icdar_preprocess.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## xml to voc"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import os\n",
    "from xml.dom.minidom import Document\n",
    "from xml.dom.minidom import parse\n",
    "import xml.dom.minidom\n",
    "import numpy as np\n",
    "import csv\n",
    "import cv2\n",
    "\n",
    "\n",
    "def WriterXMLFiles(filename, img_name, path, box_list, label_list, w, h, d):\n",
    "\n",
    "    # dict_box[filename]=json_dict[filename]\n",
    "    doc = xml.dom.minidom.Document()\n",
    "    root = doc.createElement('annotation')\n",
    "    doc.appendChild(root)\n",
    "\n",
    "    foldername = doc.createElement(\"folder\")\n",
    "    foldername.appendChild(doc.createTextNode(\"competition-dataset-eu_img\"))\n",
    "    root.appendChild(foldername)\n",
    "\n",
    "    nodeFilename = doc.createElement('filename')\n",
    "    nodeFilename.appendChild(doc.createTextNode(img_name))\n",
    "    root.appendChild(nodeFilename)\n",
    "\n",
    "    pathname = doc.createElement(\"path\")\n",
    "    pathname.appendChild(doc.createTextNode(\"xxxx\"))\n",
    "    root.appendChild(pathname)\n",
    "\n",
    "    sourcename=doc.createElement(\"source\")\n",
    "\n",
    "    databasename = doc.createElement(\"database\")\n",
    "    databasename.appendChild(doc.createTextNode(\"icdar2013\"))\n",
    "    sourcename.appendChild(databasename)\n",
    "\n",
    "    annotationname = doc.createElement(\"annotation\")\n",
    "    annotationname.appendChild(doc.createTextNode(\"xxx\"))\n",
    "    sourcename.appendChild(annotationname)\n",
    "\n",
    "    imagename = doc.createElement(\"image\")\n",
    "    imagename.appendChild(doc.createTextNode(\"xxx\"))\n",
    "    sourcename.appendChild(imagename)\n",
    "\n",
    "    flickridname = doc.createElement(\"flickrid\")\n",
    "    flickridname.appendChild(doc.createTextNode(\"0\"))\n",
    "    sourcename.appendChild(flickridname)\n",
    "\n",
    "    root.appendChild(sourcename)\n",
    "\n",
    "    nodesize = doc.createElement('size')\n",
    "    nodewidth = doc.createElement('width')\n",
    "    nodewidth.appendChild(doc.createTextNode(str(w)))\n",
    "    nodesize.appendChild(nodewidth)\n",
    "    nodeheight = doc.createElement('height')\n",
    "    nodeheight.appendChild(doc.createTextNode(str(h)))\n",
    "    nodesize.appendChild(nodeheight)\n",
    "    nodedepth = doc.createElement('depth')\n",
    "    nodedepth.appendChild(doc.createTextNode(str(d)))\n",
    "    nodesize.appendChild(nodedepth)\n",
    "    root.appendChild(nodesize)\n",
    "\n",
    "    segname = doc.createElement(\"segmented\")\n",
    "    segname.appendChild(doc.createTextNode(\"0\"))\n",
    "    root.appendChild(segname)\n",
    "\n",
    "    for (box, label) in zip(box_list, label_list):\n",
    "\n",
    "        nodeobject = doc.createElement('object')\n",
    "        nodename = doc.createElement('name')\n",
    "        nodename.appendChild(doc.createTextNode(str(label)))\n",
    "        nodeobject.appendChild(nodename)\n",
    "        nodebndbox = doc.createElement('bndbox')\n",
    "        nodex1 = doc.createElement('xmin')\n",
    "        nodex1.appendChild(doc.createTextNode(str(box[0])))\n",
    "        nodebndbox.appendChild(nodex1)\n",
    "        nodey1 = doc.createElement('ymin')\n",
    "        nodey1.appendChild(doc.createTextNode(str(box[1])))\n",
    "        nodebndbox.appendChild(nodey1)\n",
    "        nodex2 = doc.createElement('xmax')\n",
