Data_Preparaion.ipynb

{
 "cells": [
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Data Preparation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Count of Each Class/Character\n",
    "\n",
    "It is expected/assumed that the dataset is not split. The below script will read the data from parent dataset directory and process accordingly."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from collections import defaultdict\n",
    "import os, shutil\n",
    "\n",
    "dataset_path = \"./BanglaLekha_Dataset\"\n",
    "\n",
    "label_freq = defaultdict(int)\n",
    "if os.path.exists(os.path.join(dataset_path)):\n",
    "    classes = os.listdir(dataset_path)\n",
    "    for i in classes :\n",
    "        label_freq[i] = len(os.listdir(os.path.join(dataset_path, i)))\n",
    "    plt.bar(label_freq.keys(), label_freq.values())\n",
    "    plt.xlabel('Character ID')\n",
    "    plt.ylabel('Count')\n",
    "    plt.title('Class Frequency Distribution')\n",
    "    plt.ylim(min(label_freq.values())-10, max(label_freq.values())+10)\n",
    "    plt.plot()\n",
    "else:\n",
    "    print(\"Dataset folder missing\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Average, Maximum, Minimum Count"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Average number of characters present per class = 1977.4405\n",
      "Number of Classes having less images than average => 26\n",
      "Maximum images a class has : 1988\n",
      "Minimum images a class has : 1940\n"
     ]
    }
   ],
   "source": [
    "avg_img_per_class = sum(label_freq.values())/len(label_freq.values())\n",
    "print(f\"Average number of characters present per class = {avg_img_per_class:.4f}\")\n",
    "\n",
    "print(\"Number of Classes having less images than average =>\", end=' ')\n",
    "count = 0\n",
    "for (i, v) in label_freq.items():\n",
    "    if v <= avg_img_per_class:\n",
    "        count +=1\n",
    "print(f\"{count}\")\n",
    "\n",
    "min_img = min(label_freq.values())\n",
    "max_img = max(label_freq.values())\n",
    "print(f\"Maximum images a class has : {max_img}\\nMinimum images a class has : {min_img}\")\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Splitting into Train and Test\n",
    "\n",
    "If the dataset is not split into **Train** and **Test**, then this snippet will do the job. [This Repo](https://www.kaggle.com/datasets/ayanwap7/banglalekhaisolated-train-test-split) on Kaggle comes with Train and Test Splits already performed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Total Training samples = 162120\n",
      "Total Validation/Test samples = 3985\n"
     ]
    }
   ],
   "source": [
    "classes = list(label_freq.keys())\n",
    "train_size = min_img - 10\n",
    "\n",
    "if not os.path.exists(os.path.join(dataset_path, 'train')):\n",
    "    os.mkdir(os.path.join(dataset_path, 'train'))\n",
    "\n",
    "    for i in classes :\n",
    "        os.mkdir(os.path.join(dataset_path, 'train', i))\n",
    "\n",
    "if not os.path.exists(os.path.join(dataset_path, 'test')):\n",
    "    os.mkdir(os.path.join(dataset_path, 'test'))\n",
    "\n",
    "    for i in classes :\n",
    "        os.mkdir(os.path.join(dataset_path, 'test', i))\n",
    "\n",
    "if len(os.listdir(os.path.join(dataset_path, 'train', classes[0])))==0 :\n",
    "    for i in classes :\n",
    "        all_images = os.listdir(os.path.join(dataset_path, i))\n",
    "\n",
    "        for img in all_images[:train_size]:\n",
    "            shutil.move(os.path.join(dataset_path, i, img), os.path.join(dataset_path, 'train', i, img))\n",
    "        \n",
    "        for img in all_images[train_size:]:\n",
    "            shutil.move(os.path.join(dataset_path, i, img), os.path.join(dataset_path, 'test', i, img))\n",
    "        \n",
    "        if len(os.listdir(os.path.join(dataset_path, i)))==0:\n",
    "            os.rmdir(os.path.join(dataset_path, i))\n",
    "\n",
    "train_size = 0\n",
    "test_size = 0\n",
    "for i in classes :\n",
    "    train_size += len(os.listdir(os.path.join(dataset_path, 'train', i)))\n",
    "    test_size += len(os.listdir(os.path.join(dataset_path, 'test', i)))\n",
    "\n",
    "print(f\"Total Training samples = {train_size}\\nTotal Validation/Test samples = {test_size}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "To handle the **Class Imbalance** problem, I have performed the above EDA and put **equal number of samples** from each class in the Train Split. The rest of the samples from each class is kept in the Validation/Test split. You can see the output of the below cell to get the numbers."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "TRAIN :\n",
      "Maximum number of items in a class = 1930\n",
      "Minimum number of items in a class = 1930\n",
      "\n",
      "VALIDATION/TEST :\n",
      "Maximum number of items in a class = 58\n",
