Image_Prediction_Pipeline.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from IPython.display import Image\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import os \n",
    "import cv2\n",
    "from datetime import datetime\n",
    "import keras\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import Counter\n",
    "import itertools\n",
    "from keras import backend as K\n",
    "from keras.layers import Input, Conv2D, MaxPool2D, Dense,MaxPooling2D,Bidirectional\n",
    "from keras.layers import AveragePooling2D, Flatten, Activation\n",
    "from keras.layers import BatchNormalization, Dropout\n",
    "from keras.layers import Concatenate, Add, Multiply, Lambda\n",
    "from keras.layers import UpSampling2D, Reshape\n",
    "from keras.layers import add,concatenate\n",
    "from keras.layers import Reshape\n",
    "from keras.models import Model\n",
    "from keras.layers import LSTM,GRU\n",
    "import tensorflow as tf\n",
    "import warnings\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_data=pd.read_csv('sampled_data_100.csv') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(100, 2)\n"
     ]
    }
   ],
   "source": [
    "test_data.drop(['Unnamed: 0'],axis=1,inplace=True)\n",
    "print(test_data.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "def preprocess(test_img):\n",
    "    test_img_resized=cv2.resize(test_img,(170,32))\n",
    "    test_image=test_img_resized[:,:,1]\n",
    "    test_image=test_image.T \n",
    "    test_image=np.expand_dims(test_image,axis=-1)\n",
    "    test_image=np.expand_dims(test_image, axis=0)\n",
    "    test_image=test_image/255\n",
    "    return test_image"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ImageName</th>\n",
       "      <th>Labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Test_data/1_BASEMAN.jpg</td>\n",
       "      <td>BASEMAN</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Test_data/2_DIRECTS.jpg</td>\n",
       "      <td>DIRECTS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Test_data/3_lidless.jpg</td>\n",
       "      <td>LIDLESS</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Test_data/4_riposte.jpg</td>\n",
       "      <td>RIPOSTE</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Test_data/5_Smudges.jpg</td>\n",
       "      <td>SMUDGES</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 ImageName   Labels\n",
       "0  Test_data/1_BASEMAN.jpg  BASEMAN\n",
       "1  Test_data/2_DIRECTS.jpg  DIRECTS\n",
       "2  Test_data/3_lidless.jpg  LIDLESS\n",
       "3  Test_data/4_riposte.jpg  RIPOSTE\n",
       "4  Test_data/5_Smudges.jpg  SMUDGES"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "letters= '0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def words_from_labels(labels):\n",
    "    txt=[]\n",
    "    for ele in labels:\n",
    "        if ele == len(letters): \n",
    "            txt.append(\"\")\n",
    "        else:\n",
    "            txt.append(letters[ele])\n",
    "    return \"\".join(txt)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "def decode_label(out):\n",
    "    out_best = list(np.argmax(out[0,2:], axis=1))\n",
    "\n",
    "    out_best = [k for k, g in itertools.groupby(out_best)]  \n",
    "\n",
    "    outstr=words_from_labels(out_best)\n",
    "    return outstr"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "img_h=32\n",
    "img_w=170\n",
    "img_c=1\n",
    "num_classes=len(letters)+1\n",
    "batch_size=64\n",
    "max_length=15"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "def model_create(drop_out_rate=0.35):\n",
    "    \n",
    "    if K.image_data_format() == 'channels_first':\n",
    "        input_shape = (1, img_w, img_h)\n",
    "    else:\n",
    "        input_shape = (img_w, img_h, 1)\n",
    "       \n",
    "    model_input=Input(shape=input_shape,name='img_input',dtype='float32')\n",
    "\n",
    "    model = Conv2D(64, (3, 3), padding='same', name='conv1', kernel_initializer='he_normal')(model_input) \n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "    model = MaxPooling2D(pool_size=(2, 2), name='max1')(model) \n",
    "\n",
    "    model = Conv2D(128, (3, 3), padding='same', name='conv2', kernel_initializer='he_normal')(model) \n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "    model = MaxPooling2D(pool_size=(2, 2), name='max2')(model) \n",
    "\n",
    "    model = Conv2D(256, (3, 3), padding='same', name='conv3', kernel_initializer='he_normal')(model) \n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "\n",
    "    model = Conv2D(256, (3, 3), padding='same', name='conv4', kernel_initializer='he_normal')(model)\n",
    "    model=Dropout(drop_out_rate)(model)\n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "    model = MaxPooling2D(pool_size=(1, 2), name='max3')(model)  \n",
    "\n",
    "    model = Conv2D(512, (3, 3), padding='same', name='conv5', kernel_initializer='he_normal')(model) \n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "    \n",
    "    model = Conv2D(512, (3, 3), padding='same', name='conv6')(model)\n",
    "    model=Dropout(drop_out_rate)(model)\n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)\n",
    "    model = MaxPooling2D(pool_size=(1, 2), name='max4')(model) \n",
    "\n",