    "        nodex2.appendChild(doc.createTextNode(str(box[2])))\n",
    "        nodebndbox.appendChild(nodex2)\n",
    "        nodey2 = doc.createElement('ymax')\n",
    "        nodey2.appendChild(doc.createTextNode(str(box[3])))\n",
    "        nodebndbox.appendChild(nodey2)\n",
    "\n",
    "        # ang = doc.createElement('angle')\n",
    "        # ang.appendChild(doc.createTextNode(str(angle)))\n",
    "        # nodebndbox.appendChild(ang)\n",
    "        nodeobject.appendChild(nodebndbox)\n",
    "        root.appendChild(nodeobject)\n",
    "    fp = open(path + filename, 'w')\n",
    "    doc.writexml(fp, indent='\\n')\n",
    "    fp.close()\n",
    "\n",
    "\n",
    "def load_annoataion(p):\n",
    "    '''\n",
    "    load annotation from the text file\n",
    "    :param p:\n",
    "    :return:\n",
    "    '''\n",
    "    text_page = []\n",
    "    text_polys = []\n",
    "    text_tags = []\n",
    "    if not os.path.exists(p):\n",
    "        return np.array(text_polys, dtype=np.float32)\n",
    "    with open(p, 'r') as f:\n",
    "        dom = xml.dom.minidom.parse(p)\n",
    "        root = dom.documentElement\n",
    "        tables = root.getElementsByTagName('table')\n",
    "\n",
    "        for table in tables:\n",
    "            region = table.getElementsByTagName('region')\n",
    "            page = region[0].getAttribute('page')\n",
    "            bbox = table.getElementsByTagName('bounding-box')\n",
    "            box_x1 = bbox[0].getAttribute('x1')\n",
    "            box_y1 = bbox[0].getAttribute('y1')\n",
    "            box_x2 = bbox[0].getAttribute('x2')\n",
    "            box_y2 = bbox[0].getAttribute('y2')\n",
    "            \n",
    "            text_page.append(page)\n",
    "            text_polys.append([box_x1, box_y1, box_x2, box_y2])\n",
    "            text_tags.append(1)\n",
    "\n",
    "        return np.array(text_page, dtype=np.int32), np.array(text_polys, dtype=np.int32), np.array(text_tags, dtype=np.str)\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "True\n",
      "gt_eu-001_1.xml\n",
      "gt_eu-001_2.xml\n",
      "gt_eu-001_3.xml\n",
      "gt_eu-002_1.xml\n",
      "gt_eu-003_1.xml\n",
      "gt_eu-004_2.xml\n",
      "gt_eu-004_3.xml\n",
      "gt_eu-004_4.xml\n",
      "gt_eu-004_6.xml\n",
      "gt_eu-004_7.xml\n",
      "gt_eu-004_8.xml\n",
      "gt_eu-004_9.xml\n",
      "gt_eu-004_10.xml\n",
      "gt_eu-004_11.xml\n",
      "gt_eu-004_12.xml\n",
      "gt_eu-004_14.xml\n",
      "gt_eu-005_2.xml\n",
      "gt_eu-006_1.xml\n",
      "gt_eu-006_2.xml\n",
      "gt_eu-006_3.xml\n",
      "gt_eu-007_1.xml\n",
      "gt_eu-007_2.xml\n",
      "gt_eu-007_3.xml\n",
      "gt_eu-007_5.xml\n",
      "gt_eu-008_1.xml\n",
      "gt_eu-009a_1.xml\n",
      "gt_eu-010_1.xml\n",
      "gt_eu-011_3.xml\n",
      "gt_eu-012_3.xml\n",
      "gt_eu-012_4.xml\n",
      "gt_eu-012_5.xml\n",
      "gt_eu-013_8.xml\n",
      "gt_eu-013_3.xml\n",
      "gt_eu-013_4.xml\n",
      "gt_eu-013_5.xml\n",
      "gt_eu-014_2.xml\n",
      "gt_eu-015_1.xml\n",
      "gt_eu-015_2.xml\n",
      "gt_eu-016_3.xml\n",
      "gt_eu-017_2.xml\n",
      "gt_eu-018_1.xml\n",
      "gt_eu-019_3.xml\n",
      "gt_eu-020_2.xml\n",
      "gt_eu-020_3.xml\n",
      "gt_eu-021_3.xml\n",
      "gt_eu-021_7.xml\n",
      "gt_eu-022_2.xml\n",
      "gt_eu-023_3.xml\n",
      "gt_eu-024_2.xml\n",
      "gt_eu-025_2.xml\n",
      "gt_eu-025_3.xml\n",
      "gt_eu-026_4.xml\n",
      "gt_eu-026_5.xml\n",
      "gt_eu-026_6.xml\n",
      "gt_eu-027_3.xml\n"
