create_OCR_dataset_COMPETION+ADDITIONAL_DATA.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6a86538d",
   "metadata": {
    "papermill": {
     "duration": 0.191459,
     "end_time": "2023-05-08T10:52:01.336394",
     "exception": false,
     "start_time": "2023-05-08T10:52:01.144935",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import cv2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "5e73d02d",
   "metadata": {
    "papermill": {
     "duration": 0.014866,
     "end_time": "2023-05-08T10:52:27.779599",
     "exception": false,
     "start_time": "2023-05-08T10:52:27.764733",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "os.makedirs(\"./all_data/en_train_filtered/__results___files\",exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "3ff8e9ec",
   "metadata": {
    "papermill": {
     "duration": 0.014717,
     "end_time": "2023-05-08T10:52:27.914747",
     "exception": false,
     "start_time": "2023-05-08T10:52:27.900030",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "os.makedirs(\"./all_data/en_val/__results___files\",exist_ok=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "3d909085",
   "metadata": {
    "papermill": {
     "duration": 0.007803,
     "end_time": "2023-05-08T10:52:27.952772",
     "exception": false,
     "start_time": "2023-05-08T10:52:27.944969",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "import os\n",
    "import json\n",
    "from collections import Counter\n",
    "from tqdm import tqdm\n",
    "import glob"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "bbd4c566",
   "metadata": {},
   "outputs": [],
   "source": [
    "# !pip install pylatexenc\n",
    "from pylatexenc.latex2text import LatexNodes2Text\n",
    "latexconv = LatexNodes2Text()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "a11bdad4",
   "metadata": {
    "papermill": {
     "duration": 1139.917433,
     "end_time": "2023-05-08T11:11:27.878025",
     "exception": false,
     "start_time": "2023-05-08T10:52:27.960592",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 24 µs, sys: 0 ns, total: 24 µs\n",
      "Wall time: 28.1 µs\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "\n",
    "import os\n",
    "import json\n",
    "from collections import Counter\n",
    "from tqdm import tqdm\n",
    "def convert_to_bbox(polygon):\n",
    "    return [polygon['x0'], polygon['y0'], polygon['x1'], polygon['y1'],polygon['x2'],polygon['y2'],polygon['x3'],polygon['y3']]\n",
    "\n",
    "import hashlib\n",
    "import random\n",
    "from PIL import Image, ImageDraw\n",
    "import numpy as np\n",
    "import os\n",
    "\n",
    "def point_to_bbox(point, k=5):\n",
    "    x = point['x']\n",
    "    y = point['y'] \n",
    "    return [x, y]\n",
    "\n",
    "def point_to_bbox_tuple(x,y, k=5):\n",
    "    return [x, y]\n",
    "def get_id(texts,i):\n",
    "    for k in texts:\n",
    "        if i==k[\"id\"]:\n",
    "            return k\n",
    "def get_bounding_box(points):\n",
    "    xs = [p[0] for p in points]\n",
    "    ys = [p[1] for p in points]\n",
    "    x_min = min(xs)\n",
    "    y_min = min(ys)\n",
    "    x_max = max(xs)\n",
    "    y_max = max(ys)\n",
    "    return [x_min, y_min, x_max - x_min, y_max - y_min]\n",
    "def create_square(point, k=5):\n",
    "    if type(point)==dict:\n",
    "        x,y = point[\"x\"],point[\"y\"]\n",
    "    else:\n",
    "        x,y =point\n",
    "    half_k = k / 2.0\n",
    "\n",
    "    # Define the corners of the square\n",
    "    corners = [x - half_k, y - half_k,x + half_k, y - half_k,x + half_k, y + half_k,x - half_k, y + half_k]\n",
    "    return corners\n",
    "def polygon_to_bbox(polygon):\n",
    "    # Create list of x and y coordinates\n",
    "    x_coords = polygon[0::2]\n",
    "    y_coords = polygon[1::2]\n",
    "\n",
    "    # Compute minimum and maximum coordinates\n",