    "    model = Conv2D(512, (2, 2), padding='same', kernel_initializer='he_normal', name='con7')(model)\n",
    "    model=Dropout(0.25)(model)\n",
    "    model = BatchNormalization()(model)\n",
    "    model = Activation('relu')(model)    \n",
    "\n",
    "    model = Reshape(target_shape=((42, 1024)), name='reshape')(model)  \n",
    "    model = Dense(64, activation='relu', kernel_initializer='he_normal', name='dense1')(model)  \n",
    "\n",
    "    model=Bidirectional(LSTM(256, return_sequences=True, kernel_initializer='he_normal'), merge_mode='sum')(model)\n",
    "    model=Bidirectional(LSTM(256, return_sequences=True, kernel_initializer='he_normal'), merge_mode='concat')(model)\n",
    "\n",
    "    model = Dense(num_classes, kernel_initializer='he_normal',name='dense2')(model) \n",
    "    y_pred = Activation('softmax', name='softmax')(model)\n",
    "\n",
    "    return Model(inputs=[model_input], outputs=y_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_data_single_image_Prediction(model,test_img_path):\n",
    "    start=datetime.now()\n",
    "    test_img=cv2.imread(test_img_path)\n",
    "    test_image=preprocess(test_img)\n",
    "    model_output=model.predict(test_image)\n",
    "    predicted_output=decode_label(model_output)\n",
    "    print(\"Predicted Text in the Image: \", predicted_output)\n",
    "    print(\"Time Taken for Processing: \",datetime.now()-start)    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def test_data_multiple_image_Prediction(model,test_img_names,test_labels,total):\n",
    "    start=datetime.now()\n",
    "    accuracy=0\n",
    "    letter_acc=0\n",
    "    letter_cnt=0\n",
    "    count=0\n",
    "    for i in range(len(test_labels)):\n",
    "        test_img=cv2.imread(test_img_names[i])\n",
    "        test_image=preprocess(test_img)\n",
    "        model_output=model.predict(test_image)\n",
    "        predicted_output=decode_label(model_output)\n",
    "        actual_output=test_labels[i]\n",
    "        count+=1\n",
    "        for j in range(min(len(predicted_output),len(actual_output))):\n",
    "            if predicted_output[j]==actual_output[j]:\n",
    "                letter_acc+=1\n",
    "        letter_cnt+=max(len(predicted_output),len(actual_output))\n",
    "        if actual_output==predicted_output:\n",
    "            accuracy+=1\n",
    "        print(\"-\"*80)\n",
    "        print(\"Actual Text: \",actual_output,\"   Predicted Text: \",predicted_output)\n",
    "    print(\"=\"*80)\n",
    "    print(\"Model Output Accuracy: \",(accuracy/total)*100, \" %\")\n",
    "    print(\"Model Output Letter Accuracy: \",(letter_acc/letter_cnt)*100, \" %\")\n",
    "    print(\"Time Taken for Processing: \",datetime.now()-start)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "model=model_create()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Model: \"model\"\n",
      "_________________________________________________________________\n",
      " Layer (type)                Output Shape              Param #   \n",
      "=================================================================\n",
      " img_input (InputLayer)      [(None, 170, 32, 1)]      0         \n",
      "                                                                 \n",
      " conv1 (Conv2D)              (None, 170, 32, 64)       640       \n",
      "                                                                 \n",
      " batch_normalization (BatchN  (None, 170, 32, 64)      256       \n",
      " ormalization)                                                   \n",
      "                                                                 \n",
      " activation (Activation)     (None, 170, 32, 64)       0         \n",
      "                                                                 \n",
      " max1 (MaxPooling2D)         (None, 85, 16, 64)        0         \n",
      "                                                                 \n",
      " conv2 (Conv2D)              (None, 85, 16, 128)       73856     \n",
      "                                                                 \n",
      " batch_normalization_1 (Batc  (None, 85, 16, 128)      512       \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_1 (Activation)   (None, 85, 16, 128)       0         \n",
      "                                                                 \n",
      " max2 (MaxPooling2D)         (None, 42, 8, 128)        0         \n",
      "                                                                 \n",
      " conv3 (Conv2D)              (None, 42, 8, 256)        295168    \n",
      "                                                                 \n",
      " batch_normalization_2 (Batc  (None, 42, 8, 256)       1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_2 (Activation)   (None, 42, 8, 256)        0         \n",
      "                                                                 \n",
      " conv4 (Conv2D)              (None, 42, 8, 256)        590080    \n",
      "                                                                 \n",
      " dropout (Dropout)           (None, 42, 8, 256)        0         \n",
      "                                                                 \n",
