EAI6010_Burial_Records_Project.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Importing Packages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import numpy as np \n",
    "import pandas as pd \n",
    "import matplotlib.pyplot as plt\n",
    "import tensorflow as tf\n",
    "from tensorflow.keras.models import Sequential\n",
    "from tensorflow.keras.layers import Dense, Flatten, Dropout, Activation, Conv2D, MaxPooling2D, Input\n",
    "from tensorflow.keras.preprocessing.image import load_img\n",
    "from tensorflow.keras.preprocessing.image import img_to_array\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.model_selection import StratifiedKFold\n",
    "\n",
    "# Vgg16 utils\n",
    "from tensorflow.keras.applications.vgg16 import VGG16\n",
    "from tensorflow.keras.applications.vgg16 import preprocess_input as vgg_preprocess \n",
    "\n",
    "# Resnet utils \n",
    "from tensorflow.keras.applications.resnet50 import ResNet50\n",
    "from tensorflow.keras.applications.resnet50 import preprocess_input as resnet_preprocess\n",
    "\n",
    "# utils to dump \n",
    "from pickle import dump\n",
    "# import glob\n",
    "import glob"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Labelling the Forms"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "data_dir = \"/Users/varun/trunk/kumud/COURSES/EAI6010/Final_Project/forms/\"\n",
    "\n",
    "X = []\n",
    "Y = []\n",
    "for img_path in glob.glob(data_dir+\"*/*.jpg\"):\n",
    "    img = img_path.split(\"/\")\n",
    "    label = None\n",
    "    if img[-2] == \"Type-1 blocks\":\n",
    "        label = 1\n",
    "    elif img[-2] == \"Type-2 Empty\":\n",
    "        label = 2\n",
    "    elif img[-2] == \"Type-3 lines\":\n",
    "        label = 3\n",
    "    elif img[-2] == \"HandWritten\":\n",
    "        label = 0\n",
    "    else:\n",
    "        print(\"Unknown label type\", img[-2])\n",
    "    Y.append(label)\n",
    "    X.append(img_to_array(load_img(img_path,target_size=(224, 224))))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "120 120\n",
      "(224, 224, 3)\n"
     ]
    }
   ],
   "source": [
    "print(len(X),len(Y))\n",
    "print(X[1].shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Classification"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "dense (Dense)                (None, 224, 224, 64)      256       \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 224, 224, 128)     8320      \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 6422528)           0         \n",
      "_________________________________________________________________\n",
      "dense_2 (Dense)              (None, 4)                 25690116  \n",
      "=================================================================\n",
      "Total params: 25,698,692\n",
      "Trainable params: 25,698,692\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n",
      "Train on 96 samples\n",
      "Epoch 1/10\n",
      "96/96 [==============================] - 14s 146ms/sample - loss: 50882.3219 - accuracy: 0.3646\n",
      "Epoch 2/10\n",
      "96/96 [==============================] - 9s 97ms/sample - loss: 20409.7503 - accuracy: 0.3646\n",
      "Epoch 3/10\n",
      "96/96 [==============================] - 10s 100ms/sample - loss: 5191.6914 - accuracy: 0.4583\n",
      "Epoch 4/10\n",
      "96/96 [==============================] - 10s 105ms/sample - loss: 1120.8408 - accuracy: 0.5521\n",
      "Epoch 5/10\n",
      "96/96 [==============================] - 11s 111ms/sample - loss: 435.5080 - accuracy: 0.6667\n",
      "Epoch 6/10\n",
      "96/96 [==============================] - 9s 98ms/sample - loss: 147.9889 - accuracy: 0.7812\n",
      "Epoch 7/10\n",
      "96/96 [==============================] - 11s 117ms/sample - loss: 58.5417 - accuracy: 0.8542\n",
      "Epoch 8/10\n",
      "96/96 [==============================] - 10s 99ms/sample - loss: 72.2298 - accuracy: 0.9271\n",
      "Epoch 9/10\n",
      "96/96 [==============================] - 9s 96ms/sample - loss: 8.6465 - accuracy: 0.9271\n",
      "Epoch 10/10\n",
      "96/96 [==============================] - 9s 94ms/sample - loss: 25.2433 - accuracy: 0.8958\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "<tensorflow.python.keras.callbacks.History at 0x1a4b0566a0>"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Build first model\n",
    "\n",
    "model1 = Sequential([\n",
    "    Dense(64,input_shape = (224, 224, 3), activation='relu'),\n",
    "    Dense(128, activation ='relu'),\n",
    "    Flatten(),\n",
    "    Dense(4, activation='softmax')])\n",
    "\n",
    "model1.summary()\n",
    "model1.compile(optimizer=tf.keras.optimizers.Adam(0.001),\n",
    "                  loss='categorical_crossentropy',\n",
    "                  metrics=['accuracy'])\n",
    "\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, Y, test_size=0.20, random_state = 123)\n",
    "\n",
    "y_train_cat = tf.keras.utils.to_categorical(y_train)\n",
    "y_test_cat = tf.keras.utils.to_categorical(y_test)\n",
    "model1.fit(np.array(X_train), np.array(y_train_cat), epochs=10, batch_size=10,verbose =1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Confusion Matrix\n",
      "[[0 0 3 0]\n",
      " [0 8 0 0]\n",
      " [0 0 7 0]\n",
      " [0 0 2 4]]\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      " HandWritten       0.00      0.00      0.00         3\n",
      "Type 1 Block       1.00      1.00      1.00         8\n",
      "Type 2 Empty       0.58      1.00      0.74         7\n",
      "type 3 Lines       1.00      0.67      0.80         6\n",
      "\n",
      "   micro avg       0.79      0.79      0.79        24\n",
      "   macro avg       0.65      0.67      0.63        24\n",
      "weighted avg       0.75      0.79      0.75        24\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/varun/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1143: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\r",
      "24/1 [================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================] - 1s 41ms/sample - loss: 428.0718 - accuracy: 0.7917\n"
     ]
    }
   ],
   "source": [
    "# Accuracy on test data\n",