     ]
    }
   ],
   "source": [
    "pdf_path = r'D:/tablebank/icdar2013-competition-dataset-with-gt/competition-dataset-eu/'\n",
    "img_path = r'D:/tablebank/icdar2013-competition-dataset-with-gt/competition-dataset-eu_img/'\n",
    "xml_path = r'D:/tablebank/icdar2013-competition-dataset-with-gt/competition-dataset-eu/'\n",
    "gt_path = r'D:/tablebank/icdar2013-competition-dataset-with-gt/competition-dataset-eu_gt/'\n",
    "\n",
    "print(os.path.exists(pdf_path))\n",
    "txts = os.listdir(xml_path)\n",
    "for count, t in enumerate(txts):\n",
    "    if t.split('-reg')[-1]=='.xml':\n",
    "        pages, boxes, labels = load_annoataion(os.path.join(xml_path, t))\n",
    "\n",
    "\n",
    "        page_set = set(pages)\n",
    "        for p in page_set:\n",
    "            boxes_tmp = boxes[pages == p]\n",
    "            labels_tmp = labels[pages == p]    \n",
    "\n",
    "            img_name = img_path+t.split('-reg')[0]+'_'+str(p)+'.jpg'\n",
    "            img = cv2.imread(img_name)\n",
    "            if img is None:\n",
    "                continue;\n",
    "            # if img != None:\n",
    "            h, w, d = img.shape \n",
    "            output_name = 'gt_' + t.split('-reg')[0]+'_'+str(p)+'.xml'\n",
    "            img_name = t.split('-reg')[0]+'_'+str(p)+'.jpg'\n",
    "            print(output_name) \n",
    "            #WriterXMLFiles(output_name, img_name, gt_path, boxes_tmp, labels_tmp, w, h, d)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## voc to coco"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "ename": "ImportError",
     "evalue": "cannot import name 'geometry'\nIt seems that scikit-image has not been built correctly.\n\nYour install of scikit-image appears to be broken.\nTry re-installing the package following the instructions at:\nhttp://scikit-image.org/docs/stable/install.html ",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[1;32mF:\\Anaconda\\lib\\site-packages\\skimage\\__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m    152\u001b[0m     \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 153\u001b[1;33m         \u001b[1;32mfrom\u001b[0m \u001b[1;33m.\u001b[0m\u001b[0m_shared\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mgeometry\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    154\u001b[0m         \u001b[1;32mdel\u001b[0m \u001b[0mgeometry\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mImportError\u001b[0m: cannot import name 'geometry'",
      "\nDuring handling of the above exception, another exception occurred:\n",
      "\u001b[1;31mImportError\u001b[0m                               Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-1-4e73b46e2282>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m     12\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mmatplotlib\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mpyplot\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mplt\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     13\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m---> 14\u001b[1;33m \u001b[1;32mimport\u001b[0m \u001b[0mskimage\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mio\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0mio\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     15\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     16\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[0mcv2\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mF:\\Anaconda\\lib\\site-packages\\skimage\\__init__.py\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m    