      " batch_normalization_3 (Batc  (None, 42, 8, 256)       1024      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_3 (Activation)   (None, 42, 8, 256)        0         \n",
      "                                                                 \n",
      " max3 (MaxPooling2D)         (None, 42, 4, 256)        0         \n",
      "                                                                 \n",
      " conv5 (Conv2D)              (None, 42, 4, 512)        1180160   \n",
      "                                                                 \n",
      " batch_normalization_4 (Batc  (None, 42, 4, 512)       2048      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_4 (Activation)   (None, 42, 4, 512)        0         \n",
      "                                                                 \n",
      " conv6 (Conv2D)              (None, 42, 4, 512)        2359808   \n",
      "                                                                 \n",
      " dropout_1 (Dropout)         (None, 42, 4, 512)        0         \n",
      "                                                                 \n",
      " batch_normalization_5 (Batc  (None, 42, 4, 512)       2048      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_5 (Activation)   (None, 42, 4, 512)        0         \n",
      "                                                                 \n",
      " max4 (MaxPooling2D)         (None, 42, 2, 512)        0         \n",
      "                                                                 \n",
      " con7 (Conv2D)               (None, 42, 2, 512)        1049088   \n",
      "                                                                 \n",
      " dropout_2 (Dropout)         (None, 42, 2, 512)        0         \n",
      "                                                                 \n",
      " batch_normalization_6 (Batc  (None, 42, 2, 512)       2048      \n",
      " hNormalization)                                                 \n",
      "                                                                 \n",
      " activation_6 (Activation)   (None, 42, 2, 512)        0         \n",
      "                                                                 \n",
      " reshape (Reshape)           (None, 42, 1024)          0         \n",
      "                                                                 \n",
      " dense1 (Dense)              (None, 42, 64)            65600     \n",
      "                                                                 \n",
      " bidirectional (Bidirectiona  (None, 42, 256)          657408    \n",
      " l)                                                              \n",
      "                                                                 \n",
      " bidirectional_1 (Bidirectio  (None, 42, 512)          1050624   \n",
      " nal)                                                            \n",
      "                                                                 \n",
      " dense2 (Dense)              (None, 42, 37)            18981     \n",
      "                                                                 \n",
      " softmax (Activation)        (None, 42, 37)            0         \n",
      "                                                                 \n",
      "=================================================================\n",
      "Total params: 7,350,373\n",
      "Trainable params: 7,345,893\n",
      "Non-trainable params: 4,480\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "model.load_weights('Best_Img_recog_LSTM_Adam_model_run_weights.h5')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(xyz)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_image_0='Test_data/3_lidless.jpg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(test_image_0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 67ms/step\n",
      "Predicted Text in the Image:  LIDLESS\n",
      "Time Taken for Processing:  0:00:00.110602\n"
     ]
    }
   ],
   "source": [
    "test_data_single_image_Prediction(model,test_image_0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_image_1='Test_Data/84_COMPOSEDLY.jpg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(test_image_1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 84ms/step\n",
      "Predicted Text in the Image:  COMPOSEOLY\n",
      "Time Taken for Processing:  0:00:00.154076\n"
     ]
    }
   ],
   "source": [
    "test_data_single_image_Prediction(model,test_image_1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_image_2='Test_data/99_EVANGELISTS.jpg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(test_image_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 73ms/step\n",
      "Predicted Text in the Image:  ERANCELETS\n",
      "Time Taken for Processing:  0:00:00.140825\n"
     ]
    }
   ],
   "source": [
    "test_data_single_image_Prediction(model,test_image_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "demoimg='demoaimages/36.jpg'\n",
    "Image(demoimg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_data_single_image_Prediction(model,test_image_2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_img_paths=test_data['ImageName'].values\n",
    "test_img_labels=test_data['Labels'].values\n",
