Telugu_script_alphabets.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "uUjp5yraMXUP"
   },
   "source": [
    "## Telugu Script - Alphabets Recognition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "c9eabH_ivbjd"
   },
   "source": [
    "https://en.wikipedia.org/wiki/Telugu_script"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "DGNXXdMmMndh"
   },
   "source": [
    "Telugu script, an abugida from the Brahmic family of scripts, is used to write the Telugu language, a Dravidian language spoken in the Indian states of Andhra Pradesh and Telangana as well as several other neighbouring states."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "YaKAKU7yBWid",
    "outputId": "f09c8cdb-d2e6-4e80-d3c8-133b45144e3e"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount(\"/content/drive\", force_remount=True).\n"
     ]
    }
   ],
   "source": [
    "from google.colab import drive\n",
    "drive.mount('/content/drive')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "vAqnShVRAb8q",
    "outputId": "f01201c2-79ec-4417-e578-72a1929d089b"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "Using TensorFlow backend.\n"
     ]
    }
   ],
   "source": [
    "# Importing all necessary libraries \n",
    "from sklearn.model_selection import train_test_split\n",
    "from keras.models import Sequential,Model\n",
    "from keras.utils.np_utils import to_categorical\n",
    "from keras.preprocessing.image import ImageDataGenerator\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "from keras.models import Sequential\n",
    "from keras.layers import Dense, Dropout, Flatten,BatchNormalization,Input\n",
    "from keras.layers import Conv2D, MaxPooling2D\n",
    "from keras import backend as K\n",
    "from keras.layers import Activation, Dense\n",
    "from keras import optimizers\n",
    "from keras.layers import Flatten\n",
    "import tensorflow as tf\n",
    "import glob\n",
    "import os\n",
    "from PIL import Image\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "%matplotlib inline\n",
    "from numpy import expand_dims"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "cGqW47m9Ab85"
   },
   "outputs": [],
   "source": [
    "# get images from drive\n",
    "images = glob.glob(\"/content/drive/My Drive/DLCP/telugu_script/vowels\"+\"/*.png\") + glob.glob(\"/content/drive/My Drive/DLCP/telugu_script/consonants\"+\"/*.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "_MJWIr7NAb9A"
   },
   "outputs": [],
   "source": [
    "# sort images \n",
    "images.sort()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 935
    },
    "colab_type": "code",
    "id": "cXssBiwLAb9E",
    "outputId": "5c358c02-4f5b-4ac0-8c3d-5e6d215b5e2f"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['/content/drive/My Drive/DLCP/telugu_script/consonants/ala.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/anah.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ba.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/bha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/cha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/chha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/da.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/dda.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ddha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/dha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ga.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/gha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ingya.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ja.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/jha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ka.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/kha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/la.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ma.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/na.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ngya.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/pa.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/pha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ra.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/raw.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/sa.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/sha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/shha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ta.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/tha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/thha.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/tta.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/va.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/consonants/ya.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/a.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/aa.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ah.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ai.