154\u001b[0m         \u001b[1;32mdel\u001b[0m \u001b[0mgeometry\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    155\u001b[0m     \u001b[1;32mexcept\u001b[0m \u001b[0mImportError\u001b[0m \u001b[1;32mas\u001b[0m \u001b[0me\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 156\u001b[1;33m         \u001b[0m_raise_build_error\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0me\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m    157\u001b[0m     \u001b[1;32mfrom\u001b[0m \u001b[1;33m.\u001b[0m\u001b[0mutil\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mdtype\u001b[0m \u001b[1;32mimport\u001b[0m \u001b[1;33m*\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    158\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;32mF:\\Anaconda\\lib\\site-packages\\skimage\\__init__.py\u001b[0m in \u001b[0;36m_raise_build_error\u001b[1;34m(e)\u001b[0m\n\u001b[0;32m    135\u001b[0m     raise ImportError(\"\"\"%s\n\u001b[0;32m    136\u001b[0m \u001b[0mIt\u001b[0m \u001b[0mseems\u001b[0m \u001b[0mthat\u001b[0m \u001b[0mscikit\u001b[0m\u001b[1;33m-\u001b[0m\u001b[0mimage\u001b[0m \u001b[0mhas\u001b[0m \u001b[1;32mnot\u001b[0m \u001b[0mbeen\u001b[0m \u001b[0mbuilt\u001b[0m \u001b[0mcorrectly\u001b[0m\u001b[1;33m.\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m--> 137\u001b[1;33m %s\"\"\" % (e, msg))\n\u001b[0m\u001b[0;32m    138\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m    139\u001b[0m \u001b[1;32mtry\u001b[0m\u001b[1;33m:\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mImportError\u001b[0m: cannot import name 'geometry'\nIt seems that scikit-image has not been built correctly.\n\nYour install of scikit-image appears to be broken.\nTry re-installing the package following the instructions at:\nhttp://scikit-image.org/docs/stable/install.html "
     ]
    }
   ],
   "source": [
    "\n",
    "# -*- coding:utf-8 -*-\n",
    "\n",
    "# !/usr/bin/env python\n",
    "\n",
    " \n",
    "\n",
    "import argparse\n",
    "\n",
    "import json\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import skimage.io as io\n",
    "\n",
    "import cv2\n",
    "\n",
    "from labelme import utils\n",
    "\n",
    "import numpy as np\n",
    "\n",
    "import glob\n",
    "\n",
    "import PIL.Image\n",
    "\n",
    "import os,sys\n",
    "\n",
    " \n",
    "\n",
    " \n",
    "\n",
    "class PascalVOC2coco(object):\n",
    "\n",
    "    def __init__(self, xml=[], save_json_path='./new.json'):\n",
    "\n",
    "        '''\n",
    "\n",
    "        :param xml: 所有Pascal VOC的xml文件路径组成的列表\n",
    "\n",
    "        :param save_json_path: json保存位置\n",
    "\n",
    "        '''\n",
    "\n",
    "        self.xml = xml\n",
    "\n",
    "        self.save_json_path = save_json_path\n",
    "\n",
    "        self.images = []\n",
    "\n",
    "        self.categories = []\n",
    "\n",
    "        self.annotations = []\n",
    "\n",
    "        # self.data_coco = {}\n",
    "\n",
    "        self.label = []\n",
    "\n",
    "        self.annID = 1\n",
    "\n",
    "        self.height = 0\n",
    "\n",
    "        self.width = 0\n",
    "\n",
    "        self.ob = []\n",
    "\n",
    " \n",
    "\n",
    "        self.save_json()\n",
    "\n",
    " \n",
    "\n",
    "    def data_transfer(self):\n",
    "\n",
    "        for num, json_file in enumerate(self.xml):\n",
    "\n",
    " \n",
    "\n",
    "            # 进度输出\n",
    "\n",
    "            sys.stdout.write('\\r>> Converting image %d/%d' % (\n",