    "y_pred = model1.predict(np.array(X_test))\n",
    "y_pred = np.argmax(y_pred, axis=1)\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "print('Confusion Matrix')\n",
    "print(confusion_matrix(y_test, y_pred))\n",
    "print('Classification Report')\n",
    "target_names = ['HandWritten','Type 1 Block', 'Type 2 Empty', 'type 3 Lines']\n",
    "print(classification_report(y_test, y_pred, target_names=target_names))\n",
    "score = model1.evaluate(np.array(X_test), np.array(y_test_cat), verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 720x720 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt     \n",
    "\n",
    "ax= plt.subplot()\n",
    "plt.figure(figsize = (10,10))\n",
    "sns.heatmap(cm, annot=True, ax = ax,fmt=\"d\",linewidths=.5); #annot=True to annotate cells\n",
    "sns.set(font_scale=1.4)#for label size\n",
    "\n",
    "\n",
    "# labels, title and ticks\n",
    "ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); \n",
    "ax.set_title('Confusion Matrix'); \n",
    "ax.xaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']); ax.yaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']);\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Transfer Learning - VGG"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "# extract features from each photo in the directory\n",
    "#load the model, extracting features from each photo\n",
    "# prepare image for VGG model, get features, Storing Features\n",
    "\n",
    "def extract_features_vgg(X):\n",
    "    in_layer = Input(shape=(224, 224, 3))\n",
    "    model = VGG16(include_top=False, input_tensor=in_layer)\n",
    "    features = []\n",
    "    for image in X:\n",
    "        image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))\n",
    "        feature = model.predict(image, verbose=0)\n",
    "        features.append(feature)\n",
    "    return features\n",
    "\n",
    "\n",
    "X_vgg_features = extract_features_vgg(X)\n",
    "y_cat = tf.keras.utils.to_categorical(Y)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(hidden_size = 128):\n",
    "    import tensorflow.keras as keras\n",
    "    fnn_model = keras.Sequential()\n",
    "    fnn_model.add(keras.layers.Flatten())\n",
    "    fnn_model.add(keras.layers.Dense(hidden_size, activation='relu'))\n",
    "    fnn_model.add(keras.layers.Dropout(0.2))\n",
    "    fnn_model.add(keras.layers.Dense(4, activation='softmax'))\n",
    "    fnn_model.compile(optimizer=tf.keras.optimizers.Adam(0.001),loss='categorical_crossentropy',metrics=['accuracy'])\n",
    "    return fnn_model\n",
    "\n",
    "from sklearn.model_selection import train_test_split\n",
    "X_train, X_test, y_train, y_test = train_test_split(X_vgg_features, Y, test_size = 0.2,random_state = 1234)\n",
    "\n",
    "fnn_model = build_model(hidden_size=128)\n",
    "fnn_model.fit(np.array(X_vgg_features), y_cat , epochs=10, batch_size=128,verbose =0)\n",
    "y_pred = fnn_model.predict(np.array(X_test))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3 1 2 2 0 1 1 3 1 0 1 3 2 2 1 3 2 2 0 2 0 1 3 3]\n",
      "Confusion Matrix\n",
      "[[4 0 0 0]\n",
      " [0 7 0 0]\n",
      " [0 0 7 0]\n",
      " [0 0 0 6]]\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      " HandWritten       1.00      1.00      1.00         4\n",
      "Type 1 Block       1.00      1.00      1.00         7\n",
      "Type 2 Empty       1.00      1.00      1.00         7\n",
      "type 3 Lines       1.00      1.00      1.00         6\n",
      "\n",
      "   micro avg       1.00      1.00      1.00        24\n",
      "   macro avg       1.00      1.00      1.00        24\n",
      "weighted avg       1.00      1.00      1.00        24\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_pred = np.argmax(y_pred, axis=1)\n",
    "print(y_pred)\n",
    "len(y_pred)\n",
    "\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "print('Confusion Matrix')\n",
    "print(confusion_matrix(y_test, y_pred))\n",
    "print('Classification Report')\n",
    "target_names = ['HandWritten','Type 1 Block', 'Type 2 Empty', 'type 3 Lines']\n",
    "print(classification_report(y_test, y_pred, target_names=target_names))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 720x720 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm = confusion_matrix(y_test, y_pred)\n",
    "\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt     \n",
    "\n",
    "ax= plt.subplot()\n",
    "plt.figure(figsize = (10,10))\n",
    "sns.heatmap(cm, annot=True, ax = ax,fmt=\"d\",linewidths=.5); #annot=True to annotate cells\n",
    "sns.set(font_scale=1.4)#for label size\n",
    "\n",
    "\n",
    "# labels, title and ticks\n",
    "ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); \n",
    "ax.set_title('Confusion Matrix'); \n",
    "ax.xaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']); ax.yaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']);\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[3 3 1 2 2 2 1 2 3 2 1 1 1 1 2 2 3 1 2 2 2 3 1 3]\n",
      "Confusion Matrix\n",
      "[[ 0  0  0  1]\n",
      " [ 0  8  0  0]\n",
      " [ 0  0 10  0]\n",
      " [ 0  0  0  5]]\n",
      "Classification Report\n",
      "              precision    recall  f1-score   support\n",
      "\n",
      " HandWritten       0.00      0.00      0.00         1\n",
      "Type 1 Block       1.00      1.00      1.00         8\n",
      "Type 2 Empty       1.00      1.00      1.00        10\n",
      "type 3 Lines       0.83      1.00      0.91         5\n",
      "\n",
      "   micro avg       0.96      0.96      0.96        24\n",
      "   macro avg       0.71      0.75      0.73        24\n",
      "weighted avg       0.92      0.96      0.94        24\n",
      "\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/Users/varun/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1143: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/Users/varun/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1143: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n",
      "/Users/varun/anaconda3/lib/python3.6/site-packages/sklearn/metrics/classification.py:1143: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples.\n",
      "  'precision', 'predicted', average, warn_for)\n"
