devbox
知新_RL
这个屌丝很懒，什么也没留下！
关注作者
热门标签
热门文章
当前位置:   article > 正文

数据格式处理：数据转换(yolo转voc和voc转coco)及voc数据删除等_voc 格式转coco

作者：知新_RL | 2024-07-23 10:59:48

赞

踩

voc 格式转coco

yolo转voc：

  1. #!/usr/bin/env python
  2. # -*- coding: utf-8 -*-
  3. # @Time        :2021/8/31 下午3:06
  4. # @Author      :
  5. # @ProjectName :mmdetection-master
  6. # @File        :yolo2voc.py
  7. # @Description :
  8. import os
  9. import codecs
  10. import cv2
  11. import random
  12. import xml.etree.ElementTree as ET
  13. from xml.dom.minidom import parse
  14.  
  15. def vocXMLFormat(filePath, imgName, imgW, imgH, objNames, locs, depth=3, truncated=0, difficult=0):
  16.     """
  17.     Generate xml annotation file,eg:
  18.         vocXMLFormat("0000.xml", "0001.png", 608, 608, ["person", "TV"], [[24, 32, 156, 145], [124, 76, 472, 384]])
  19.     Args:
  20.         filePath:The path of the file saved in the generated xml
  21.         imgName:xml file annotates the name of the picture
  22.         imgW:
  23.         imgH:
  24.         objNames:The object contained in the picture, format: ["object label 1", "object label 2"...]
  25.         locs:The picture contains the coordinates of the object, format: [[x,y,w,h],[x,y,w,h]...]
  26.         depth:
  27.         truncated:
  28.         difficult:
  30.     Returns:
  32.     """
  33.     if (objNames == None) or (locs == None):
  34.         print("The objNames or locs is None!!!")
  35.         return
  36.     with codecs.open(filePath, 'w', 'utf-8') as xml:
  37.         xml.write('<?xml version="1.0" encoding="UTF-8"?>\n')
  38.         xml.write('<annotation>\n')
  39.         xml.write('\t<folder>' + 'voc format' + '</folder>\n')
  40.         xml.write('\t<filename>' + imgName + '</filename>\n')
  41.         xml.write('\t<path>' + imgName + '</path>\n')
  42.         xml.write('\t<source>\n')
  43.         xml.write('\t\t<database>weiz</database>\n')
  44.         xml.write('\t</source>\n')
  45.         xml.write('\t<size>\n')
  46.         xml.write('\t\t<width>' + str(imgW) + '</width>\n')
  47.         xml.write('\t\t<height>' + str(imgH) + '</height>\n')
  48.         xml.write('\t\t<depth>' + str(depth) + '</depth>\n')
  49.         xml.write('\t</size>\n')
  50.         xml.write('\t<segmented>0</segmented>\n')
  51.  
  52.         for ind, name in enumerate(objNames):
  53.             xml.write('\t<object>\n')
  54.             xml.write('\t\t<name>' + name + '</name>\n')
  55.             xml.write('\t\t<pose>Unspecified</pose>\n')
  56.             xml.write('\t\t<truncated>' + str(truncated) + '</truncated>\n')
  57.             xml.write('\t\t<difficult>' + str(difficult) + '</difficult>\n')
  58.             xml.write('\t\t<bndbox>\n')
  59.             xml.write('\t\t\t<xmin>' + str(locs[ind][0]) + '</xmin>\n')
  60.             xml.write('\t\t\t<ymin>' + str(locs[ind][1]) + '</ymin>\n')
  61.             xml.write('\t\t\t<xmax>' + str(locs[ind][0] + locs[ind][2]) + '</xmax>\n')
  62.             xml.write('\t\t\t<ymax>' + str(locs[ind][1] + locs[ind][3]) + '</ymax>\n')
  63.             xml.write('\t\t</bndbox>\n')
  64.             xml.write('\t</object>\n')
  65.  
  66.         xml.write('</annotation>')
  67.     xml.close()
  68.     print("The {} accomplish!".format(filePath))
  69.  
  70.  
  71. def read_line(file_name):
  72.     """
  73.     read file by line
  74.     :param file_name:
  75.     :return:
  76.     """
  77.     lines = []
  78.     file = open(file_name)
  79.     for line in file:
  80.         line = line.strip('\n')
  81.         lines.append(line)
  82.     file.close()
  83.     return lines
  84.  
  85.  
  86. def showAnnotions(image, locs, labels):
  87.     """
  88.     show annotions
  89.     :param locs:
  90.     :param labels:
  91.     :return:
  92.     """
  93.     for ind, (x, y, w, h) in enumerate(locs):
  94.         cv2.rectangle(image, (x, y), (x + w, y + h), (0, 0, 255), 2)