    "\n",
    "                num + 1, len(self.xml)))\n",
    "\n",
    "            sys.stdout.flush()\n",
    "\n",
    " \n",
    "\n",
    "            self.json_file = json_file\n",
    "\n",
    "            #print(\"self.json\", self.json_file)\n",
    "\n",
    "            self.num = num \n",
    "\n",
    "            #print(self.num)\n",
    "\n",
    "            path = os.path.dirname(self.json_file)\n",
    "\n",
    "            #print(path)\n",
    "\n",
    "            path = os.path.dirname(path)\n",
    "\n",
    "            #print(path)\n",
    "\n",
    "            # path=os.path.split(self.json_file)[0]\n",
    "\n",
    "            # path=os.path.split(path)[0]\n",
    "\n",
    "            obj_path = glob.glob(os.path.join(path, 'SegmentationObject', '*.png'))\n",
    "\n",
    "            #print(obj_path)\n",
    "\n",
    "            with open(json_file, 'r') as fp:\n",
    "\n",
    "                #print(fp)\n",
    "\n",
    "                flag = 0\n",
    "\n",
    "                for p in fp:\n",
    "\n",
    "                    #print(p)\n",
    "\n",
    "                    # if 'folder' in p:\n",
    "\n",
    "                    #     folder =p.split('>')[1].split('<')[0]\n",
    "\n",
    "                    f_name = 1\n",
    "\n",
    "                    if 'filename' in p:\n",
    "\n",
    "                        self.filen_ame = p.split('>')[1].split('<')[0]\n",
    "\n",
    "                        #print(self.filen_ame)\n",
    "\n",
    "                        f_name = 0\n",
    "\n",
    " \n",
    "\n",
    "                        self.path = os.path.join(path, 'SegmentationObject', self.filen_ame.split('.')[0] + '.png')\n",
    "\n",
    "                        #if self.path not in obj_path:\n",
    "\n",
    "                        #    break\n",
    "\n",
    " \n",
    "\n",
    " \n",
    "\n",
    "                    if 'width' in p:\n",
    "\n",
    "                        self.width = int(p.split('>')[1].split('<')[0])\n",
    "\n",
    "                        #print(self.width)\n",
    "\n",
    "                    if 'height' in p:\n",
    "\n",
    "                        self.height = int(p.split('>')[1].split('<')[0])\n",
    "\n",
    " \n",
    "\n",
    "                        self.images.append(self.image())\n",
    "\n",
    "                        #print(self.image())\n",
    "\n",
    " \n",
    "\n",
    "                    if flag == 1:\n",
    "\n",
    "                        self.supercategory = self.ob[0]\n",
    "\n",
    "                        if self.supercategory not in self.label:\n",
    "\n",
    "                            self.categories.append(self.categorie())\n",
    "\n",
    "                            self.label.append(self.supercategory)\n",
    "\n",
    " \n",
    "\n",
    "                        # 边界框\n",
    "\n",
    "                        x1 = int(self.ob[1]);\n",
    "\n",
    "                        y1 = int(self.ob[2]);\n",
    "\n",
    "                        x2 = int(self.ob[3]);\n",
    "\n",
    "                        y2 = int(self.ob[4])\n",
    "\n",
    "                        self.rectangle = [x1, y1, x2, y2]\n",
    "\n",
    "                        self.bbox = [x1, y1, x2 - x1, y2 - y1]  # COCO 对应格式[x,y,w,h]\n",
    "\n",
    " \n",
    "\n",
    "                        self.annotations.append(self.annotation())\n",
    "\n",
    "                        self.annID += 1\n",
    "\n",
    "                        self.ob = []\n",
    "\n",