     ]
    }
   ],
   "source": [
    "y_pred1 = np.argmax(y_pred, axis=1)\n",
    "print(y_pred1)\n",
    "len(y_pred1)\n",
    "\n",
    "from sklearn.metrics import classification_report, confusion_matrix\n",
    "print('Confusion Matrix')\n",
    "print(confusion_matrix(y_test, y_pred1))\n",
    "print('Classification Report')\n",
    "target_names = ['HandWritten','Type 1 Block', 'Type 2 Empty', 'type 3 Lines']\n",
    "print(classification_report(y_test, y_pred1, target_names=target_names))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "text/plain": [
       "<Figure size 720x720 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_pred1 = np.argmax(y_pred, axis=1)\n",
    "cm = confusion_matrix(y_test, y_pred1)\n",
    "\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt     \n",
    "\n",
    "ax= plt.subplot()\n",
    "plt.figure(figsize = (10,10))\n",
    "sns.heatmap(cm, annot=True, ax = ax,fmt=\"d\",linewidths=.5); #annot=True to annotate cells\n",
    "sns.set(font_scale=1.4)#for label size\n",
    "\n",
    "\n",
    "# labels, title and ticks\n",
    "ax.set_xlabel('Predicted labels');ax.set_ylabel('True labels'); \n",
    "ax.set_title('Confusion Matrix'); \n",
    "ax.xaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']); ax.yaxis.set_ticklabels(['HW','Type-1', 'Type-2', 'type-3']);\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\swamireddy.v\\Anaconda3\\lib\\site-packages\\sklearn\\model_selection\\_split.py:657: Warning: The least populated class in y has only 2 members, which is too few. The minimum number of members in any class cannot be less than n_splits=6.\n",
      "  % (min_groups, self.n_splits)), Warning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The accuracy for 1 is 1.00\n",
      "The accuracy for 2 is 0.90\n",
      "The accuracy for 3 is 1.00\n",
      "The accuracy for 4 is 0.95\n",
      "The accuracy for 5 is 1.00\n",
      "The accuracy for 6 is 1.00\n",
      "The average accuracy on 10-fold cross validation is 0.98 with standard deviation 0.04\n"
     ]
    }
   ],
   "source": [
    "kfold = StratifiedKFold(n_splits=6, shuffle=True, random_state=1234)\n",
    "cvscores_vgg = []\n",
    "for train, test in kfold.split(X, Y):\n",
    "    fnn_model = build_model(hidden_size=128)\n",
    "    fnn_model.fit(np.array(X_vgg_features)[train], y_cat[train] , epochs=10, batch_size=128,verbose =0)\n",
    "    score = fnn_model.evaluate(np.array(X_vgg_features)[test],y_cat[test],verbose=0)\n",
    "    cvscores_vgg.append(score[1])\n",
    "    \n",
    "    \n",
    "for i,score in enumerate(cvscores_vgg):\n",
    "    print(\"The accuracy for {} is {:.2f}\".format(i+1,score))\n",
    "    \n",
    "print(\"The average accuracy on 10-fold cross validation is {:.2f} with standard deviation {:.2f}\"\\\n",
    "      .format(np.mean(cvscores_vgg),np.std(cvscores_vgg)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "def extract_features_resnet(X):\n",
    "# load the model\n",
    "    in_layer = Input(shape=(224, 224, 3))\n",
    "    model = ResNet50(include_top=False, input_tensor=in_layer)\n",
    "    # extract features from each photo\n",
    "    features = []\n",
    "    for image in X:\n",
    "        # reshape data for the model\n",
    "        image = image.reshape((1, image.shape[0], image.shape[1], image.shape[2]))\n",
    "        # prepare the image for the ResNet model\n",
    "        image = resnet_preprocess(image)\n",
    "        # get features\n",
    "        feature = model.predict(image, verbose=0)\n",
    "        # store feature\n",
    "        features.append(feature)\n",
    "    return features\n",
    "\n",
    "y_cat = tf.keras.utils.to_categorical(Y)\n",
    "\n",
    "\n",
    "X_resnet_features = extract_features_resnet(X)\n",
    "#Ref: https://www.tensorflow.org/guide/keras/overview\n",
    "# build our own model\n",
    "def build_model(hidden_size = 128):\n",
    "    # build a feed forward NN \n",
    "    import tensorflow.keras as keras\n",
    "\n",
    "    fnn_model = keras.Sequential()\n",
    "    # Add a densely-connected layer \n",
    "    fnn_model.add(keras.layers.Flatten())\n",
    "    fnn_model.add(keras.layers.Dense(hidden_size, activation='relu'))\n",
    "    fnn_model.add(keras.layers.Dropout(0.2))\n",
    "    # Add a softmax layer with 4 output units:\n",
    "    fnn_model.add(keras.layers.Dense(4, activation='softmax'))\n",
    "\n",
    "    fnn_model.compile(optimizer=tf.keras.optimizers.Adam(0.001),\n",
    "                  loss='categorical_crossentropy',\n",
    "                  metrics=['accuracy'])\n",
    "    return fnn_model\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\swamireddy.v\\Anaconda3\\lib\\site-packages\\sklearn\\model_selection\\_split.py:657: Warning: The least populated class in y has only 2 members, which is too few. The minimum number of members in any class cannot be less than n_splits=6.\n",
      "  % (min_groups, self.n_splits)), Warning)\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "The accuracy for Resnet fold 1 is 1.00\n",
      "The accuracy for Resnet fold 2 is 0.95\n",
      "The accuracy for Resnet fold 3 is 1.00\n",
      "The accuracy for Resnet fold 4 is 0.95\n",
      "The accuracy for Resnet fold 5 is 1.00\n",
      "The accuracy for Resnet fold 6 is 1.00\n",
      "The average accuracy on 6-fold cross validation for Resnet is 0.98 with standard deviation 0.02\n"
     ]
    }
   ],
   "source": [
    "cvscores_resnet = []\n",
    "for train, test in kfold.split(X, Y):\n",
    "    fnn_model = build_model(hidden_size=256)\n",
    "    fnn_model.fit(np.array(X_resnet_features)[train], y_cat[train] , epochs=10, batch_size=32,verbose =0)\n",
    "    score = fnn_model.evaluate(np.array(X_resnet_features)[test],y_cat[test],verbose=0)\n",
    "    cvscores_resnet.append(score[1])\n",
    "    \n",
    "# print 6 fold cross validation acccuracy for Resnet50\n",
    "for i,score in enumerate(cvscores_resnet):\n",
    "    print(\"The accuracy for Resnet fold {} is {:.2f}\".format(i+1,score))\n",
    "    \n",
    "# taking average of accuracy of k fold cv\n",
    "print(\"The average accuracy on 6-fold cross validation for Resnet is {:.2f} with standard deviation {:.2f}\"\\\n",