  95.         cv2.putText(image, labels[ind], (x, y), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 1)
  96.  
  97.     cv2.imshow("iamge", image)
  98.     cv2.waitKey(30)
  99.  
  100.  
  101. def yolo2voc(txt_path, image_path, save_xml_path, label_path):
  102.     """
  104.     Args:
  105.         txt_path:
  106.         image_path:
  107.         save_xml_path:
  108.         label_path:
  110.     Returns:
  112.     """
  113.     txt_list = os.listdir(txt_path)
  114.     label_list = read_line(label_path)
  115.     for txt_name in txt_list:
  116.         image_name, _ = os.path.splitext(txt_name)
  117.         for extension in ["png", "jpg"]:  # read image
  118.             image_name_path = image_name + '.' + extension
  119.             image_name_path = os.path.join(image_path, image_name_path)
  120.             if os.path.exists(image_name_path):
  121.                 image = cv2.imread(image_name_path)
  122.                 image_h, image_w, image_channel = image.shape
  123.                 break
  124.             else:
  125.                 continue
  126.  
  127.         txt_name_path = os.path.join(txt_path, txt_name)
  128.         lines = read_line(txt_name_path)
  129.         object_locs = []
  130.         object_labels = []
  131.         for line in lines:
  132.             ind, center_x, center_y, w, h = line.split(' ')
  133.             x = (float(center_x) - float(w) / 2) * image_w
  134.             y = (float(center_y) - float(h) / 2) * image_h
  135.             w = float(w) * image_w
  136.             h = float(h) * image_h
  137.             object_locs.append([int(x), int(y), int(w), int(h)])
  138.             object_labels.append(label_list[int(ind)])
  139.  
  140.         # showAnnotions(image, object_locs, object_labels)
  141.         xml_name_path = os.path.join(save_xml_path, image_name + ".xml")
  142.         vocXMLFormat(xml_name_path, os.path.split(image_name_path)[-1], image_w, image_h, object_labels, object_locs,
  143.                      image_channel)
  144.  
  145.  
  146. def fractionalSet(imagePath, txtSavePath):
  147.     """
  149.     Args:
  150.         imagePath:
  151.         txtSavePath:
  153.     Returns:
  155.     """
  156.     testPercent = 0.1
  157.     valPercent = 0.1
  158.     trainPercent = 0.8
  159.  
  160.     imageNames = os.listdir(imagePath)
  161.     testNum = int(len(imageNames) * testPercent)
  162.     print("testNum:", testNum)
  163.     valNum = int(len(imageNames) * valPercent)
  164.     print("valNum:", valNum)
  165.     trainNum = len(imageNames) - testNum - valNum
  166.     print("trainNum", trainNum)
  167.     trainValNum = trainNum + valNum
  168.     print("trainValNum:", trainValNum)
  169.  
  170.     test = random.sample(imageNames, testNum)
  171.     imageNamesTrainVal = list(set(imageNames) - set(test))
  172.     val = random.sample(imageNamesTrainVal, valNum)
  173.     train = list(set(imageNamesTrainVal) - set(val))
  174.  
  175.     ftrainval_path = os.path.join(txtSavePath, "trainval.txt")
  176.     ftest_path = os.path.join(txtSavePath, "test.txt")
  177.     ftrain_path = os.path.join(txtSavePath, "train.txt")
  178.     fval_path = os.path.join(txtSavePath, "val.txt")
  179.  
  180.     ftrainval = open(ftrainval_path, 'w')
  181.     ftest = open(ftest_path, 'w')
  182.     ftrain = open(ftrain_path, 'w')
  183.     fval = open(fval_path, 'w')
  184.  
  185.     for imageName in imageNames:
  186.         if imageName in test:
  187.             imageName = imageName.split('.')[0] + "\n"
  188.             ftest.write(imageName)
  189.         else:
  190.             if imageName in val:
  191.                 imageName = imageName.split('.')[0] + "\n"
  192.                 fval.write(imageName)
  193.             else:
  194.                 imageName = imageName.split('.')[0] + "\n"
  195.                 ftrain.write(imageName)
  196.             ftrainval.write(imageName)
  197.  
  198.     ftrainval.close()
  199.     ftrain.close()