      "Minimum number of items in a class = 10\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "train_dict = defaultdict(int)\n",
    "test_dict = defaultdict(int)\n",
    "\n",
    "if os.path.exists(os.path.join(dataset_path, 'train')):\n",
    "    classes = os.listdir(os.path.join(dataset_path, 'train'))\n",
    "    for i in classes :\n",
    "        train_dict[i] = len(os.listdir(os.path.join(dataset_path, 'train', i)))\n",
    "        test_dict[i] = len(os.listdir(os.path.join(dataset_path, 'test', i)))\n",
    "\n",
    "print(f\"TRAIN :\\nMaximum number of items in a class = {max(train_dict.values())}\\nMinimum number of items in a class = {min(train_dict.values())}\")\n",
    "print(f\"\\nVALIDATION/TEST :\\nMaximum number of items in a class = {max(test_dict.values())}\\nMinimum number of items in a class = {min(test_dict.values())}\")\n",
    "\n",
    "plt.bar(train_dict.keys(), train_dict.values())\n",
    "#plt.xlabel('Character ID')\n",
    "plt.ylabel('Count')\n",
    "plt.title('Class Frequency Distribution')\n",
    "plt.ylim(min(train_dict.values())-10, max(train_dict.values())+10)\n",
    "plt.show()\n",
    "\n",
    "plt.bar(test_dict.keys(), test_dict.values(), color='red')\n",
    "#plt.xlabel('Character ID')\n",
    "plt.ylabel('Count')\n",
    "plt.title('Class Frequency Distribution')\n",
    "plt.ylim(min(test_dict.values())-10, max(test_dict.values())+10)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Random Image Visualization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Actual Image size = (224, 224)\n"
     ]
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "from PIL import Image\n",
    "\n",
    "random_index_class = str(np.random.randint(1, 84))\n",
    "all_images = os.listdir(os.path.join(dataset_path, 'train', random_index_class))\n",
    "\n",
    "random_index_image = np.random.randint(1, len(all_images))\n",
    "a = Image.open(os.path.join(dataset_path, 'train', random_index_class, all_images[random_index_image]))\n",
    "plt.imshow(np.asarray(a))\n",
    "\n",
    "print(f\"Actual Image size = {a.size}\")"
   ]
  },
  {
   "attachments": {},
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Resizing Images\n",
    "Run this cell again ONLY if you wish to resize the images. Else DON'T RUN.\n",
    "\n",
    "The Height and Width of all the images in this dataset is less than 224. Therefore, for compatibility, we need to pad and resize the images."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Execution time = 8m 13.6s\n",
    "import torchvision.transforms.functional as TF\n",
    "import math\n",
    "\n",
    "def resize(img, size):\n",
    "    left = math.ceil((size - img.size[0])/2)\n",
    "    top = math.floor((size - img.size[1])/2)\n",
    "    right = math.floor((size - img.size[0])/2)\n",
    "    bottom = math.ceil((size - img.size[1])/2)\n",
    "    resized = TF.pad(img, (left, top, right, bottom))\n",
    "    return resized\n",
    "\n",
    "target = 224\n",
    "\n",
    "for split in ['train', 'test']:\n",
    "    for label in classes :\n",
    "        all_images = os.listdir(os.path.join(dataset_path, split, label))\n",
    "        for image in all_images :\n",
    "            img = Image.open(os.path.join(dataset_path, split, label, image))\n",
    "            resized = resize(img, target)\n",
    "            resized.save(os.path.join(dataset_path, split, label, image))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Reverse Split\n",
    "\n",
    "If the dataset is split before and you wish to split the data again, this snippet will populate all the images from the split folders into parent folder again."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os, shutil\n",
    "\n",
    "dataset_path = \"./BanglaLekha_Dataset\"\n",
    "\n",
    "classes = os.listdir(os.path.join(dataset_path, 'train'))\n",
    "for cl in classes:\n",
    "    os.mkdir(os.path.join(dataset_path, cl))\n",
    "\n",
    "for split in ['train', 'test']:\n",
    "    for cl in classes:\n",
    "        images = os.listdir(os.path.join(dataset_path, split, cl))\n",
    "        for img in images:\n",
    "            shutil.move(os.path.join(dataset_path, split, cl, img), os.path.join(dataset_path, cl, img))\n",
    "        if len(os.listdir(os.path.join(dataset_path, split, cl))) == 0:\n",
    "            shutil.rmtree(os.path.join(dataset_path, split, cl))\n",
    "\n",
    "    if len(os.listdir(os.path.join(dataset_path, split))) == 0:\n",
    "        shutil.rmtree(os.path.join(dataset_path, split))"
   ]
  }
 ],
 "metadata": {
  "interpreter": {
   "hash": "8af0c7b516e23976b6ef6c4f3f67ca7449b209a47950dd9299e28344133359b8"
  },
  "kernelspec": {
   "display_name": "Python 3.10.4 64-bit",
   "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"
  },
  "orig_nbformat": 4
 },
 "nbformat": 4,
 "nbformat_minor": 2
}