    "      .format(np.mean(cvscores_resnet),np.std(cvscores_resnet)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for type 1 Blocks\n",
    "from PIL import Image, ImageOps\n",
    " \n",
    "def crop(image_path, border,saved_location):\n",
    "    img = Image.open(image_path)\n",
    "    cropped_image = ImageOps.crop(img, border)\n",
    "    cropped_image.save(saved_location)\n",
    "    \n",
    "if __name__ == '__main__':\n",
    "    img_dir = 'E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Type-1 blocks'\n",
    "    for img_path in glob.glob(img_dir+\"\\\\*.jpg\"):\n",
    "        img = img_path.split(\"\\\\\")\n",
    "        location = 'E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Croppedimages\\\\Type-1 blocks\\\\'\n",
    "        saved_location = location + img[-1]\n",
    "        crop(img_path, (0, 0, 0, 410),saved_location)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for type 2 Empty\n",
    "from PIL import Image, ImageOps\n",
    " \n",
    "def crop(image_path, border,saved_location):\n",
    "    img = Image.open(image_path)\n",
    "    cropped_image = ImageOps.crop(img, border)\n",
    "    cropped_image.save(saved_location)\n",
    "    \n",
    "if __name__ == '__main__':\n",
    "    img_dir = 'E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Type-2 Empty'\n",
    "    for img_path in glob.glob(img_dir+\"\\\\*.jpg\"):\n",
    "        img = img_path.split(\"\\\\\")\n",
    "        location = 'E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Croppedimages\\\\Type-2 Empty\\\\'\n",
    "        saved_location = location + img[-1]\n",
    "        crop(img_path, (0, 0, 100, 50),saved_location)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# for type 3 lines\n",
    "from PIL import Image, ImageOps\n",
    " \n",
    "def crop(image_path, border,saved_location):\n",
    "    img = Image.open(image_path)\n",
    "    cropped_image = ImageOps.crop(img, border)\n",
    "    cropped_image.save(saved_location)\n",
    "\n",
    "if __name__ == '__main__':\n",
    "    img_dir = 'E:\\\\Analytics\\\\AIapplications\\\\AIimages\\Type-3 lines'\n",
    "    for img_path in glob.glob(img_dir+\"\\\\*.jpg\"):\n",
    "        img = img_path.split(\"\\\\\")\n",
    "        location = \"E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Croppedimages\\\\Type-3 lines\\\\\"\n",
    "        saved_location = location + img[-1]\n",
    "        crop(img_path, (0, 0, 0, 250),saved_location)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "DECEDENT AGE INTERMENT NO\n",
      "DABA, AYEHUSH H (FEMALE) 68 NA\n",
      "DATE OF INTERMENT SECTION BLOCK LOT-TER- ROW GRICA/ NI VETERAN\n",
      "3/11/2016 33 225 ROW 2 6 NO\n",
      "FUNERAL HOME VAULT\n",
      "GATLING GR BOX #30\n",
      "\n",
      "LOT OWNER RELATION\n",
      "NEXT OF KIN RELATION\n",
      "\n",
      "TADESSE DABA HUSBAND\n",
      "ADDRESS\n",
      "\n",
      "7217 S MAY\n",
      "\n",
      "CHICAGO, IL 60617 702-292-0263\n",
      "MARKER SINGLE (CICOMPANION OmonuMENT \"““\" C)in Clout\n",
      "\n",
      " \n",
      "\n",
      "mre OGRANITE 0 BRONZE/GRANITE 0 VETERAN\n",
      "\n",
      " \n",
      "\n",
      "DATE SET EMAIL $4 q,\n",
      "OsctaseD\n",
      "\n",
      "DABA, LEMANEE\n",
      "\n",
      "rT ERNE BST 08\n",
      "DATE OF INTERMENT BECTON\n",
      "\n",
      "LOT-TER on VITHRAR\n",
      "6/24/96 33 233R0Wi| 6 Pe [vo\n",
      "OMOUMAGATLING'S CHAPEL\n",
      "GRAVE BOX #30\n",
      "\n",
      " \n",
      "\n",
      "VAULT\n",
      "\n",
      "LoTownin\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "NEXT OF mi\n",
      "\n",
      "TADESSE DABA, FATHER =“\n",
      "7217 S. MAY, CHGO., IL 60621\n",
      "ECEASED INTERMENT WO\n",
      "\n",
      "DABA, SHORRO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF INTERMENT SECTION LOT-TER an ‘age VETERAN\n",
      "9/19/96 33 233R0W1 2 18 NO\n",
      "URDERTARER\n",
      "\n",
      "GATLING'S CHAPEL\n",
      "GRAVE BOX #30\n",
      "\n",
      "LOT OWNER RELATION\n",
      "\n",
      "VAULT\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "WEXT OF We\n",
      "\n",
      "TADESSE DABA, -MeeHER FATHEC\n",
      "7217 §. MAY, CHICAGO, IL 60621\n",
      "\n",
      "— BT 6. MY AEX Lradlépen Peony\n",
      "San fH+G3-36/9: “\n",
      "\n",
      "774_=\n",
      "DICEAS! eRatS Oot\n",
      "DABE, PATRICIA JEAN (CREMAINS)|\n",
      "DATE OF TERMENT SECTION LOT-TER an ag WEVERAR\n",
      "5/11/96 | | U [B10 | |\n",
      "URDERTARER\n",
      "\n",
      "FANELY HAS 02443:\n",
      "waar PLASTIC CUNTAINER\n",
      "LOT OWKER atlanon\n",
      "wevormm “ RODNEY G. DABE, III aov Mann\n",
      "\n",
      " \n",
      "\n",
      "T8921 BAKER AVE., CUUNTHY CLUS AITLs,\n",
      "REMARKS IL 60478\n",
      "permit#\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DACHASID ON ERT BO\n",
      "DABNEY, ABRAHAM L. 619859\n",
      "DATE OF ImTEAME ST sEcnon 1O1-TER aa aos VETERAR\n",
      "12/8/97 28 16 | 6 {si yes\n",
      "\n",
      "ial GATLINGS CHAPEL\n",
      "VAULT\n",
      "#30 Gr. Bx.\n",
      "LOT Owen a@tanoa\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "EXT OF ime\n",
      "\n",
      "atLAnom\n",
      "BRENDA A WALLACE DAUGHTER\n",
      "11442 S$ LOWE CHGO 60628\n",
      "\n",
      "Remanas\n",
      "DECEDENT ace INTERMENT NO\n",
      "DABNEY, ERWIN OTIS NA\n",
      "DATE OF INTERMENT LOT-TER ROW GAICR/ MS VETERAN\n",
      "11/28/2015 30 ROW 2 [5 NO\n",
      "FUNERAL HOME VAULT\n",
      "GR BOX #30\n",
      "LOT OWNER RELATION\n",
      "NEXT OF KIN RELATION\n",
      "JUANITA DABNEY\n",
      "aopness 22075 BROOK AVE\n",
      "MARKER T)stNGLE [CICOMPANION Cimonument “““* Cin Dour\n",
      "\n",
      " \n",
      "\n",
      "TYPE\n",
      "\n",
      "OOGRANITE (J BRONZE/GRANITE O) VETERAN\n",
      "\n",
      "DATE SET EMAIL\n",
      "DECEASED (NTERMENT RO\n",
      "DABNEY, FLORENCE S.\n",
      "\n",
      "DATE Gr ‘SSTERMENT secnon LOT-TER an ABE VETERAM\n",
      "‘Oct. 31, 1989 | 34 29 10 | 84 NO\n",
      "UMDERTARER Taylor F.H.\n",
      "\n",
      "wautr cv #9\n",
      "\n",
      "LOT OWNER RELATION\n",
      "\n",
