv2D)                 (None, 38, 38, 64)   8256        add_10[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "BN_6 (BatchNormalization)       (None, 38, 38, 64)   256         Conv_9[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "upsample_3 (UpSampling2D)       (None, 76, 76, 128)  0           add_10[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_14 (Activation)      (None, 38, 38, 64)   0           BN_6[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "Conv_8 (Conv2D)                 (None, 76, 76, 64)   8256        upsample_3[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "upsample_4 (UpSampling2D)       (None, 76, 76, 64)   0           activation_14[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_11 (Add)                    (None, 76, 76, 64)   0           Conv_8[0][0]                     \n",
      "                                                                 upsample_4[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "Conv_11 (Conv2D)                (None, 76, 76, 32)   2080        add_11[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "BN_7 (BatchNormalization)       (None, 76, 76, 32)   128         Conv_11[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "Conv_10 (Conv2D)                (None, 76, 76, 32)   2080        add_11[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "activation_15 (Activation)      (None, 76, 76, 32)   0           BN_7[0][0]                       \n",
      "__________________________________________________________________________________________________\n",
      "upsample_5 (UpSampling2D)       (None, 152, 152, 32) 0           Conv_10[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "upsample_6 (UpSampling2D)       (None, 152, 152, 32) 0           activation_15[0][0]              \n",
      "__________________________________________________________________________________________________\n",
      "add_12 (Add)                    (None, 152, 152, 32) 0           upsample_5[0][0]                 \n",
      "                                                                 upsample_6[0][0]                 \n",
      "__________________________________________________________________________________________________\n",
      "Conv_12 (Conv2D)                (None, 150, 150, 41) 11849       add_12[0][0]                     \n",
      "__________________________________________________________________________________________________\n",
      "BN_8 (BatchNormalization)       (None, 150, 150, 41) 164         Conv_12[0][0]                    \n",
      "__________________________________________________________________________________________________\n",
      "activation_16 (Activation)      (None, 150, 150, 41) 0           BN_8[0][0]                       \n",
      "==================================================================================================\n",
      "Total params: 73,165\n",