    "                        flag = 0\n",
    "\n",
    "                    elif f_name == 1:\n",
    "\n",
    "                        if 'name' in p:\n",
    "\n",
    "                            self.ob.append(p.split('>')[1].split('<')[0])\n",
    "\n",
    " \n",
    "\n",
    "                        if 'xmin' in p:\n",
    "\n",
    "                            self.ob.append(p.split('>')[1].split('<')[0])\n",
    "\n",
    " \n",
    "\n",
    "                        if 'ymin' in p:\n",
    "\n",
    "                            self.ob.append(p.split('>')[1].split('<')[0])\n",
    "\n",
    " \n",
    "\n",
    "                        if 'xmax' in p:\n",
    "\n",
    "                            self.ob.append(p.split('>')[1].split('<')[0])\n",
    "\n",
    " \n",
    "\n",
    "                        if 'ymax' in p:\n",
    "\n",
    "                            self.ob.append(p.split('>')[1].split('<')[0])\n",
    "\n",
    "                            flag = 1\n",
    "\n",
    " \n",
    "\n",
    "                    '''\n",
    "\n",
    "                    if '<object>' in p:\n",
    "\n",
    "                        # 类别\n",
    "\n",
    "                        print(next(fp))\n",
    "\n",
    "                        d = [next(fp).split('>')[1].split('<')[0] for _ in range(7)]\n",
    "\n",
    "                        self.supercategory = d[0]\n",
    "\n",
    "                        if self.supercategory not in self.label:\n",
    "\n",
    "                            self.categories.append(self.categorie())\n",
    "\n",
    "                            self.label.append(self.supercategory)\n",
    "\n",
    "\n",
    "\n",
    "                        # 边界框\n",
    "\n",
    "                        x1 = int(d[-4]);\n",
    "\n",
    "                        y1 = int(d[-3]);\n",
    "\n",
    "                        x2 = int(d[-2]);\n",
    "\n",
    "                        y2 = int(d[-1])\n",
    "\n",
    "                        self.rectangle = [x1, y1, x2, y2]\n",
    "\n",
    "                        self.bbox = [x1, y1, x2 - x1, y2 - y1]  # COCO 对应格式[x,y,w,h]\n",
    "\n",
    "\n",
    "\n",
    "                        self.annotations.append(self.annotation())\n",
    "\n",
    "                        self.annID += 1\n",
    "\n",
    "                     '''\n",
    "\n",
    " \n",
    "\n",
    "        sys.stdout.write('\\n')\n",
    "\n",
    "        sys.stdout.flush()\n",
    "\n",
    " \n",
    "\n",
    "    def image(self):\n",
    "\n",
    "        image = {}\n",
    "\n",
    "        image['height'] = self.height\n",
    "\n",
    "        image['width'] = self.width\n",
    "\n",
    "        image['id'] = self.num + 1\n",
    "\n",
    "        image['file_name'] = self.filen_ame\n",
    "\n",
    "        return image\n",
    "\n",
    " \n",
    "\n",
    "    def categorie(self):\n",
    "\n",
    "        categorie = {}\n",
    "\n",
    "        categorie['supercategory'] = self.supercategory\n",
    "\n",
    "        categorie['id'] = len(self.label) + 1  # 0 默认为背景\n",
    "\n",
    "        categorie['name'] = self.supercategory\n",
    "\n",
    "        return categorie\n",
    "\n",
    " \n",
    "\n",
    "    def annotation(self):\n",
    "\n",
    "        annotation = {}\n",
    "\n",
    "        # annotation['segmentation'] = [self.getsegmentation()]\n",
    "\n",
    "        annotation['segmentation'] = [list(map(float, self.getsegmentation()))]\n",
    "\n",
    "        annotation['iscrowd'] = 0\n",
    "\n",