      "ENT OF CLA Toes\n",
      "\n",
      "Michelle V. Agins Granddaughter ma\n",
      "372 DeKalb 6B\n",
      "\n",
      " \n",
      "\n",
      "REMARES: Brooklyn, N.Y. 11265\n",
      "oeciasiD ONT E Re wT RD\n",
      "\n",
      "DABNEY, JERMAINE LEMONT\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "OATE OF INTEARE NT section LOT- \"OW aa veTiAan\n",
      "10/9/2003 {240A sf 2 30 ‘| NO\n",
      "UNDEATAGER\n",
      "\n",
      "LEAK & SONS\n",
      "WAULT\n",
      "\n",
      "GRAVE BOX #30\n",
      "LOT Gwnea etLAnou\n",
      "ad VICTORIA DEAN (FIANCE) nuaros\n",
      "\n",
      " \n",
      "\n",
      "9236 S. DREXEL\n",
      "CHICAGO, IL 60619\n",
      "773/768-5158\n",
      "\n",
      "AEMARGS\n",
      "DECEASED\n",
      "\n",
      "DABNEY, MOSES\n",
      "\n",
      "Tem ATT OO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "NA\n",
      "DATE OF I TERBEL RET Become Lot-TER a | nar vaTanany\n",
      "11/27/06 | 29 [ac BeR 2 | 68 | no\n",
      "UNDEATALER\n",
      "TAYLOR\n",
      "vaurr\n",
      "GR BOX #30\n",
      "LOT ownte mLATOS\n",
      "EMMA DABNEY (WIFE)\n",
      "ExT OF Ki RELATION\n",
      "11648 SOUTH BISHOP\n",
      "CHICAGO, IL. 60643 (773) 468-0358\n",
      "OaciasID\n",
      "\n",
      "DABNEY, THADDEUS, W.\n",
      "\n",
      "dant 8 Oram OFT Bex\n",
      "\n",
      "NA\n",
      "\n",
      " \n",
      "\n",
      "viTiaan\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF MITE RaSG HT srcrnom - aa agi\n",
      "\"Row 2\n",
      "4/8/09 | 30 [fPok F | 5 46 No\n",
      "\n",
      "UNDTATAKER\n",
      "\n",
      "GATLING'S\n",
      "vault\n",
      "\n",
      "GR_BOX #30\n",
      "LOT Oweda ANOS\n",
      "\n",
      "ORA DABNEY (MOTHER)\n",
      "ReLAnOs\n",
      "\n",
      "suvorast 5430 SOUTH MARSHFIELD\n",
      "\n",
      "CHICAGO, IL. 60609\n",
      "\n",
      "(773) 434-7887\n",
      "DECEASED INTERMENT NO\n",
      "\n",
      "DABNEY, THOMAS\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "CATE OF INTERMENT SECTION LOT-TER arn AGE VETERAN\n",
      "12/28/92 33 26 14 | 62 NO\n",
      "\n",
      "UNDERTAKER GATLING' S\n",
      "\n",
      "var GR. BOX 430\n",
      "\n",
      "LOT CWNEA RELATION\n",
      "\n",
      "MOTOFEN MARY DABNEY = _—_—sWIFE muanon\n",
      "\n",
      " \n",
      "\n",
      "9531 S. PRAIRIE, CHGO., IL 60628\n",
      "312/785-5841\n",
      "\n",
      "REMARKS\n",
      "DECTASED DIT ERMEOT OO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DACE, FRANK\n",
      "DATE Of meTLARAEOT sicnom LOT-TEa | aa | |\n",
      "3/11/95 34 W% 101 {SOUTH 66 YES\n",
      "omererae™\" HALSTED FUNERAL HOME\n",
      "“n\" GR. BOX #30 ome\n",
      "Tor owner . ae st Frank Dac e da\n",
      "ao uae 773- 734-06 #9\n",
      "RSELATION\n",
      "\n",
      " \n",
      "\n",
      "METOFEN(WIFE) JENNIE DACE\n",
      "\n",
      "7606 S. CREIGER AVE.\n",
      "BEMARES CHICAGO, IL. 60649\n",
      "DeceasED crv maeg art pe\n",
      "\n",
      "DADDIO, SYLVIA NA\n",
      "OATE OF meTe man ost petcnos Lor-Tia on aoe viTiaas\n",
      "3/4/10 27 [ies 548 | 1 NA | NO\n",
      "Gaowmramta BEVERLY RIDGE\n",
      "vay\" MONTICELLO #30\n",
      "\n",
      "LOT Owma\n",
      "\n",
      " \n",
      "\n",
      "JOANN DADDIO (DAUGHTER)\n",
      "BERT ORK 9701 HOYNE ananoe\n",
      "CHICAGO, IL. 60643 (773) 445-8773\n",
      "DEectasio\n",
      "DADE, CHONI C. | N/A\n",
      "DATE Of WNTEARMENT stction LOT-TeR aa Ld veTEenan\n",
      "3/22/2003 18 400A | 1 Ta | NO\n",
      "UNDIATAREA\n",
      "LEAK\n",
      "vault\n",
      "GR_BX #30\n",
      "LOT Ownen ailanoa\n",
      "ERT OF Rime RELATION\n",
      "MR. & MRS. ANTHONY DADE canmare\n",
      "8014 S. TROY\n",
      "Rian CHICAGO, IL. 60652\n",
      "\n",
      " \n",
      "\n",
      "unat foo saa\n",
      "\n",
      "Hit hte\n",
      "DICLasiD WAT gene BT 90\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DADE, EDMOND W. NA\n",
      "\n",
      "OATH OF TU AMENT s2cnom Lol-TiA an aa wiTanan\n",
      "10/04/04 33 7 ROW 3 | 3 |%0 NO\n",
      "UNDIATARaA LEAK\n",
      "vue GR BOX #30\n",
      "LOT Ownta aLLATIOS\n",
      "SAMUEL A. WALKER (UNCLE)\n",
      "\n",
      "surorane 8923 S. UNIVERSITY snanos\n",
      "\n",
      " \n",
      "\n",
      "CHICAGO, IL. 60619 (773) 731-4030\n",
      "\n",
      "_harker\n",
      "Orceased\n",
      "DADE, LOVE NA\n",
      "OAT? Oo mute mast wt |e [36. 2] aa ie ee\n",
      "5/8/10 29 | RPok et a 1 oo | yo\n",
      "UNOURTARER\n",
      "GATLING'S\n",
      "vauLT\n",
      "GR BOX #30\n",
      "lor ownsa agLaTIoN\n",
      "ROBERT DADE (HUSBAND)\n",
      "ext OF km RELATOw\n",
      "\n",
      "3214 WEST 63RD PLACE\n",
      "CHICAGO, IL. 60620\n",
      "\n",
      " \n",
      "\n",
      "REMARAE\n",
      "OsceaseD TERI AT AO\n",
      "\n",
      "DADE, ROBERT\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "OATH OF mTeaaint maicnos ‘Bk 1 VETERAR\n",
      "2/17/2012 30 | Brock B \"5 “01 NO\n",
      "UNDE ATAKER CARTER\n",
      "ven GRBOX#30\n",
      "Lot OwnEa anano\n",
      "RELATION\n",
      "\n",
      " \n",
      "\n",
      "MUXT OF Hi BARBARA HOWARD (DAUGHTER)\n",
      "2128 RIDGE LANE UNIT 3\n",
      "ameanns WOODRIDGE, IL. 60517\n",
      "(773) 440-5054\n",
      "GECEASED DNTERMENT BO\n",
      "DAENITZ, CELLIA E.\n",
      "GATE OF INTERMENT szcnon LOT-TER an aaa VETERAN\n",
      "lay 5, 1990 18 150 7 87 No\n",
      "TRoenTaKEn\n",
      "Hallowel & James F.H.\n",
      "VAULT\n",
      "Universal #29\n",
      "LOT OWNER RELATION\n",
      "EXT OF mime RELATION\n",
      "\n",
      "11030 German Ct.\n",
      "\n",
      "Leonard Skerbinek\n",
      "\n",
      "Son\n",
      "\n",
      "Willow Springs, IL 60480\n",
      "Oactas@o tara aad est eat\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DAFFIN, DORRA O'STAR NA\n",
      "DATE Of INTE MORENT secnos Lol-Tua Qa 93 veTioas\n",
      "10/31/2003 33 54 row-3 | 12 ]3 MOS. [ NO\n",
      "UNDE RTAGER CARTER\n",
      "wAULT No ROX\n",
      "Lol Owner elLAnos\n",
      "alanos\n",
      "\n",
      " \n",
      "\n",
      "SRT OF amt LAVADA DAFFIN (MOM)\n",
      "7411 S. BLACKSTONE\n",