  200.     fval.close()
  201.     ftest.close()
  202.  
  203.  
  204. def convert(size, box):
  205.     """
  206.     ·µ»Ø¹éÒ»»¯ºóµÄÖÐÐĵ㣬¿í¸ßÖµ
  207.     :param size:
  208.     :param box:
  209.     :return:
  210.     """
  211.     dw = 1./(size[0])
  212.     dh = 1./(size[1])
  213.     x = (box[0] + box[1])/2.0 - 1
  214.     y = (box[2] + box[3])/2.0 - 1
  215.     w = box[1] - box[0]
  216.     h = box[3] - box[2]
  217.     x = x*dw
  218.     w = w*dw
  219.     y = y*dh
  220.     h = h*dh
  221.     return (x,y,w,h)
  222.  
  223.  
  224. def count_label_num(xml_path):
  225.     """
  227.     Args:
  228.         xml_path:
  230.     Returns:
  232.     """
  233.     classes = []
  234.     num = []
  235.  
  236.     xml_list = os.listdir(xml_path)
  237.     for xml_name in xml_list:
  238.         xml_name_path = os.path.join(xml_path, xml_name)
  239.  
  240.         et = ET.parse(xml_name_path)
  241.         root = et.getroot()
  242.         for obj in root.iter('object'):
  243.             cls = obj.find("name").text
  244.  
  245.             try:
  246.                 cls_id = classes.index(cls)
  247.             except ValueError:
  248.                 classes.append(cls)
  249.                 num.append(0)
  250.                 print("add new label:{}".format(cls))
  251.                 cls_id = classes.index(cls)
  252.  
  253.             num[cls_id] += 1
  254.  
  255.     return classes, num
  256.  
  257.  
  258. def checkLabel(imagePath, labelPath):
  259.     """
  260.     Ëæ»ú³é²é¼ì²é±ê×¢Êý¾ÝÊÇ·ñÕýÈ·
  261.     :param imagePath:
  262.     :param labelPath:
  263.     :return:
  264.     """
  265.     labelNames = os.listdir(labelPath)
  266.     while True:
  267.         index = random.randint(0, len(labelNames) - 1)
  268.         labelPath_tmp = os.path.join(labelPath, labelNames[index])
  269.  
  270.         tree = ET.parse(labelPath_tmp)
  271.         root = tree.getroot()
  272.         imageName = root.find('filename').text
  273.         labels = []   # ±êÇ©ÐÅÏ¢[[x,y,w,h,label]]
  274.         for obj in root.iter('object'):
  275.             x1 = int(obj.find("bndbox").find("xmin").text)
  276.             y1 = int(obj.find("bndbox").find("ymin").text)
  277.             x2 = int(obj.find("bndbox").find("xmax").text)
  278.             y2 = int(obj.find("bndbox").find("ymax").text)
  279.             name = obj.find("name").text
  280.             labels.append([x1, y1, x2, y2, name])
  281.  
  282.         imagePath_tmp = os.path.join(imagePath, imageName)
  283.         img = cv2.imread(imagePath_tmp)
  284.         for label in labels:
  285.             cv2.rectangle(img, (label[0], label[1]), (label[2], label[3]), (0, 0, 255))
  286.             cv2.putText(img, label[-1], (label[0], label[1]), cv2.FONT_HERSHEY_SIMPLEX, 0.8, (255, 0, 0), 2)
  287.         cv2.imshow(imageName, img)
  288.         k = cv2.waitKey(0)
  289.  
  290.         # °´Esc¼üÍ˳ö
  291.         if k == 27:
  292.             break
  293.         # °´ÓÒ¼ýÍ·ºÍϼýÍ·¼ÌÐø¼ì²â
  294.         if k != 1 and k != 2:
  295.             if k == 27:
  296.                 break
  297.             k = cv2.waitKey(0)
  298.         if k == 27:
  299.             break
  300.         cv2.destroyAllWindows()
  301.  
  302.  
  303.  
  304. txt_path = "/home/zyy/桌面/label"
  305. image_path = "./defect4/JPEGImages"
  306. save_xml_path = "./defect4/Annotations"
  307. image_sets = "./defect4/ImageSets/Main"
  308. label_path = "./defect4/label.txt"
  309. setNames = ["test.txt", "val.txt", "train.txt", "trainval.txt"]
  310. def main():
  311.     # yolo2voc(txt_path, image_path, save_xml_path, label_path)
  312.     # fractionalSet(image_path, image_sets)
  313.  
  314.     # classes, num = count_label_num(save_xml_path)
  315.     # print(classes)
  316.     # print(num)
  317.  
  318.     checkLabel(image_path, save_xml_path)
  319.  
  320.  
  321. if __name__ == "__main__":
  322.     main()