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/alu.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/aluu.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/am.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/an.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/au.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/e.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ee.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/i.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ii.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/o.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/oo.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ru.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/ruu.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/u.png',\n",
       " '/content/drive/My Drive/DLCP/telugu_script/vowels/uu.png']"
      ]
     },
     "execution_count": 5,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "images"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "dY-Ot4gcLIbL",
    "outputId": "3f351018-c433-4a08-989f-04279543b504"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "54"
      ]
     },
     "execution_count": 6,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(images)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "-keGeXEvAb9I"
   },
   "outputs": [],
   "source": [
    "# convert images to grayscale and resize to 28X28\n",
    "images_array = []\n",
    "for img in images:\n",
    "    img = Image.open(img)\n",
    "    # img = img.convert('RGB')\n",
    "    img = img.convert('L')\n",
    "    img = img.resize((28,28), Image.ANTIALIAS)\n",
    "    images_array.append(np.array(img))\n",
    "images_array = np.array(images_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 118
    },
    "colab_type": "code",
    "id": "-DnF0GlQAb9V",
    "outputId": "1bd27669-896e-4db8-d8a6-4b868df4dda3"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAC+kAAABRCAYAAAAUjwa/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nEEIIIURA/D/2yex2X5kwYwAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 3888x72 with 54 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot images\n",
    "plt.figure(figsize=(54, 1))\n",
    "for i in range(54):\n",
    "    plt.subplot(1,54,i+1)\n",
    "    plt.imshow(images_array[i],cmap=\"gray\")\n",
    "    plt.axis(\"off\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "4Y51ecoYAb9c"
   },
   "outputs": [],
   "source": [
    "# create a large dataset with duplicates \n",
    "labels = []\n",
    "final_array = []\n",
    "labels_array = []\n",
    "for i,img in enumerate(images_array):\n",
    "    for j in range(10000):\n",
    "        final_array.append(img)\n",
    "        labels_array.append(os.path.basename(images[i])[:-4])\n",
    "final_array = np.array(final_array)\n",
    "labels_array = np.array(labels_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "EJaiN2RH0bPG"
   },
   "outputs": [],
   "source": [
    "# Label Encoding\n",
    "le = LabelEncoder()\n",
    "le.fit(labels_array)\n",
    "\n",
    "def encode(le, labels):\n",
    "    enc = le.transform(labels)\n",
    "    return tf.keras.utils.to_categorical(enc, num_classes=54)\n",
    "def decode(le, one_hot):\n",
    "    dec = np.argmax(one_hot,axis=1)\n",
    "    return le.inverse_transform(dec)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "XPUhkL4euJ6v"
   },
   "outputs": [],
   "source": [
    "labels_array_en = encode(le, labels_array)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "jZhY6oJvAb9t"
   },
   "outputs": [],
   "source": [
    "# split the dataset into train and test\n",
    "X_train,X_test,Y_train,Y_test = train_test_split(final_array,labels_array_en,shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "R5icTKlJAb9x"
   },
   "outputs": [],
   "source": [
    "# reshape to be [samples][width][height][channels]\n",
    "X_train = X_train.reshape((X_train.shape[0], 28, 28, 1)).astype('float32')\n",
    "X_test = X_test.reshape((X_test.shape[0], 28, 28, 1)).astype('float32')\n",
    "# X_train = trainX.reshape(-1,28,28,1)\n",
    "# X_test = testX.reshape(-1,28,28,1)\n",
    "\n",
    "# normalize inputs from 0-255 to 0-1\n",
    "X_train = X_train / 255\n",
    "X_test = X_test / 255"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 34
    },
    "colab_type": "code",
    "id": "aijVAzC8UmJg",
    "outputId": "d72fea27-1aba-415e-b8fa-4fcdb51c4928"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(405000, 28, 28, 1) (135000, 28, 28, 1) (405000, 54) (135000, 54)\n"
     ]
    }
   ],
   "source": [
    "print(X_train.shape, X_test.shape, Y_train.shape, Y_test.shape)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "gugWXryEck8V"
   },
   "source": [
    "## Image agumentation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 71
    },
    "colab_type": "code",
    "id": "Ne_x6jq8aFCi",
    "outputId": "cdde062e-e9de-4f9a-b587-d2e0eef3aafc"