      "nm CHICAGO, IL 60619 (773)752-1247\n",
      "DECEASED\n",
      "\n",
      "DAGDICIAN, ROSE\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF INTERMENT SECTION LOT-TER Gn ae VETERAN\n",
      "Dec. 11, 1989] 19A M178 | 2 88 NO\n",
      "\n",
      "ONDERTARES Smith F.H.\n",
      "\n",
      "vaury Gr. Box 30\n",
      "\n",
      "LOT OWNER Q@ELATION\n",
      "\n",
      "werrorent Richard Dagdigian Son fEATION\n",
      "\n",
      " \n",
      "\n",
      "6219 1]. Forest Glenn\n",
      "\n",
      "Chicago,IL 60646\n",
      "SxTERteeart a0\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DECLAsID\n",
      "DAGGER, KENNIE\n",
      "BATH OF TE Aaa aT ercnon om-TrA aol wivTanan\n",
      "1/24/05 28 | = oe 92 | NO\n",
      "UNDIRTARER\n",
      "GAIDAS FUNERALS\n",
      "VAULT\n",
      "SIMPLEX #29\n",
      "Ol Ownmin aUaAnoa\n",
      "JAMES DAGGER (SON)\n",
      "suronm 10615 S. 83RD AVENUE meuanos\n",
      "\n",
      " \n",
      "\n",
      "PALOS HILLS, IL 60465 (708) 974-0313\n",
      "REMARKS\n",
      "~e e\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "OtcEanES PVE RRRLAET BO\n",
      "DAGGS, ERNEST .\n",
      "OaTE OF imTERSE NT SECTION LOT-TEA w ‘b an aor ew\n",
      "5/22/97 34 242 1 NA NO\n",
      "UNDERTARLA\n",
      "\n",
      "R\n",
      "vauir\n",
      "CRAVE BOX #30\n",
      "LOT OWNER aELATION\n",
      "ACY OF a RLLATION\n",
      "VIVIAN DAGGS WIFE\n",
      "8207 S. LAFLIN, CHGO,, IT 60620\n",
      "REMARES\n",
      "\n",
      "(773)651-6125\n",
      "\n",
      " \n",
      "\n",
      "B/G\n",
      "\n",
      "MARKER\n",
      "DECEASED INTERMENT 0D\n",
      "\n",
      "DAHLIN, ANNA“V.\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF INTERMENT secrON LOT-TER an eat VETERAM\n",
      "August 24, 1989] 27 B= 360 1 3\n",
      "URDERTAKER\n",
      "\n",
      "TIGRE\n",
      "\n",
      "Colonial Bunereal\n",
      "VAULT\n",
      "\n",
      " \n",
      "\n",
      "LOT OWNER RELATION\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "PENT OF Kity RELATION\n",
      "Daughter: Lalliaan Francis\n",
      "19052 Loomis Homewood, I1 60430\n",
      "DECEAZEC DeTERMENT BO\n",
      "\n",
      "DAHLSTROM, ELLEN Gy,\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF INTERMENT BECTION LOT-TER Qn AGE VETERAS\n",
      "Nov. 14, 1987 19 650 76 No\n",
      "SS\n",
      "UNDERTAKER\n",
      "Hare FH New Lisbon, Wis\n",
      "VAULT °\n",
      "30 Monticello\n",
      "LOT OWMER RELATION\n",
      "Same\n",
      "MEXT OF SUN RRATION\n",
      "Judy Ericson (Daughter)\n",
      "\n",
      " \n",
      "\n",
      "Rt. #3, Box 47 - Tomah, Wis, 54660\n",
      "INTERMENT NO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DECEASED\n",
      "Dahnke, Louise Dorothy\n",
      "DATE OF INTERMENT SECTION LOT-TER GA AGE VETERAN\n",
      "Feb. 16, 1985 11 E% 85 84 No\n",
      "UNDERTAKER\n",
      "Edgar F.H.\n",
      "VAULT\n",
      "#9 Grave Box\n",
      "LOT OWNER RELATION\n",
      "Mae C. Dahnke & Lillian M. Heacock\n",
      "MEXT OF KIN RELATION\n",
      "\n",
      "Mrs. Lillian D, Heacock\n",
      "10500 S. Walden Parkway - Chicago, 60643\n",
      "\n",
      " \n",
      "\n",
      "REMARKS\n",
      "rv ansenert 080\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DEctasiDd\n",
      "DAIL, HAZIKIAH\n",
      "OATH OF im TE RAGE WT gicnos .OT-TER R W an aot viTinas\n",
      "12/09/2002 2 240a FQN] 2 74 NO\n",
      "UNDSRTANER GATLING :\n",
      "VAULT GR BK #30\n",
      "Lol ownia egLaTOn\n",
      "muTores!” MANTHA DAIL WIFE “™A™*\n",
      "T2308\" S EMERALD TTS 264-8314\n",
      "\n",
      "CHICAGO TL 60628\n",
      "o@rceasaD\n",
      "DAILEY, ANNIE B.\n",
      "DATE OF TL AMENT Sicrion LOT-TER on wiThnan\n",
      "8/25/04 44 ROW 2] 5\n",
      "UNDERTAKER LEAK\n",
      "vee GR BOX #30\n",
      "Lor ownta RELAnOS\n",
      "ANNA _COLEMAN (DAUGHTER)\n",
      "auanoa\n",
      "\n",
      " \n",
      "\n",
      "BET OF aie\n",
      "\n",
      "8505 S. DANTE\n",
      "CHICAGO, IL. 60619 (773) _ 221-8386\n",
      "\n",
      " \n",
      "\n",
      "“EAL\n",
      "oeceasEeD oulsemg art oO\n",
      "DALLEY, JAE NABA (STILLBORN) |\n",
      "OATH OF inTIAMEAT eiction 1OT-TER Qn aa VETiAa®\n",
      "5/2/2003 33 54 ROW] | 5 ler nen NO\n",
      "UNDIATARER\n",
      "\n",
      "CARTER FUNERAL HOME\n",
      "waut\n",
      "\n",
      "NONE\n",
      "LOT OwnEa ACLATION\n",
      "\n",
      "adLAnOs\n",
      "\n",
      "suroras  BARFEALE DAILEY (FATHER)\n",
      "789 N. 125 W. LOT 5\n",
      "VALPARISO, IN 46385\n",
      "\n",
      "REMARKS\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "‘OTT RAME ATT BI\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DECEASED\n",
      "DAILEY, LEON NA\n",
      "OATE OF ETLAMENT stcTion: eor-rga an act | witTinan\n",
      "7/21/2003 | 18 125 | A 53 NO\n",
      "USRDIATASER\n",
      "LEAK\n",
      "vault\n",
      "GR_BOX #30\n",
      "LOT Owner elLaATog\n",
      "atanons\n",
      "\n",
      " \n",
      "\n",
      "muvormin”  GERALDINE DAILEY (SISTER)\n",
      "10229 S. YALE AVE.\n",
      "CHICAGO, IL 60628 (773)785-1829°\n",
      "\n",
      "REMARKS\n",
      "OeceaseD art gnaeg wT 960\n",
      "\n",
      "DAILEY, LEON JR.\n",
      "\n",
      " \n",
      "\n",
      "DATE OF Te ARENT atcnon LOT-TER ak veTuRans\n",
      "4/4/97 W348 \"5 25 NO\n",
      "\n",
      " \n",
      "\n",
      "UNDERTAKER\n",
      "\n",
      "JOHNSON FUNERAL HOME\n",
      "\n",
      " \n",
      "\n",
      "vat\n",
      "\n",
      "GRAVE BOX #30\n",
      "\n",
      " \n",
      "\n",
      "LOT Gwmen MELA ee\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "RELATION\n",
      "\n",
      "serovem’ VERNA BERRYMAN (MOTHER)\n",
      "\n",
      " \n",
      "\n",
      "1245 W. MORRIS, CHGO., IL 50626\n",
      "\n",
      " \n",
      "\n",
      "memaned 773/381-9442\n",
      "OacuasiD Teg OTTO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DAILEY, LILLI NA\n",
      "pari cr sicnos aor-Taa agi vweTasan\n",
      "10/26/10 \"| 21 [Se ROW1 \"y NA NO\n",
      "UBDERTARER\n",
      "TAYLOR\n",
      "oma\n",
      "GR BOX #30\n",
      "LO) Ownmia ananos\n",