    "    x_min, x_max = min(x_coords), max(x_coords)\n",
    "    y_min, y_max = min(y_coords), max(y_coords)\n",
    "\n",
    "    # Calculate the center of the bounding box\n",
    "    x_center = (x_min + x_max) / 2.0\n",
    "    y_center = (y_min + y_max) / 2.0\n",
    "\n",
    "    # Calculate the width and height of the bounding box\n",
    "    width = x_max - x_min\n",
    "    height = y_max - y_min\n",
    "\n",
    "    # Return the bounding box in the format [x, y, width, height]\n",
    "    return [x_center, y_center, width, height]\n",
    "def get_line_bboxes(v,ctype=None):\n",
    "    texts = v[\"text\"]\n",
    "    polygon_info = []\n",
    "    point_info = []\n",
    "    for tinf in v[\"axes\"][\"x-axis\"][\"ticks\"]:\n",
    "        bbox_label = get_id(texts,tinf[\"id\"])\n",
    "        polygon_info.append(bbox_label)\n",
    "\n",
    "    for tinf in v[\"axes\"][\"y-axis\"][\"ticks\"]:\n",
    "        bbox_label = get_id(texts,tinf[\"id\"])\n",
    "        polygon_info.append(bbox_label)\n",
    "    return polygon_info\n",
    "import re\n",
    "\n",
    "def sanitize_filename(filename):\n",
    "    # This regex will match any character that is not a letter, number, space, hyphen, underscore, or dot.\n",
    "    invalid_chars_pattern = r'[\\\\/:*?\"<>|]'\n",
    "    # Replace invalid characters with an underscore.\n",
    "    sanitized_filename = re.sub(invalid_chars_pattern, '_', filename)\n",
    "    return sanitized_filename\n",
    "\n",
    "def save_cropped_patches(image_path, list_of_dicts,savepath,imno,savednames,savedtext):\n",
    "    # Open the image\n",
    "    img = Image.open(image_path)\n",
    "    im_basename = os.path.basename(image_path).split(\".\")[0]\n",
    "    # For each dict in the list\n",
    "    for d in list_of_dicts:\n",
    "        polygon = d['polygon']\n",
    "\n",
    "        # Create polygon points in the order [(x0, y0), (x1, y1), (x2, y2), (x3, y3)]\n",
    "        poly_points = [(polygon['x0'], polygon['y0']), (polygon['x1'], polygon['y1']), (polygon['x2'], polygon['y2']), (polygon['x3'], polygon['y3'])]\n",
    "\n",
    "        # Create a blank mask image\n",
    "        polygon_img = Image.new('L', (img.width, img.height), 0)\n",
    "        \n",
    "        # Draw the polygon on the mask image\n",
    "        ImageDraw.Draw(polygon_img).polygon(poly_points, outline=1, fill=1)\n",
    "\n",
    "        # Get the bounding box of the polygon\n",
    "        bbox = polygon_img.getbbox()\n",
    "\n",
    "        # Crop the bounding box directly from the original image\n",
    "        bbox_cropped_image = img.crop(bbox)\n",
    "        \n",
    "        ttxt = latexconv.latex_to_text(d['text'])\n",
    "        if len(ttxt) == 0 or any([x in im_basename for x in \"()/\"]):\n",
    "            continue\n",
    "\n",
    "        # Save the cropped bounding box with the filename convention: {dict[\"text\"]}_{randomhash}.png\n",
    "        bbox_cropped_image.convert('RGB').save(os.path.join(savepath, f\"{im_basename}_{imno}.jpg\"))\n",
    "        savednames.append(f\"{im_basename}_{imno}.jpg\")\n",
    "        savedtext.append(ttxt)\n",
    "        imno += 1\n",
    "    return imno\n",
    "\n",
    "train_image_names = []\n",
    "train_texts = []\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "54e1a7a2",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_images = sorted(glob.glob(\"./ICPR2022_CHARTINFO_UB_PMC_TRAIN_v1.0/images/*/*\"))\n",
    "train_annots = sorted(glob.glob(\"./ICPR2022_CHARTINFO_UB_PMC_TRAIN_v1.0/annotations_JSON/*/*\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "c5c1f0fb",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_images1 = {}\n",
    "for x in sorted(glob.glob(\"./ICPR2022_CHARTINFO_UB_UNITEC_PMC_TEST_v2.1/chart_images/*/*/*\")):\n",
    "    train_images1[\"\".join(os.path.split(x)[-1].split(\".\")[:-1])] = x\n",