      "Trainable params: 72,123\n",
      "Non-trainable params: 1,042\n",
      "__________________________________________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "# Complile Model with the same fixed set of hyperparameters\n",
    "autoencoder.compile(optimizer = 'adam',loss = 'categorical_crossentropy',metrics = ['accuracy'])\n",
    "autoencoder.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Code for storing history of model \n",
    "checkpoint = ModelCheckpoint(model_name+'/'+model_name+'.h5',monitor='val_loss',verbose = 2,save_best_only = True,save_weights_only = True, mode = 'auto')\n",
    "earlystop = EarlyStopping(monitor='val_loss', min_delta=0, patience = 4, verbose=0, mode='auto')\n",
    "tensorboard = TensorBoard(log_dir='./'+model_name, histogram_freq=0, batch_size=batch, write_graph=True, write_grads=False, write_images=True, embeddings_freq=0, embeddings_layer_names=None, embeddings_metadata=None)\n",
    "rp = ReduceLROnPlateau(monitor='val_loss', factor=0.2, patience = 4, min_lr=0.001)\n",
    "csvlogger = CSVLogger(model_name+'/logs.csv', separator=',', append= True)\n",
    "callbacks = [checkpoint,tensorboard, earlystop, csvlogger, rp]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# Fine-tune the model with your dataset for segmentation\n",
    "autoencoder.load_weights(model_name+'/'+model_name+'.h5')\n",
    "auto_encoder_train = autoencoder.fit(X_train,Y_train, batch_size = batch , epochs=epochs,callbacks = callbacks,validation_data = (X_test,Y_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "autoencoder.save_weights(model_name + 'my_weights.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "pd.read_csv(model_name+'/logs.csv').plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 68,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQUAAAD8CAYAAAB+fLH0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nD6To4XyuW/AtAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQUAAAD8CAYAAAB+fLH0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n/x+GoeDEhzjadAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAQUAAAD8CAYAAAB+fLH0AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz\nAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDIuMi4yLCBo\ndHRwOi8vbWF0cGxvdGxpYi5vcmcvhp/UCwAAEP9JREFUeJzt3X+s3XV9x/Hn28IFW6QF+bGuMFug\n0zGzjabB7uqMkdlVxsBlutQZ12gXYoYTx4yUkcxmiQnMDZXEYZi44cJEhhobw9htKmwxHcyCYMEC\nrVWhUqk6ta4YS/W9P77fi+fT3dt7fnzPr/b5SG7OOd/zOd/z7re3r36+53zOeUdmIknTXjDsAiSN\nFkNBUsFQkFQwFCQVDAVJBUNBUqFvoRARayLi8YjYFREb+vU8kpoV/VinEBHzgCeA1wF7gC8Bb87M\nrzb+ZJIa1a+ZwoXArszcnZkHgduBy/r0XJIadFyf9rsEeKrl9h7gFbMNnliwMF+46Iw+lSIJYP/T\nu76bmafPNa5foRAzbCvOUyLicuBygBMXns6qd3yoT6VIApj6q0u+2c64fp0+7AHObrl9FvB064DM\nvDkzV2bmyokFC/tUhqRO9SsUvgQsj4hlETEBrAU29em5JDWoL6cPmXkoIt4J/DswD/h4Zj7aj+eS\n1Kx+vaZAZt4F3NWv/UvqD1c0SioYCpIKhoKkgqEgqWAoSCoYCpIKhoKkgqEgqWAoSCoYCpIKhoKk\ngqEgqWAoSCoYCpIKhoKkgqEgqWAoSCoYCpIKhoKkgqEgqdB1KETE2RFxT0TsiIhHI+LKevupEbE5\nInbWl6c0V66kfutlpnAI+IvM/BVgFXBFRJwPbAC2ZOZyYEt9W9KY6DoUMnNvZj5YX/8RsIOqh+Rl\nwK31sFuBN/RapKTBaeQ1hYhYClwA3A+cmZl7oQoOwM6x0hjpORQi4iTg08C7M3N/B4+7PCK2RcS2\ngwd+2GsZkhrSUyhExPFUgXBbZn6m3vxMRCyu718M7JvpsTaYlUZTL+8+BHALsCMzb2i5axOwrr6+\nDvhc9+VJGrReekm+EngrsD0iHqq3/SVwHXBHRKwHngTe1FuJkgap61DIzC8CMcvdF3W7X0nD5YpG\nSQVDQVLBUJBUMBQkFQwFSQVDQVLBUJBUMBQkFQwFSQVDQVLBUJBUMBQkFQwFSQVDQVLBUJBUMBQk\nFQwFSQVDQVLBUJBUMBQkFZpoBjMvIr4cEZ+vby+LiPvrBrOfioiJ3suUNChNzBSupOojOe164IN1\ng9nvA+sbeA5JA9Jrh6izgN8FPlbfDuC1wJ31EBvMSmOm15nCh4D3Aj+rb78Y+EFmHqpv76HqRC1p\nTPTSNu4SYF9mPtC6eYahOcvjbTArjaBe28ZdGhEXAycCJ1PNHBZFxHH1bOEs4OmZHpyZNwM3Ayxc\nsnzG4JA0eF3PFDLzmsw8KzOXAmuBL2TmW4B7gDfWw2wwK42ZfqxTuBq4KiJ2Ub3GcEsfnkNSn/Ry\n+vC8zLwXuLe+vhu4sIn9Sho8VzRKKhgKkgqGgqSCoSCpYChIKhgKkgqGgqSCoSCpYChIKhgKkgqG\ngqSCoSCpYChIKhgKkgqGgqSCoSCpYChIKjTyzUu9WrD/AJNf2DbUGrZObO3sAa96V38KkYbMmYKk\ngqEgqdBr27hFEXFnRDwWETsi4jcj4tSI2Fw3mN0cEac0Vayk/ut1pvBh4O7MfBnw61SNZjcAW+oG\ns1vq25LGRC9t404GXk3d1yEzD2bmD4DLqBrLgg1mpbHTy0zhHOA7wD9GxJcj4mMRsQA4MzP3AtSX\nZzRQp6QB6SUUjgNWADdl5gXAATo4VWhtMPvsc8/2UIakJvUSCnuAPZl5f337TqqQeCYiFgPUl/tm\nenBm3pyZKzNz5fzj5/dQhqQmdb14KTO/HRFPRcRLM/Nx4CLgq/XPOuA6xqjB7OTBybbHbp3YCl+8\nEYDla752xLE7//fDPdUlDVqvKxr/DLgtIiaA3cDbqGYfd0TEeuBJ4E09PoekAeopFDLzIWDlDHdd\n1Mt+R93kwcnnl0XvvPvcOUbf+POrLo1uxOdftrrR/V3y2FSj+xt3I/HZh3HUyekGlKccRzLX6chM\njrVTlCdXTDS6v6k/hCtumCvcK8fCsXaZs6SCoSCpYChIKhgKkgqGgqSCoSCpYChIKhgKkgojsXjp\nwAsOzPnFqZ0uFhpXO+8+t+MFTMtPuvKYWFSjwRiJUGhHa2gcKwHRkTZWSx5ry6zPOw927Rp2FePH\n0wdJhbGZKbTqpEdDp1PxuT/g1MXHrDvUTg2Ha+/PeWVndYzoKcl55/Vn7Dhrp29Kux/7cqYgqWAo\nSCoYCpIKhoKkgqEgqWAoSCoYCpIKPa1TiIg/B/4ESGA71bc5LwZuB04FHgTempkHj7SfE07+ydxf\nld7Be/fdfM9hJ4/denfXux8ro7p8evVff74Pe+1sDcdIqFexNr3Ct5dekkuAdwErM/PlwDxgLXA9\n8MG6wez3gfVNFCppMHpd0Xgc8MKIeA6YD+wFXgv8UX3/rcBG4KYen6en//2PBZ2ugmzneFb7rD9T\ncYx9bmIktPN5lj7opUPUtyLib6kavvyYahXlA8APMvNQPWwPsKTnKkdM6z+obpYkj4KOQ+SkZqfX\ngzotabru6f31rf4OgqBfHwzs5fThFKq288uAXwQWAK+fYWjO8vjnG8z++MBPuy1DUsN6effht4Gv\nZ+Z3MvM54DPAJLAoIqZnIGcBT8/04NYGsy9cMK+HMiQ1qZdQeBJYFRHzIyL4eYPZe4A31mPGpsGs\npErXoVC3oL+T6m3H7fW+bgauBq6KiF3Ai4FbGqhT0oD02mD2fcD7Dtu8G7iwl/1KGh5XNEoqGAqS\nCoaCpMJYfkfj0a7d1ZuDXDg113N187X0z++7jYVATS9C6lW7nwsp/pwN/3316xvODYUeDXP59Sit\nrGx9/o6PSTur+NZ0WFCXOgm/ToOq/aXlw+Xpg6SCMwUNXT9nW03/z9vTjKgN3c7+Nt67sbEanClI\nKhgKkgqGgqSCoSCpYChIKhgKkgqGgqSCoSCp4OKlo8BHrrqi48esXt2/5hX9Xqo7ip8NaVI3da9e\nvXrOMVNTU23ty1CodfsL1M/VeKdvemvf9j01NfeHCboNjn7uux3jGgYwGrV7+iCpEJkzfgP7QC1c\nuDBXrVo17DI60q/v3O/GxntPHnYJPdn4mv1zjmn9mLC6MzU19UBmrpxrnDMFSYU5QyEiPh4R+yLi\nkZZtp0bE5ojYWV+eUm+PiLgxInZFxFciYkU/i5fUvHZmCv/E//+Kiw3AlrqJ7Jb6NlQdopbXP5fT\nQA9JSYM1Zyhk5n8C/3PY5suomsdSX76hZfsnsnIfVbeoxU0VK6n/un1N4czM3AtQX55Rb18CPNUy\n7qhsMCsdzZpepxAzbJu1wSzVKQYnnnhiw2VI6la3M4Vnpk8L6st99fY9wNkt49pqMDsxMdFlGZKa\n1u1MYRNV89jrKJvIbgLeGRG3A68Afjh9mjEs7ayua+d98lG28TX7R2qtQj+OZ+u6kLn+rIP4+zya\n103MGQoR8UngNcBpEbGHqnfkdcAdEbGeqvv0m+rhdwEXA7uAZ4G39aFmSX00Zyhk5ptnueuiGcYm\n0PmnczrUyWrCyTGfBYyLQc62xn1mN+rG6gNRo7S0ePRc1caYG56/5j+s3kwenDxqTyFc5iypYChI\nKhgKkgqGgqSCoSCpYChIKhgKkgqGgqTCSCxeWvCzBS5M6tkNcw+puXBJRzISoaDe+Q9dTfH0QVLB\nUJBUMBQkFQwFSQVDQVLBUJBUMBQkFQwFSQVDQVKh2wazH4iIx+omsp+NiEUt911TN5h9PCJ+p1+F\nS+qPbhvMbgZenpm/BjwBXAMQEecDa4FfrR/z9xExr7FqJfVdVw1mM3MqMw/VN++j6gQFVYPZ2zPz\nJ5n5dar+Dxc2WK+kPmviA1FvBz5VX19CFRLTbDCro1Y73cdWr757AJXA8jVfm3PM1FR7++rphcaI\nuBY4BNw2vWmGYbM2mI2IbRGx7dnnnu2lDEkN6joUImIdcAnwlrozFHTZYHb+8fO7LUNSw7oKhYhY\nA1wNXJqZrf/NbwLWRsQJEbEMWA78d+9lShqUbhvMXgOcAGyOCID7MvMdmfloRNwBfJXqtOKKzPxp\nv4qX1LxuG8zecoTx7wfe30tRkobHFY2SCoaCpIKhIKlgKEgqGAqSCvZ9kLo0qCXMg+ZMQVLBUJBU\nMBQkFQwFSQVDQVLBUJBUMBQkFQwFSQVDQVLBUJBUcJmzdBTYefe5bYya+xufwZmCpMM4U5C6NHlw\ncs4xWye2DqCSZjlTkFToqsFsy33viYiMiNPq2xERN9YNZr8SESv6UbSk/um2wSwRcTbwOuDJls2v\np+r1sBy4HLip9xIlDVJXDWZrHwTeS9kW7jLgE1m5D1gUEYsbqVTSQHTbIepS4FuZ+fBhdy0Bnmq5\nbYNZacx0/O5DRMwHrgVWz3T3DNtmbTBLdYrBwhMWdlqGpD7pZqZwLrAMeDgivkHVRPbBiPgFbDAr\njb2OQyEzt2fmGZm5NDOXUgXBisz8NlWD2T+u34VYBfwwM/c2W7I0PiYPTra1nmGUtPOW5CeB/wJe\nGhF7ImL9EYbfBewGdgH/APxpI1VKGphuG8y23r+05XoCV/RelnR0mTw4OTarG13mLA3I1NRUW+NW\nr57pNfzZLV/T3ged2nx6lzlLKhkKkgqGgqSCoSCpYChIKhgKkgqGgqSCoSCpENUixCEXEfEd4ADw\n3WHX0uI0rGcuo1aT9RzZSzLz9LkGjUQoAETEtsxcOew6plnP3EatJutphqcPkgqGgqTCKIXCzcMu\n4DDWM7dRq8l6GjAyrylIGg2jNFOQNAKGHgoRsSYiHq8byGwYUg1nR8Q9EbEjIh6NiCvr7Rsj4lsR\n8VD9c/EAa/pGRGyvn3dbve3UiNgcETvry1MGVMtLW47BQxGxPyLePejjM1NjotmOySAaE81Szwci\n4rH6OT8bEYvq7Usj4sctx+qjTdfTmMwc2g8wj6oV7jnABPAwcP4Q6lhM9T2TAC8CngDOBzYC7xnS\nsfkGcNph2/4G2FBf3wBcP6S/s28DLxn08QFeDawAHpnrmAAXA/9G9Q3jq4D7B1TPauC4+vr1LfUs\nbR03yj/DnilcCOzKzN2ZeRC4naqhzEBl5t7MfLC+/iNgB6PZr+Iy4Nb6+q3AG4ZQw0XA1zLzm4N+\n4py5MdFsx6TvjYlmqiczpzLzUH3zPqpvNB8rww6FkWseExFLgQuA++tN76yngh8f1HS9lsBURDxQ\n98gAODPrb8euL88YYD3T1gKfbLk9rOMzbbZjMgq/W2+nmq1MWxYRX46I/4iI3xpwLW0bdii03Txm\nECLiJODTwLszcz9VL8xzgd8A9gJ/N8ByXpmZK6j6c14REa8e4HPPKCImgEuBf603DfP4zGWov1sR\ncS1wCLit3rQX+KXMvAC4CviXiDh5UPV0Ytih0HbzmH6LiOOpAuG2zPwMQGY+k5k/zcyfUX1l/YWD\nqiczn64v9wGfrZ/7mekpcH25b1D11F4PPJiZz9S1De34tJjtmAztdysi1gGXAG/J+gWFzPxJZn6v\nvv4A1WtpvzyIejo17FD4ErA8IpbV/wutpWooM1AREcAtwI7MvKFle+s56O8Djxz+2D7VsyAiXjR9\nnerFq0eojs26etg64HODqKfFm2k5dRjW8TnMbMdkKI2JImINcDVwaWY+27L99IiYV18/h6oz++5+\n19OVYb/SSfUq8RNUyXntkGp4FdXU8ivAQ/XPxcA/A9vr7ZuAxQOq5xyqd2IeBh6dPi7Ai4EtwM76\n8tQBHqP5wPeAhS3bBnp8qAJpL/Ac1Uxg/WzHhOr04SP179V2YOWA6tlF9VrG9O/RR+uxf1D/XT4M\nPAj83jB+19v5cUWjpMKwTx8kjRhDQVLBUJBUMBQkFQwFSQVDQVLBUJBUMBQkFf4PD/Ddrpu8TFsA\nAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "index = 12\n",
    "inp = np.zeros((1,input_Shape[0],input_Shape[1],3))\n",
    "inp[0] = X_test[index]\n",
    "predicted = autoencoder.predict(inp)\n",
    "out = map_img(predicted[0])\n",
    "plt.imshow(X_test[index])\n",
    "plt.show()\n",
    "os.listdir()\n",
    "lb = plt.imread('./gt/' + x_test[index])\n",
    "plt.imshow(cv2.resize(lb,(input_Shape[0],input_Shape[1])))\n",
    "plt.show()\n",
    "plt.imshow(out)\n",
    "plt.show()\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}