      "PUBLIC GUARDIAN\n",
      "auanos\n",
      "\n",
      "ONT OF nom\n",
      "\n",
      " \n",
      "\n",
      "Marker.\n",
      "DIceaseD\n",
      "DAILEY, MARY NA\n",
      "DATE OF TERRE AT s1cnon 1ol-Tin on aa vitunan\n",
      "1/12/10 [ L 50 | ae | 16 | NO\n",
      "UNDE ATAKeR\n",
      "W. W. HOLT\n",
      "vat\n",
      "BOX\n",
      "LOT OWEER aLaATOS\n",
      "BETTY MAYS (SISTER)\n",
      "oT OF ame ReLATOS\n",
      "\n",
      "15812 HOYNE\n",
      "\n",
      " \n",
      "\n",
      "HARVEY, IL. 60426 (708) 339-9146\n",
      "\n",
      "aimanas\n",
      "OECEABID\n",
      "DAILEY, PAULINE H. |\n",
      "DATE OF weTE masz eST mchos LOT-TER oak veTunan\n",
      "8/5/2002 17 610 Rows} 3 | 52 [ko\n",
      "UMNOERTARAR\n",
      "GATLING\n",
      "VAULT\n",
      "GR BOX #30\n",
      "LOT Ownga anno\n",
      "mexTowam™ JOYCE PITTS (DAUGHTER) aeLanoe\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "REMARES\n",
      "GLENWOOD, IL.\n",
      "\n",
      "7208.2757=-5148\n",
      "\n",
      "60425\n",
      "\n",
      "Warde,\n",
      "OrceasiD\n",
      "DAILEY, ROSA AMANDA\n",
      "DATE OF TERME WT orcnon 10T-TEA aa agi woTtaan\n",
      "5/12/2004 12 j71 ROoW1} 3 | 81 | NO\n",
      "UBDEATARGR\n",
      "CARTER FUNERAL HOME\n",
      "\n",
      "WAULT\n",
      "\n",
      "' GRAVE BOX #30\n",
      "LOT Ownan elanoe\n",
      "MeTorane BETTY JONES (DAUGHTER) auanos\n",
      "\n",
      " \n",
      "\n",
      "7431 S. MERRILL.\n",
      "CHICAGO, IL 60649\n",
      "773/643-4117\n",
      "\n",
      "AEMARKS\n",
      "Burial Permit\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DECEASED TT FRAGA OT\n",
      "DAILEY, SARA #618866\n",
      "DATE OF ITE RAE NT SECTION LOl-TER an agi vitenas\n",
      "11/21/97 31 |2SuB5 5 Ina NO\n",
      "UMDIATAREA\n",
      "GRIFFIN FUNERAL HOME\n",
      "WAMLT\n",
      "GRAVE BOX #30\n",
      "LOT OwnEA atLATION\n",
      "NEXT OF wim THEODORE DAILEY (SON) nnAnoN\n",
      "7334 S. WHIPPLE, CHGO., IL 60629\n",
      "\n",
      " \n",
      "\n",
      "REMAREG\n",
      "\n",
      "773/471-1799\n",
      "\n",
      "manjenw\n",
      "O8ctasiD\n",
      "DAILY, DOLORES M.\n",
      "\n",
      "Ort emnes oT BO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "CATE OF RETELRASL IT stcnonm 10T-TER an nai ViTinas\n",
      "1/2/2001 137 |Nwt663 | 1 73 NO\n",
      "UNDE ATARER\n",
      "BLAKE-LAMB\n",
      "vant MONARCH #30\n",
      "LOl Gwwen ORLA TION\n",
      "@LET OF Ree rRoanom\n",
      "THOMAS DAILY SON\n",
      "3815 WEST 80TH ST\n",
      "REMARES\n",
      "CHEO = TI 60652 (1135 ogo tNeo\n",
      "orceasio\n",
      "DAILY, SCOTT E. ee)\n",
      "DATH Of ETLAMENT escrion LOT-TER acl viTigas\n",
      "10/2/04 aa NE} 210 A | NA\n",
      "UBDERTAGER\n",
      "NONE\n",
      "VAULT\n",
      "MARBLE URN\n",
      "Lolownin etLanog\n",
      "GINGER SANFORD (DAUGHTER)\n",
      "BEET OF aie alana\n",
      "\n",
      " \n",
      "\n",
      "5753 N. RIVER ROAD\n",
      "\n",
      " \n",
      "\n",
      "WATERVILLE, OH 43566 878=\n",
      "DECEASED CNTERMENT 80\n",
      "\n",
      "DAINTY, FREDERICK J.\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF INTERMENT secnon LoT-TER aa ant VETERAN\n",
      "March 24, 198 27 SWk648 79 No\n",
      "SS\n",
      "UNGERTAKER\n",
      "\n",
      "Evergreen Mortuary & Cemetery Tucson, Az.\n",
      "VAULT\n",
      "\n",
      "# 9 Grave Box\n",
      "LOT OWNER RELATION\n",
      "\n",
      "Same\n",
      "EXT OF Kite\n",
      "\n",
      " \n",
      "\n",
      "RELATION\n",
      "Katherine Dainty - (Wife) 602-624-0961\n",
      "\n",
      "BOLBXAEXOKASTAXRAXXEUNSHY XAAXXBERGBX\n",
      "\n",
      "Remarks 2827 N. Fontana Tucson, Az. 85705\n",
      "IMTERMAAT WO\n",
      "\n",
      "pretaseD’ DAINTY, KATHERINE\n",
      "WETERAS\n",
      "\n",
      "No\n",
      "\n",
      "DATE OF INTERMENT sECTION Lot-TEA an aot |\n",
      "Pr, Pea\n",
      "\n",
      "12/22/98 27 6 4An sii\n",
      "\n",
      "sneenia*& ASEWIKOW HOOVERSON\n",
      "vat “MONTECELLO #30\n",
      "\n",
      "LOT GwHEa\n",
      "\n",
      "RELATION\n",
      "\n",
      "MENT OF IM RELATION\n",
      "\n",
      "FRED DAINTY (SON)\n",
      "371 Fourth St,Prairie du Sac, Wisc\n",
      "53578 NA\n",
      "\n",
      "REMARKS\n",
      "TERME NT SO\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "OCceaseD\n",
      "DAIREN, GOLDEN NA\n",
      "OAT OF MeTERAcERT cnc .ot-Tes aa nok vavTanan\n",
      "175172015 | 21 J 13ROW6 | 6 | 66 | NO\n",
      "Guognranan\n",
      "SOUTHWEST\n",
      "var GRBOX#30\n",
      "Lol Owns manos\n",
      "ieixT OF am GWENDOLYN JORDAN  (DAUGHT ERAT\n",
      "120) TUCKAHOE PL.\n",
      "ae MUNSTER, IN. 46321\n",
      "\n",
      " \n",
      "\n",
      "(708)296-0751\n",
      "DAITCH, DAVID I.\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "GATE CF INTERMENT SECTION LOT-TER aa ont VETERAN\n",
      "Oct. 3, 1990 35 524 2 78 Yes\n",
      "UNDERTAKER\n",
      "\n",
      "Zimmerman & Sandeman F.H.\n",
      "VAULT\n",
      "\n",
      "Gr. Box 30\n",
      "LOT OWNER RELATION\n",
      "EXT OF Ki RELATION\n",
      "\n",
      "Donna Daztch Wife\n",
      "\n",
      " \n",
      "\n",
      "9721 S.Ridgeland 2B Oak Lawn,IL 60453\n",
      "\n",
      " \n",
      "\n",
      "Pda.\n",
      "oeceasiD (arama est 060\n",
      "\n",
      "DAKIN, HANNAH LOIS\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE Of WeTE Real wt gicnos .OT-TER on aed WOTERAR\n",
      "11/6/2001 27 lsu 521 | 1 | NA NO\n",
      "\n",
      "USDERTARER\n",
      "\n",
      "KRUEGER\n",
      "VAULT\n",
      "\n",
      "GRAVE BOX #30\n",
      "Lor Owmss sano\n",
      "Sexton gem GLORIA HAYWOOD enanes\n",
      "\n",
      " \n",
      "\n",
      "46592 WwW. (ZLOin OLR.\n",
      "BLUE ISLAND, TIL BOLO6\n",
      "\n",
      "708-385-0827\n",
      "DECEASED (NTERMENT HO\n",
      "\n",
      "DAKO, CAROLINE (CREMAINS)\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "DATE OF ONTEALAENT BECTON LOT-TER aa \"AGE VETERAE\n",
      "Oct. 19,1989 |35 E4515 2 NA NA\n",
      "URDERTARER\n",
      "\n",
      "Roller F.H. Mountain Home, Ark,\n",
      "VAULT\n",