    "        annotation['image_id'] = self.num + 1\n",
    "\n",
    "        # annotation['bbox'] = list(map(float, self.bbox))\n",
    "\n",
    "        annotation['bbox'] = self.bbox\n",
    "\n",
    "        annotation['category_id'] = self.getcatid(self.supercategory)\n",
    "\n",
    "        annotation['id'] = self.annID\n",
    "\n",
    "        return annotation\n",
    "\n",
    " \n",
    "\n",
    "    def getcatid(self, label):\n",
    "\n",
    "        for categorie in self.categories:\n",
    "\n",
    "            if label == categorie['name']:\n",
    "\n",
    "                return categorie['id']\n",
    "\n",
    "        return -1\n",
    "\n",
    " \n",
    "\n",
    "    def getsegmentation(self):\n",
    "\n",
    " \n",
    "\n",
    "        try:\n",
    "\n",
    "            mask_1 = cv2.imread(self.path, 0)\n",
    "\n",
    "            mask = np.zeros_like(mask_1, np.uint8)\n",
    "\n",
    "            rectangle = self.rectangle\n",
    "\n",
    "            mask[rectangle[1]:rectangle[3], rectangle[0]:rectangle[2]] = mask_1[rectangle[1]:rectangle[3],\n",
    "\n",
    "                                                                         rectangle[0]:rectangle[2]]\n",
    "\n",
    " \n",
    "\n",
    "            # 计算矩形中点像素值\n",
    "\n",
    "            mean_x = (rectangle[0] + rectangle[2]) // 2\n",
    "\n",
    "            mean_y = (rectangle[1] + rectangle[3]) // 2\n",
    "\n",
    " \n",
    "\n",
    "            end = min((mask.shape[1], int(rectangle[2]) + 1))\n",
    "\n",
    "            start = max((0, int(rectangle[0]) - 1))\n",
    "\n",
    " \n",
    "\n",
    "            flag = True\n",
    "\n",
    "            for i in range(mean_x, end):\n",
    "\n",
    "                x_ = i;\n",
    "\n",
    "                y_ = mean_y\n",
    "\n",
    "                pixels = mask_1[y_, x_]\n",
    "\n",
    "                if pixels != 0 and pixels != 220:  # 0 对应背景 220对应边界线\n",
    "\n",
    "                    mask = (mask == pixels).astype(np.uint8)\n",
    "\n",
    "                    flag = False\n",
    "\n",
    "                    break\n",
    "\n",
    "            if flag:\n",
    "\n",
    "                for i in range(mean_x, start, -1):\n",
    "\n",
    "                    x_ = i;\n",
    "\n",
    "                    y_ = mean_y\n",
    "\n",
    "                    pixels = mask_1[y_, x_]\n",
    "\n",
    "                    if pixels != 0 and pixels != 220:\n",
    "\n",
    "                        mask = (mask == pixels).astype(np.uint8)\n",
    "\n",
    "                        break\n",
    "\n",
    "            self.mask = mask\n",
    "\n",
    " \n",
    "\n",
    "            return self.mask2polygons()\n",
    "\n",
    " \n",
    "\n",
    "        except:\n",
    "\n",
    "            return [0]\n",
    "\n",
    " \n",
    "\n",
    "    def mask2polygons(self):\n",
    "\n",
    "        contours = cv2.findContours(self.mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)  # 找到轮廓线\n",
    "\n",
    "        bbox=[]\n",
    "\n",
    "        for cont in contours[1]:\n",
    "\n",
    "            [bbox.append(i) for i in list(cont.flatten())]\n",
    "\n",
    "            # map(bbox.append,list(cont.flatten()))\n",
    "\n",
    "        return bbox # list(contours[1][0].flatten())\n",
    "\n",
    " \n",
    "\n",
    "    # '''\n",
    "\n",
    "    def getbbox(self, points):\n",
    "\n",
    "        # img = np.zeros([self.height,self.width],np.uint8)\n",
    "\n",