    "total=100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 66ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  WHAT    Predicted Text:  WHAT\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  MEMOREX    Predicted Text:  MEMOREX\n",
      "1/1 [==============================] - 0s 115ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  COMPACT    Predicted Text:  COMPACT\n",
      "1/1 [==============================] - 0s 114ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PROFESSIONAL    Predicted Text:  PROFESSIONAL\n",
      "1/1 [==============================] - 0s 113ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  RECORDABLE    Predicted Text:  RECORDABLE\n",
      "1/1 [==============================] - 0s 98ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  EUROPE    Predicted Text:  EUROPE\n",
      "1/1 [==============================] - 0s 88ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PRODUCTS    Predicted Text:  PRODUCTS\n",
      "1/1 [==============================] - 0s 96ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  COMPATIBLE    Predicted Text:  COMPATIBLE\n",
      "1/1 [==============================] - 0s 99ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  IMPORTED    Predicted Text:  IMPORTED\n",
      "1/1 [==============================] - 0s 116ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  5    Predicted Text:  S\n",
      "1/1 [==============================] - 0s 82ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  TESCO    Predicted Text:  TESCO\n",
      "1/1 [==============================] - 0s 86ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  VALUE    Predicted Text:  VALUE\n",
      "1/1 [==============================] - 0s 100ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  WASHING    Predicted Text:  WASHING\n",
      "1/1 [==============================] - 0s 95ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  UP    Predicted Text:  ASP\n",
      "1/1 [==============================] - 0s 86ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  LIQUID    Predicted Text:  LIQUID\n",
      "1/1 [==============================] - 0s 111ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  SAFETY    Predicted Text:  SAFETY\n",
      "1/1 [==============================] - 0s 100ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  NOTICE    Predicted Text:  MOEE\n",
      "1/1 [==============================] - 0s 125ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  STOP    Predicted Text:  STOP\n",
      "1/1 [==============================] - 0s 118ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PEPSI    Predicted Text:  FACEA\n",
      "1/1 [==============================] - 0s 108ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  SMOKIN    Predicted Text:  SMOKIN\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  DISC    Predicted Text:  DISC\n",
      "1/1 [==============================] - 0s 109ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PRODUCTS    Predicted Text:  PRODUCTS\n",
      "1/1 [==============================] - 0s 59ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  EUROPE    Predicted Text:  EUROPE\n",
      "1/1 [==============================] - 0s 72ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  BECOME    Predicted Text:  BECOME\n",
      "1/1 [==============================] - 0s 46ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  JACOBSON    Predicted Text:  JACOBSON\n",
      "1/1 [==============================] - 0s 52ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  RECORDS    Predicted Text:  RECORDS\n",
      "================================================================================\n",
      "Model Output Accuracy:  84.61538461538461  %\n",
      "Model Output Letter Accuracy:  91.51515151515152  %\n",
      "Time Taken for Processing:  0:00:04.134481\n"
     ]
    }
   ],
   "source": [
    "test_data_multiple_image_Prediction(model,test_img_paths,test_img_labels,total)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [],
   "source": [
    "testimage='D:\\\\Final-yr-Proj\\\\CODE\\\\word\\\\word\\\\1\\\\73.jpg'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/jpeg": "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",
      "text/plain": [
       "<IPython.core.display.Image object>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Image(testimage)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 56ms/step\n",
      "Predicted Text in the Image:  WASHING\n",
      "Time Taken for Processing:  0:00:00.098033\n"
     ]
    }
   ],
   "source": [
    "test_data_single_image_Prediction(model,testimage)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(xyz)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_data2=pd.read_csv('demo_data.csv') "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(26, 2)\n"
     ]
    }
   ],
   "source": [
    "test_data2.drop(['Unnamed: 0'],axis=1,inplace=True)\n",
    "print(test_data2.shape)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>ImageName</th>\n",
       "      <th>Labels</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>demoaimages/img1.jpg</td>\n",
       "      <td>WHAT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>demoaimages/33.jpg</td>\n",