    "train_annots1 = {}\n",
    "for x in sorted(glob.glob(\"./ICPR2022_CHARTINFO_UB_UNITEC_PMC_TEST_v2.1/final_full_GT/*/*/*.json\")):\n",
    "    train_annots1[\"\".join(os.path.split(x)[-1].split(\".\")[:-1])] = x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "8b162489",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "22923it [01:36, 236.50it/s] \n"
     ]
    }
   ],
   "source": [
    "index=0\n",
    "for imgpth,annot in tqdm(zip(train_images,train_annots)):\n",
    "    v = json.load(open(annot))\n",
    "    if v[\"task2\"] is None:\n",
    "        continue\n",
    "    polygons = v[\"task2\"][\"output\"][\"text_blocks\"]\n",
    "    index = save_cropped_patches(imgpth,polygons,\"./all_data/en_train_filtered/__results___files/\",index,train_image_names,train_texts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "b0cbb036",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "13259it [00:30, 429.01it/s]  \n"
     ]
    }
   ],
   "source": [
    "index=0\n",
    "for imgpth,annot in tqdm(zip(train_images1,train_annots1)):\n",
    "    v = json.load(open(train_annots1[annot]))\n",
    "    if v[\"task2\"] is None or v[\"task2\"][\"output\"] is None:\n",
    "        continue\n",
    "    polygons = v[\"task2\"][\"output\"][\"text_blocks\"]\n",
    "    index = save_cropped_patches(train_images1[annot],polygons,\"./all_data/en_train_filtered/__results___files/\",index,train_image_names,train_texts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "8c67e629",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import json\n",
    "from collections import Counter\n",
    "from tqdm import tqdm\n",
    "# define directories\n",
    "IMAGE_DIR = '../datasets/dataset/train/images'\n",
    "ANNOTATIONS_DIR = '../datasets/dataset/train/annotations'\n",
    "\n",
    "# get the list of images\n",
    "image_filenames = os.listdir(IMAGE_DIR)\n",
    "\n",
    "json_files = json.load(open(\"../datasets/dataset/data.json\"))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "id": "429341c2",
   "metadata": {},
   "outputs": [],
   "source": [
    "valid_files = set([x.split(\".\")[0] for x in os.listdir(\"../datasets/dataset/keypoint/labels/valid/\")])\n",
    "train_files = set([x.split(\".\")[0] for x in os.listdir(\"../datasets/dataset/keypoint/labels/train/\")])\n",
    "valid_files = {k:json_files[k] for k in json_files if k in valid_files}\n",
    "train_files = {k:json_files[k] for k in json_files if k in train_files}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "b89dd418",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 60312/60312 [09:48<00:00, 102.40it/s]\n"
     ]
    }
   ],
   "source": [
    "ROOT=\"../datasets/dataset/keypoint/images/train\"\n",
    "index=0\n",
    "for k in tqdm(train_files):\n",
    "    imgpth=os.path.join(ROOT,k+\".jpg\")\n",
    "    v = train_files[k]\n",
    "    \n",
    "    imgpth=os.path.join(ROOT,k+\".jpg\")\n",
    "    polygons = get_line_bboxes(v, v['chart-type'])\n",
    "    index = save_cropped_patches(imgpth,polygons,\"./all_data/en_train_filtered/__results___files/\",index,train_image_names,train_texts)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "24989eff",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(223, 60312)"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(valid_files),len(train_files)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "9ce1f78b",
   "metadata": {},
   "outputs": [],
   "source": [
    "train_df = pd.DataFrame(list(zip(train_image_names, train_texts)),\n",
    "               columns =['filename', 'words'])\n",
    "train_df.to_csv('./all_data/en_train_filtered/labels.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "id": "069049d8",
   "metadata": {},
   "outputs": [],
   "source": [
    "valid_image_names,valid_texts=[],[]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "5202d145",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 223/223 [00:02<00:00, 103.89it/s]\n"