    "        # cv2.polylines(img, [np.asarray(points)], True, 1, lineType=cv2.LINE_AA)  # 画边界线\n",
    "\n",
    "        # cv2.fillPoly(img, [np.asarray(points)], 1)  # 画多边形 内部像素值为1\n",
    "\n",
    "        polygons = points\n",
    "\n",
    "        mask = self.polygons_to_mask([self.height, self.width], polygons)\n",
    "\n",
    "        return self.mask2box(mask)\n",
    "\n",
    " \n",
    "\n",
    "    def mask2box(self, mask):\n",
    "\n",
    "        '''从mask反算出其边框\n",
    "\n",
    "        mask：[h,w]  0、1组成的图片\n",
    "\n",
    "        1对应对象，只需计算1对应的行列号（左上角行列号，右下角行列号，就可以算出其边框）\n",
    "\n",
    "        '''\n",
    "\n",
    "        # np.where(mask==1)\n",
    "\n",
    "        index = np.argwhere(mask == 1)\n",
    "\n",
    "        rows = index[:, 0]\n",
    "\n",
    "        clos = index[:, 1]\n",
    "\n",
    "        # 解析左上角行列号\n",
    "\n",
    "        left_top_r = np.min(rows)  # y\n",
    "\n",
    "        left_top_c = np.min(clos)  # x\n",
    "\n",
    " \n",
    "\n",
    "        # 解析右下角行列号\n",
    "\n",
    "        right_bottom_r = np.max(rows)\n",
    "\n",
    "        right_bottom_c = np.max(clos)\n",
    "\n",
    " \n",
    "\n",
    "        # return [(left_top_r,left_top_c),(right_bottom_r,right_bottom_c)]\n",
    "\n",
    "        # return [(left_top_c, left_top_r), (right_bottom_c, right_bottom_r)]\n",
    "\n",
    "        # return [left_top_c, left_top_r, right_bottom_c, right_bottom_r]  # [x1,y1,x2,y2]\n",
    "\n",
    "        return [left_top_c, left_top_r, right_bottom_c - left_top_c,\n",
    "\n",
    "                right_bottom_r - left_top_r]  # [x1,y1,w,h] 对应COCO的bbox格式\n",
    "\n",
    " \n",
    "\n",
    "    def polygons_to_mask(self, img_shape, polygons):\n",
    "\n",
    "        mask = np.zeros(img_shape, dtype=np.uint8)\n",
    "\n",
    "        mask = PIL.Image.fromarray(mask)\n",
    "\n",
    "        xy = list(map(tuple, polygons))\n",
    "\n",
    "        PIL.ImageDraw.Draw(mask).polygon(xy=xy, outline=1, fill=1)\n",
    "\n",
    "        mask = np.array(mask, dtype=bool)\n",
    "\n",
    "        return mask\n",
    "\n",
    " \n",
    "\n",
    "    # '''\n",
    "\n",
    "    def data2coco(self):\n",
    "\n",
    "        data_coco = {}\n",
    "\n",
    "        data_coco['images'] = self.images\n",
    "\n",
    "        data_coco['categories'] = self.categories\n",
    "\n",
    "        data_coco['annotations'] = self.annotations\n",
    "\n",
    "        return data_coco\n",
    "\n",
    " \n",
    "\n",
    "    def save_json(self):\n",
    "\n",
    "        self.data_transfer()\n",
    "\n",
    "        self.data_coco = self.data2coco()\n",
    "\n",
    "        # 保存json文件\n",
    "\n",
    "        json.dump(self.data_coco, open(self.save_json_path, 'w'), indent=4)  # indent=4 更加美观显示\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "xml_file = glob.glob('./icdar2013-competition-dataset-with-gt/competition-dataset-eu_gt/*.xml')\n",
    "\n",
    "PascalVOC2coco(xml_file, './eu.json')"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.2"
  },
  "latex_envs": {
   "LaTeX_envs_menu_present": true,
   "autoclose": false,
   "autocomplete": true,
   "bibliofile": "biblio.bib",
   "cite_by": "apalike",
   "current_citInitial": 1,
   "eqLabelWithNumbers": true,
   "eqNumInitial": 1,
   "hotkeys": {
    "equation": "Ctrl-E",
    "itemize": "Ctrl-I"
   },
   "labels_anchors": false,
   "latex_user_defs": false,
   "report_style_numbering": false,
   "user_envs_cfg": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}