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/usr/local/lib/python3.6/dist-packages/keras_preprocessing/image/image_data_generator.py:341: UserWarning: This ImageDataGenerator specifies `zca_whitening` which overrides setting of`featurewise_std_normalization`.\n",
      "  warnings.warn('This ImageDataGenerator specifies '\n"
     ]
    }
   ],
   "source": [
    "train_image_gen = ImageDataGenerator(\n",
    "    featurewise_center=True,\n",
    "    featurewise_std_normalization=True,\n",
    "    zca_whitening=True,\n",
    "    rotation_range=90,\n",
    "    horizontal_flip=True, \n",
    "    vertical_flip=True,\n",
    "    # rescale=1./255,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2)\n",
    "test_image_gen = ImageDataGenerator(\n",
    "    featurewise_center=True,\n",
    "    featurewise_std_normalization=True,\n",
    "    zca_whitening=True,\n",
    "    rotation_range=90,\n",
    "    horizontal_flip=True, \n",
    "    vertical_flip=True,\n",
    "    # rescale=1./255,\n",
    "    width_shift_range=0.2,\n",
    "    height_shift_range=0.2)\n",
    "\n",
    "train_image_gen.fit(X_train, augment=True)\n",
    "test_image_gen.fit(X_test, augment=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 269
    },
    "colab_type": "code",
    "id": "tGq4CYq0Ab-F",
    "outputId": "602bb463-9445-44a8-9a93-40bbb8c77bcc"
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAVEAAAD8CAYAAADOg5fGAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 9 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot images\n",
    "for X_batch, y_batch in train_image_gen.flow(X_train, Y_train, batch_size=9):\n",
    "    # create a grid of 3x3 images\n",
    "    for i in range(0, 9):\n",
    "        plt.subplot(330 + 1 + i)\n",
    "        plt.imshow(X_batch[i].reshape(28, 28), cmap=plt.get_cmap('gray'))\n",
    "    # show the plot\n",
    "    plt.show()\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0aSxstldAb-K"
   },
   "outputs": [],
   "source": [
    "# lets set epochs and batch_size values\n",
    "epochs = 10\n",
    "batch_size = 128"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "colab_type": "text",
    "id": "tsaKyciRbTcV"
   },
   "source": [
    "## CNN"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 632
    },
    "colab_type": "code",
    "id": "6HhJwC0_nz-x",
    "outputId": "e87e2905-203d-494d-b771-f19585ee289f"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "WARNING:tensorflow:From /usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/init_ops.py:1251: calling VarianceScaling.__init__ (from tensorflow.python.ops.init_ops) with dtype is deprecated and will be removed in a future version.\n",
      "Instructions for updating:\n",
      "Call initializer instance with the dtype argument instead of passing it to the constructor\n",
      "Model: \"sequential\"\n",
      "_________________________________________________________________\n",
      "Layer (type)                 Output Shape              Param #   \n",
      "=================================================================\n",
      "conv2d (Conv2D)              (None, 28, 28, 32)        832       \n",
      "_________________________________________________________________\n",
      "conv2d_1 (Conv2D)            (None, 28, 28, 32)        25632     \n",
      "_________________________________________________________________\n",
      "max_pooling2d (MaxPooling2D) (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "dropout (Dropout)            (None, 14, 14, 32)        0         \n",
      "_________________________________________________________________\n",
      "conv2d_2 (Conv2D)            (None, 14, 14, 64)        18496     \n",
      "_________________________________________________________________\n",
      "conv2d_3 (Conv2D)            (None, 14, 14, 64)        36928     \n",
      "_________________________________________________________________\n",
      "max_pooling2d_1 (MaxPooling2 (None, 7, 7, 64)          0         \n",
      "_________________________________________________________________\n",
      "dropout_1 (Dropout)          (None, 7, 7, 64)          0         \n",
      "_________________________________________________________________\n",
      "flatten (Flatten)            (None, 3136)              0         \n",
      "_________________________________________________________________\n",
      "dense (Dense)                (None, 256)               803072    \n",
      "_________________________________________________________________\n",
      "dropout_2 (Dropout)          (None, 256)               0         \n",
      "_________________________________________________________________\n",
      "dense_1 (Dense)              (None, 54)                13878     \n",
      "=================================================================\n",
      "Total params: 898,838\n",