     ]
    }
   ],
   "source": [
    "ROOT=\"../datasets/dataset/keypoint/images/valid\"\n",
    "index=0\n",
    "for k in tqdm(valid_files):\n",
    "    imgpth=os.path.join(ROOT,k+\".jpg\")\n",
    "    img = cv2.imread(imgpth)\n",
    "    image = img.copy()\n",
    "    h,w,_ = img.shape\n",
    "    v = valid_files[k]\n",
    "    \n",
    "    imgpth=os.path.join(ROOT,k+\".jpg\")\n",
    "    polygons = get_line_bboxes(v, v['chart-type'])\n",
    "    index = save_cropped_patches(imgpth,polygons,\"./all_data/en_val/__results___files/\",index,valid_image_names,valid_texts)\n",
    "valid_df = pd.DataFrame(list(zip(valid_image_names, valid_texts)),\n",
    "               columns =['filename', 'words'])\n",
    "valid_df.to_csv('./all_data/en_val/labels.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "c9691ce9",
   "metadata": {
    "papermill": {
     "duration": 6.649407,
     "end_time": "2023-05-08T11:11:36.642853",
     "exception": false,
     "start_time": "2023-05-08T11:11:29.993446",
     "status": "completed"
    },
    "tags": []
   },
   "outputs": [],
   "source": [
    "# train_df = pd.read_csv(\"./all_data/en_train_filtered/__results___files/labels.csv\", sep='^([^,]+),', engine='python', usecols=['filename', 'words'], keep_default_na=False)\n",
    "# train_df = train_df[train_df.filename.apply(lambda x: \".jpg\" in x)]\n",
    "# train_df = train_df.dropna().reset_index(drop=True)\n",
    "# train_df.to_csv(\"./all_data/en_train_filtered/__results___files/labels.csv\",index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "484afb9f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x720 with 9 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import os\n",
    "import random\n",
    "import matplotlib.pyplot as plt\n",
    "from PIL import Image\n",
    "\n",
    "def sample_images(folder_path, tdf,num_samples=9):\n",
    "    # Get all files in the directory\n",
    "    all_files = os.listdir(folder_path)\n",
    "\n",
    "    # Filter list for only jpg and png files\n",
    "    image_files = [file for file in all_files if file.endswith(('.jpg', '.png'))]\n",
    "\n",
    "    # Randomly sample the image files\n",
    "    sampled_files = tdf.sample(num_samples)\n",
    "\n",
    "    # Determine grid size for plotting\n",
    "    grid_size = int(num_samples**0.5)\n",
    "\n",
    "    # Create subplots\n",
    "    fig, axs = plt.subplots(grid_size, grid_size, figsize=(10,10))\n",
    "\n",
    "    for i, (file,word) in enumerate(zip(sampled_files.filename,sampled_files.words)):\n",
    "        # Open image file\n",
    "        img = Image.open(os.path.join(folder_path, file.replace(\".png\",\".jpg\")))\n",
    "        \n",
    "        # Add to subplot\n",
    "        axs[i//grid_size, i%grid_size].imshow(img)\n",
    "        axs[i//grid_size, i%grid_size].axis('off')\n",
    "        axs[i//grid_size, i%grid_size].set_title(word)\n",
    "\n",
    "    # Remove empty subplots\n",
    "    if num_samples < grid_size**2:\n",
    "        for i in range(num_samples, grid_size**2):\n",
    "            fig.delaxes(axs.flatten()[i])\n",
    "            \n",
    "    # Display the plot\n",
    "    plt.show()\n",
    "\n",
    "# call the function\n",
    "sample_images('./all_data/en_train_filtered/__results___files',train_df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e51c9026",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.7.10"
  },
  "papermill": {
   "default_parameters": {},
   "duration": 4876.035627,
   "end_time": "2023-05-08T12:13:07.856916",
   "environment_variables": {},
   "exception": null,
   "input_path": "__notebook__.ipynb",
   "output_path": "__notebook__.ipynb",
   "parameters": {},
   "start_time": "2023-05-08T10:51:51.821289",
   "version": "2.4.0"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}