       "      <td>MEMOREX</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>demoaimages/34.jpg</td>\n",
       "      <td>COMPACT</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>demoaimages/38.jpg</td>\n",
       "      <td>PROFESSIONAL</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>demoaimages/36.jpg</td>\n",
       "      <td>RECORDABLE</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              ImageName        Labels\n",
       "0  demoaimages/img1.jpg          WHAT\n",
       "1    demoaimages/33.jpg       MEMOREX\n",
       "2    demoaimages/34.jpg       COMPACT\n",
       "3    demoaimages/38.jpg  PROFESSIONAL\n",
       "4    demoaimages/36.jpg    RECORDABLE"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test_data2.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [],
   "source": [
    "test_img_paths=test_data2['ImageName'].values\n",
    "test_img_labels=test_data2['Labels'].values\n",
    "total=26"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1/1 [==============================] - 0s 100ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  WHAT    Predicted Text:  WHAT\n",
      "1/1 [==============================] - 0s 79ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  MEMOREX    Predicted Text:  MEMOREX\n",
      "1/1 [==============================] - 0s 84ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  COMPACT    Predicted Text:  COMPACT\n",
      "1/1 [==============================] - 0s 57ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PROFESSIONAL    Predicted Text:  PROFESSIONAL\n",
      "1/1 [==============================] - 0s 65ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  RECORDABLE    Predicted Text:  RECORDABLE\n",
      "1/1 [==============================] - 0s 50ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  EUROPE    Predicted Text:  EUROPE\n",
      "1/1 [==============================] - 0s 53ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PRODUCTS    Predicted Text:  PRODUCTS\n",
      "1/1 [==============================] - 0s 47ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  COMPATIBLE    Predicted Text:  COMPATIBLE\n",
      "1/1 [==============================] - 0s 104ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  IMPORTED    Predicted Text:  IMPORTED\n",
      "1/1 [==============================] - 0s 67ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  5    Predicted Text:  S\n",
      "1/1 [==============================] - 0s 76ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  TESCO    Predicted Text:  TESCO\n",
      "1/1 [==============================] - 0s 60ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  VALUE    Predicted Text:  VALUE\n",
      "1/1 [==============================] - 0s 116ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  WASHING    Predicted Text:  WASHING\n",
      "1/1 [==============================] - 0s 86ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  UP    Predicted Text:  ASP\n",
      "1/1 [==============================] - 0s 67ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  LIQUID    Predicted Text:  LIQUID\n",
      "1/1 [==============================] - 0s 62ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  SAFETY    Predicted Text:  SAFETY\n",
      "1/1 [==============================] - 0s 58ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  NOTICE    Predicted Text:  MOEE\n",
      "1/1 [==============================] - 0s 68ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  STOP    Predicted Text:  STOP\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PEPSI    Predicted Text:  FACEA\n",
      "1/1 [==============================] - 0s 69ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  SMOKIN    Predicted Text:  SMOKIN\n",
      "1/1 [==============================] - 0s 70ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  DISC    Predicted Text:  DISC\n",
      "1/1 [==============================] - 0s 91ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  PRODUCTS    Predicted Text:  PRODUCTS\n",
      "1/1 [==============================] - 0s 89ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  EUROPE    Predicted Text:  EUROPE\n",
      "1/1 [==============================] - 0s 99ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  BECOME    Predicted Text:  BECOME\n",
      "1/1 [==============================] - 0s 70ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  JACOBSON    Predicted Text:  JACOBSON\n",
      "1/1 [==============================] - 0s 51ms/step\n",
      "--------------------------------------------------------------------------------\n",
      "Actual Text:  RECORDS    Predicted Text:  RECORDS\n",
      "================================================================================\n",
      "Model Output Accuracy:  84.61538461538461  %\n",
      "Model Output Letter Accuracy:  91.51515151515152  %\n",
      "Time Taken for Processing:  0:00:03.794056\n"
     ]
    }
   ],
   "source": [
    "test_data_multiple_image_Prediction(model,test_img_paths,test_img_labels,total)"
   ]
  }
 ],
 "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": "3.10.4"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}