      "Trainable params: 898,838\n",
      "Non-trainable params: 0\n",
      "_________________________________________________________________\n"
     ]
    }
   ],
   "source": [
    "model = tf.keras.models.Sequential()\n",
    "\n",
    "model.add(tf.keras.layers.Conv2D(filters = 32, kernel_size = (5,5),padding = 'Same', \n",
    "                 activation ='relu', input_shape = (28,28,1)))\n",
    "model.add(tf.keras.layers.Conv2D(filters = 32, kernel_size = (5,5),padding = 'Same', \n",
    "                 activation ='relu'))\n",
    "model.add(tf.keras.layers.MaxPool2D(pool_size=(2,2)))\n",
    "model.add(tf.keras.layers.Dropout(0.25))\n",
    "\n",
    "\n",
    "model.add(tf.keras.layers.Conv2D(filters = 64, kernel_size = (3,3),padding = 'Same', \n",
    "                 activation ='relu'))\n",
    "model.add(tf.keras.layers.Conv2D(filters = 64, kernel_size = (3,3),padding = 'Same', \n",
    "                 activation ='relu'))\n",
    "model.add(tf.keras.layers.MaxPool2D(pool_size=(2,2), strides=(2,2)))\n",
    "model.add(tf.keras.layers.Dropout(0.25))\n",
    "\n",
    "\n",
    "model.add(tf.keras.layers.Flatten())\n",
    "model.add(tf.keras.layers.Dense(256, activation = \"relu\"))\n",
    "model.add(tf.keras.layers.Dropout(0.5))\n",
    "model.add(tf.keras.layers.Dense(54, activation = \"softmax\"))\n",
    "model.summary()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "ap2roeumAb-Q"
   },
   "outputs": [],
   "source": [
    "# Build CNN\n",
    "# Define the optimizer\n",
    "optimizer = tf.keras.optimizers.RMSprop(lr=0.001, rho=0.9, epsilon=1e-08, decay=0.0)\n",
    "model.compile(loss='categorical_crossentropy', optimizer=optimizer, metrics=['accuracy'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 374
    },
    "colab_type": "code",
    "id": "K8ZHj52DLkzZ",
    "outputId": "499374bf-9656-4f84-f1cf-4b97957928c7"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Epoch 1/30\n",
      "3164/3164 [============================>.] - ETA: 0s - loss: 1.1503 - acc: 0.6377\n",
      "Epoch 00001: val_acc improved from -inf to 0.96660, saving model to /content/drive/My Drive/DLCP/telugu_script/talphabets_model.h5\n",
      "3165/3164 [==============================] - 442s 140ms/step - loss: 1.1501 - acc: 0.6378 - val_loss: 0.1000 - val_acc: 0.9666\n",
      "Epoch 2/30\n",
      "3164/3164 [============================>.] - ETA: 0s - loss: 0.2088 - acc: 0.9324\n",
      "Epoch 00002: val_acc improved from 0.96660 to 0.99848, saving model to /content/drive/My Drive/DLCP/telugu_script/talphabets_model.h5\n",
      "3165/3164 [==============================] - 442s 140ms/step - loss: 0.2087 - acc: 0.9324 - val_loss: 0.0074 - val_acc: 0.9985\n",
      "Epoch 3/30\n",
      "3164/3164 [============================>.] - ETA: 0s - loss: 0.1289 - acc: 0.9623\n",
      "Epoch 00003: val_acc did not improve from 0.99848\n",
      "3165/3164 [==============================] - 439s 139ms/step - loss: 0.1289 - acc: 0.9623 - val_loss: 0.0064 - val_acc: 0.9983\n",
      "Epoch 4/30\n",
      "3164/3164 [============================>.] - ETA: 0s - loss: 0.1143 - acc: 0.9692\n",
      "Epoch 00004: val_acc improved from 0.99848 to 0.99915, saving model to /content/drive/My Drive/DLCP/telugu_script/talphabets_model.h5\n",
      "3165/3164 [==============================] - 445s 141ms/step - loss: 0.1143 - acc: 0.9692 - val_loss: 0.0032 - val_acc: 0.9991\n",
      "Epoch 5/30\n",
      "3164/3164 [============================>.] - ETA: 0s - loss: 0.1210 - acc: 0.9710\n",
      "Epoch 00005: val_acc did not improve from 0.99915\n",
      "3165/3164 [==============================] - 442s 140ms/step - loss: 0.1209 - acc: 0.9710 - val_loss: 0.0048 - val_acc: 0.9987\n",
      "Epoch 00005: early stopping\n"
     ]
    }
   ],
   "source": [
    "checkpint = tf.keras.callbacks.ModelCheckpoint(filepath=\"/content/drive/My Drive/DLCP/telugu_script/talphabets_model.h5\",save_best_only=True,verbose=1,monitor=\"val_acc\",mode=\"max\")\n",
    "stoperly = tf.keras.callbacks.EarlyStopping(monitor=\"val_loss\",mode=\"auto\",verbose=1)\n",
    "callbaklist = [checkpint,stoperly]\n",
    "# model_history = model.fit(X_train_m3,trainy,validation_data=(X_test_m3,testy),epochs=10,batch_size=128,callbacks=callbaklist)\n",
    "model_history = model.fit_generator(train_image_gen.flow(X_train, Y_train, batch_size=batch_size),validation_data=test_image_gen.flow(X_test, Y_test, batch_size=batch_size),\n",
    "                    steps_per_epoch=len(X_train) / batch_size,validation_steps=len(X_test)/batch_size, epochs = 30,verbose = 1,callbacks=callbaklist)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 503
    },
    "colab_type": "code",
    "id": "wZJZDFWk0YKo",
    "outputId": "6d7f022d-3a01-493f-a510-711b6ef3bfc4"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fcdb0c67cf8>"
      ]
     },