      "\n",
      " \n",
      "\n",
      "LOT OWNER RELATION\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "RELATION\n",
      "sexropne Doris Anglin Executrix\n",
      "\n",
      "501-435-6255\n",
      "\n",
      "REMARED\n",
      "DECLAEID OTT CARRE ETT BC\n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      " \n",
      "\n",
      "————— ¥\n",
      "DALAS Te 60822%\n",
      "GATE OF ITI RMAENT stcnon tor-mia ROW [aa ‘Act VETERAR\n",
      "6/6/2003 109 - 1 6 78 NA\n",
      "UNDEATARER\n",
      "MIDWEST\n",
      "VAULT\n",
      "GR_ BOX #30\n",
      "LOT CwHin RILATOS\n",
      "acLAnoa\n",
      "\n",
      " \n",
      "\n",
      "suvorni™ =~ PREDEZEA WALKER (DAUGHTER)\n",
      "\n",
      "6253 S. MICHIGAN AVE.\n",
      "Ben IL 60637 (773)955-7290\n",
      "\n",
      "REMARKS\n",
      "\n",
      " \n",
      "\n",
      "Thar Lb Ls\n",
      "DECEASED rT a RAE FT FD\n",
      "\n",
      "DAL-DEGAN, VAL NA\n",
      "DATE OF RETRA MENT becrion witthan\n",
      "12/8/2003 (\"33 (e eb zH [eRyer |“ wa | “wo\n",
      "UMDEATARER Zz & SANDEMAN\n",
      "vauT\n",
      "CRYPT\n",
      "\n",
      "LoTownEa ‘RILA TIONS\n",
      "\n",
      " \n",
      "\n",
      "ALLATION\n",
      "\n",
      "ExT OF ain VERNA DAL-DEGAN (WIFE)\n",
      "2516 W. 115TH ST.\n",
      "CHICAGO, IL 60655\n",
      "\n",
      "AEM ARES\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "import cv2\n",
    "import sys\n",
    "import pytesseract\n",
    "from PIL import Image, ImageOps\n",
    "import tempfile\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "IMAGE_SIZE = 1024\n",
    "BINARY_THREHOLD = 180\n",
    "directory = \"E:\\\\Analytics\\\\AIapplications\\\\AIimages\\\\Type-1 blocks\"\n",
    "pytesseract.pytesseract.tesseract_cmd = r\"C:\\\\Program Files\\\\Tesseract-OCR\\\\tesseract.exe\"\n",
    "\n",
    "\n",
    "def process_image_for_ocr(file_path):\n",
    "    \n",
    "    temp_filename = set_image_dpi(file_path)\n",
    "    #im_new=remove_noise_and_smooth(temp_filename)\n",
    "    return (temp_filename)\n",
    "\n",
    "def set_image_dpi(file_path):\n",
    "    im = Image.open(file_path)\n",
    "    im = ImageOps.expand(im, border=16)\n",
    "    length_x, width_y = im.size\n",
    "    factor = min(1, int(IMAGE_SIZE / length_x))\n",
    "    size = factor * length_x, factor * width_y\n",
    "    im_resized = im.resize(size, Image.ANTIALIAS)\n",
    "    temp_file = tempfile.NamedTemporaryFile(delete=False, suffix='.jpg')\n",
    "    temp_filename = temp_file.name\n",
    "    im_resized.save(temp_filename, dpi=(300, 300))\n",
    "    return temp_filename\n",
    "\n",
    "def image_smoothening(img):\n",
    "    ret1, th1 = cv2.threshold(img, BINARY_THREHOLD, 255, cv2.THRESH_BINARY)\n",
    "    ret2, th2 = cv2.threshold(th1, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)\n",
    "    blur = cv2.GaussianBlur(th2, (1, 1), 0)\n",
    "    ret3, th3 = cv2.threshold(blur, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)\n",
    "    return th3\n",
    "\n",
    "def remove_noise_and_smooth(file_name):\n",
    "    img = plt.imread(file_name)\n",
    "    filtered = cv2.adaptiveThreshold(img.astype(np.uint8), 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 41,3)\n",
    "    kernel = np.ones((1, 1), np.uint8)\n",
    "    opening = cv2.morphologyEx(filtered, cv2.MORPH_OPEN, kernel)\n",
    "    closing = cv2.morphologyEx(opening, cv2.MORPH_CLOSE, kernel)\n",
    "    img = image_smoothening(img)\n",
    "    or_image = cv2.bitwise_or(img, closing)\n",
    "    plt.imshow(or_image)\n",
    "    return or_image\n",
    "\n",
    "for file in os.listdir(directory):\n",
    "    filename = os.fsdecode(file)\n",
    "    if filename.endswith(\".jpg\"): \n",
    "        image = os.path.join(directory, filename)\n",
    "        config=()\n",
    "        text = pytesseract.image_to_string(process_image_for_ocr(image))\n",
    "        print(text)\n",
    "    else:\n",
    "        continue\n",
    "     "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'oaceasiD Grt fmt rt BO\\n\\nDABNEY, JERMAINE LEMONT\\n\\nOATH OF TERRE WT\\n\\n \\n\\n10/9/2003\\n\\nUSDSRIAGIA'"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import os\n",
    "import cv2\n",
    "import sys\n",
    "import pytesseract\n",
    "from PIL import Image, ImageOps\n",
    "import tempfile\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "output_dir = 'C:\\\\Users\\\\swamireddy.v\\\\Desktop\\\\outputs'\n",
    "\n",
    "def get_string(img_path):\n",
    "    img = cv2.imread(img_path)\n",
    "\n",
    "    # Extract the file name without the file extension\n",
    "    file_name = os.path.basename(img_path).split('.')[0]\n",
    "    file_name = file_name.split()[0]\n",
    "\n",
    "    # Create a directory for outputs\n",
    "    output_path = os.path.join(output_dir, file_name)\n",
    "    if not os.path.exists(output_path):\n",
    "        os.makedirs(output_path)\n",
    "        \n",
    "    img = cv2.resize(img, None, fx=1.5, fy=1.5, interpolation=cv2.INTER_CUBIC)\n",
    "    # Convert to gray\n",
    "    img = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)\n",
    "\n",
    "    # Apply dilation and erosion to remove some noise\n",
    "    kernel = np.ones((1, 1), np.uint8)\n",
    "    img = cv2.dilate(img, kernel, iterations=1)\n",
    "    img = cv2.erode(img, kernel, iterations=1)\n",
    "    # Apply blur to smooth out the edges\n",
    "    img = cv2.GaussianBlur(img, (5, 5), 0)\n",
    "    # Apply threshold to get image with only b&w (binarization)\n",
    "    img = cv2.threshold(img, 0, 255, cv2.THRESH_BINARY + cv2.THRESH_OTSU)[1]\n",
    "    # Save the filtered image in the output directory\n",
    "    save_path = os.path.join(output_path, file_name + \"_filter_\"  + \".jpg\")\n",
    "    cv2.imwrite(save_path, img)\n",
    "\n",
    "    # Recognize text with tesseract for python\n",
    "    result = pytesseract.image_to_string(img, lang=\"eng\")\n",
    "    return result\n",
    "\n",
    "get_string('E:/Analytics/AIapplications/AIimages/Croppedimages/Type-1 blocks/iCard_021881_1_Dabney_Jermaine_Lemont.jpg')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}