     "execution_count": 22,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAecAAAHVCAYAAADLvzPyAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nIUM4AwAQMv8fb7qlLK89mIcAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot train and test loss\n",
    "fig, ax = plt.subplots(figsize=(8, 8))\n",
    "ax.plot(model_history.history[\"loss\"])\n",
    "ax.legend(\"loss\")\n",
    "ax.plot(model_history.history[\"val_loss\"])\n",
    "ax.legend(\"val_loss\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 503
    },
    "colab_type": "code",
    "id": "1MsmpdES0eC-",
    "outputId": "d97d9504-cae4-4d7e-fbc0-849cc59c38c6"
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.legend.Legend at 0x7fcdb0b01d68>"
      ]
     },
     "execution_count": 23,
     "metadata": {
      "tags": []
     },
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAe0AAAHVCAYAAADcnaM7AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 576x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "tags": []
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot train and test accuracy\n",
    "fig, ax = plt.subplots(figsize=(8, 8))\n",
    "ax.plot(model_history.history[\"acc\"])\n",
    "ax.legend(\"acc\")\n",
    "ax.plot(model_history.history[\"val_acc\"])\n",
    "ax.legend(\"val_acc\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 51
    },
    "colab_type": "code",
    "id": "zwM4CGFhc9NX",
    "outputId": "da20f714-a267-4e11-d999-015c27f7b6c6"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Test loss: 0.004903744342134823\n",
      "Test accuracy: 0.9986445\n"
     ]
    }
   ],
   "source": [
    "# predict test accuracy\n",
    "scores = model.evaluate(test_image_gen.flow(X_test, Y_test, batch_size=batch_size), verbose=0)\n",
    "print('Test loss:', scores[0])\n",
    "print('Test accuracy:', scores[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "0NIggEPPYnxJ"
   },
   "outputs": [],
   "source": [
    "# make a prediction using test-time augmentation\n",
    "def tta_prediction(datagen, model, image, n_examples):\n",
    "\t# convert image into dataset\n",
    "\tsamples = expand_dims(image, 0)\n",
    "\t# prepare iterator\n",
    "\tit = datagen.flow(samples, batch_size=n_examples)\n",
    "\t# make predictions for each augmented image\n",
    "\tyhats = model.predict_generator(it, steps=n_examples, verbose=0)\n",
    "\t# sum across predictions\n",
    "\tsummed = np.sum(yhats, axis=0)\n",
    "\t# argmax across classes\n",
    "\treturn summed"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "colab": {
     "base_uri": "https://localhost:8080/",
     "height": 357
    },
    "colab_type": "code",
    "id": "KjKDlgRCdGio",
    "outputId": "119565bf-7288-4ccc-944b-becbe6898445"
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "truth: ['gha']\n",
      "prediction: ['gha']\n",
      "truth: ['au']\n",
      "prediction: ['au']\n",
      "truth: ['da']\n",
      "prediction: ['da']\n",
      "truth: ['o']\n",
      "prediction: ['o']\n",
      "truth: ['e']\n",
      "prediction: ['e']\n",
      "truth: ['am']\n",
      "prediction: ['am']\n",
      "truth: ['aa']\n",
      "prediction: ['aa']\n",
      "truth: ['aa']\n",
      "prediction: ['aa']\n",
      "truth: ['uu']\n",
      "prediction: ['uu']\n",
      "truth: ['uu']\n",
      "prediction: ['uu']\n"
     ]
    }
   ],
   "source": [
    "# plot and print model prediciton \n",
    "#plt.figure(figsize=(10, 1))\n",
    "for index in np.random.choice(25000,10):\n",
    "    img = X_test[index]\n",
    "    # plt.subplot(1,10,index+1)\n",
    "    # plt.imshow(img,cmap=\"gray\")\n",
    "    # plt.axis(\"off\")\n",
    "    print(\"truth:\",decode(le,[Y_test[index]]))\n",
    "    pred = tta_prediction(test_image_gen,model,img,7)\n",
    "    print(\"prediction:\",decode(le,[pred]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "vdJD1O6uKXWO"
   },
   "outputs": [],
   "source": [
    "# from keras.models import load_model\n",
    "# model = load_model('model.h5')\n",
    "# model.load_weights('/content/drive/My Drive/DLCP/model.h5')\n",
    "# pred = model.predict(X_test_m3[467].reshape(1,28,28,1))\n",
    "\n",
    "# plt.figure(figsize=(10, 1))\n",
    "# for index in np.random.choice(25000,10):\n",
    "#     img = Image.open('data/' + str(index) + '.png').convert(\"L\")\n",
    "#     img = img.resize((28,28))\n",
    "#     im2arr = np.array(img)\n",
    "#     im2arr = X_test_m3[index]\n",
    "#     im2arr = im2arr.reshape(1,28,28,1)\n",
    "#     Predicting the Test set results\n",
    "#     y_pred = model.predict(im2arr,batch_size=128)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 0,
   "metadata": {
    "colab": {},
    "colab_type": "code",
    "id": "p4I44HW3uFcR"
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "accelerator": "GPU",
  "colab": {
   "collapsed_sections": [],
   "name": "Telugu_script_alphabets.ipynb",
   "provenance": []
  },
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}