voc转coco

  1. import xml.etree.ElementTree as ET
  2. import os
  3. import json
  4.  
  5. coco = dict()
  6. coco['images'] = []
  7. coco['type'] = 'instances'
  8. coco['annotations'] = []
  9. coco['categories'] = []
  10.  
  11. category_set = dict()
  12. image_set = set()
  13.  
  14. category_item_id = -1
  15. image_id = 20180000000
  16. annotation_id = 0
  17.  
  18.  
  19. def addCatItem(name):
  20.     global category_item_id
  21.     category_item = dict()
  22.     category_item['supercategory'] = 'none'
  23.     category_item_id += 1
  24.     category_item['id'] = category_item_id
  25.     category_item['name'] = name
  26.     coco['categories'].append(category_item)
  27.     category_set[name] = category_item_id
  28.     return category_item_id
  29.  
  30.  
  31. def addImgItem(file_name, size):
  32.     global image_id
  33.     if file_name is None:
  34.         raise Exception('Could not find filename tag in xml file.')
  35.     if size['width'] is None:
  36.         raise Exception('Could not find width tag in xml file.')
  37.     if size['height'] is None:
  38.         raise Exception('Could not find height tag in xml file.')
  39.     image_id += 1
  40.     image_item = dict()
  41.     image_item['id'] = image_id
  42.     image_item['file_name'] = file_name
  43.     image_item['width'] = size['width']
  44.     image_item['height'] = size['height']
  45.     coco['images'].append(image_item)
  46.     image_set.add(file_name)
  47.     return image_id
  48.  
  49.  
  50. def addAnnoItem(object_name, image_id, category_id, bbox):
  51.     global annotation_id
  52.     annotation_item = dict()
  53.     annotation_item['segmentation'] = []
  54.     seg = []
  55.     # bbox[] is x,y,w,h
  56.     # left_top
  57.     seg.append(bbox[0])
  58.     seg.append(bbox[1])
  59.     # left_bottom
  60.     seg.append(bbox[0])
  61.     seg.append(bbox[1] + bbox[3])
  62.     # right_bottom
  63.     seg.append(bbox[0] + bbox[2])
  64.     seg.append(bbox[1] + bbox[3])
  65.     # right_top
  66.     seg.append(bbox[0] + bbox[2])
  67.     seg.append(bbox[1])
  68.  
  69.     annotation_item['segmentation'].append(seg)
  70.  
  71.     annotation_item['area'] = bbox[2] * bbox[3]
  72.     annotation_item['iscrowd'] = 0
  73.     annotation_item['ignore'] = 0
  74.     annotation_item['image_id'] = image_id
  75.     annotation_item['bbox'] = bbox
  76.     annotation_item['category_id'] = category_id
  77.     annotation_id += 1
  78.     annotation_item['id'] = annotation_id
  79.     coco['annotations'].append(annotation_item)
  80.  
  81.  
  82. def _read_image_ids(image_sets_file):
  83.     ids = []
  84.     with open(image_sets_file) as f:
  85.         for line in f:
  86.             ids.append(line.rstrip())
  87.     return ids
  88.  
  89.  
  90. def parseXmlFiles_by_txt(data_dir, json_save_path, split='train'):
  91.     """
  93.     Args:
  94.         data_dir:
  95.         json_save_path:
  96.         split: 'train' 'trainval' 'test'
  98.     Returns:
  100.     """
  101.     labelfile = split + ".txt"
  102.     image_sets_file = data_dir + "/ImageSets/Main/" + labelfile
  103.     ids = _read_image_ids(image_sets_file)
  104.  
  105.     for _id in ids:
  106.         xml_file = data_dir + f"/Annotations/{_id}.xml"
  107.  
  108.         bndbox = dict()
  109.         size = dict()
  110.         current_image_id = None
  111.         current_category_id = None
  112.         file_name = None
  113.         size['width'] = None
  114.         size['height'] = None
  115.         size['depth'] = None
  116.  
  117.         tree = ET.parse(xml_file)
  118.         root = tree.getroot()
  119.         if root.tag != 'annotation':
  120.             raise Exception('pascal voc xml root element should be annotation, rather than {}'.format(root.tag))
  121.  
  122.         # elem is <folder>, <filename>, <size>, <object>
  123.         for elem in root:
  124.             current_parent = elem.tag
  125.             current_sub = None
  126.             object_name = None
  127.  
  128.             if elem.tag == 'folder':
  129.                 continue
  130.  
  131.             if elem.tag == 'filename':
  132.                 file_name = elem.text
  133.                 if file_name in category_set:
  134.                     raise Exception('file_name duplicated')
  135.  
  136.             # add img item only after parse <size> tag
  137.             elif current_image_id is None and file_name is not None and size['width'] is not None:
  138.                 if file_name not in image_set:
  139.                     current_image_id = addImgItem(file_name, size)
  140.                     print('add image with {} and {}'.format(file_name, size))
  141.                 else:
  142.                     raise Exception('duplicated image: {}'.format(file_name))
  143.                     # subelem is <width>, <height>, <depth>, <name>, <bndbox>
  144.             for subelem in elem:
  145.                 bndbox['xmin'] = None
  146.                 bndbox['xmax'] = None
  147.                 bndbox['ymin'] = None
  148.                 bndbox['ymax'] = None
  149.  
  150.                 current_sub = subelem.tag
  151.                 if current_parent == 'object' and subelem.tag == 'name':
  152.                     object_name = subelem.text
  153.                     if object_name not in category_set:
  154.                         current_category_id = addCatItem(object_name)
  155.                     else:
  156.                         current_category_id = category_set[object_name]
  157.  
  158.                 elif current_parent == 'size':
  159.                     if size[subelem.tag] is not None:
  160.                         raise Exception('xml structure broken at size tag.')
  161.                     size[subelem.tag] = int(subelem.text)
  162.  
  163.                 # option is <xmin>, <ymin>, <xmax>, <ymax>, when subelem is <bndbox>
  164.                 for option in subelem:
  165.                     if current_sub == 'bndbox':
  166.                         if bndbox[option.tag] is not None:
  167.                             raise Exception('xml structure corrupted at bndbox tag.')
  168.                         bndbox[option.tag] = int(option.text)
  169.  
  170.                 # only after parse the <object> tag
  171.                 if bndbox['xmin'] is not None:
  172.                     if object_name is None:
  173.                         raise Exception('xml structure broken at bndbox tag')
  174.                     if current_image_id is None:
  175.                         raise Exception('xml structure broken at bndbox tag')
  176.                     if current_category_id is None:
  177.                         raise Exception('xml structure broken at bndbox tag')
  178.                     bbox = []
  179.                     # x
  180.                     bbox.append(bndbox['xmin'])
  181.                     # y
  182.                     bbox.append(bndbox['ymin'])
  183.                     # w
  184.                     bbox.append(bndbox['xmax'] - bndbox['xmin'])
  185.                     # h
  186.                     bbox.append(bndbox['ymax'] - bndbox['ymin'])
  187.                     print('add annotation with {},{},{},{}'.format(object_name, current_image_id, current_category_id,
  188.                                                                    bbox))
  189.                     addAnnoItem(object_name, current_image_id, current_category_id, bbox)
  190.     json.dump(coco, open(json_save_path, 'w'))
  191.  
  192.  
  193. """Ö±½Ó´ÓxmlÎļþ¼ÐÖÐÉú³É"""
  194.  
  195.  
  196. def parseXmlFiles(xml_path, json_save_path):
  197.     for f in os.listdir(xml_path):
  198.         if not f.endswith('.xml'):
  199.             continue
  200.  
  201.         bndbox = dict()
  202.         size = dict()
  203.         current_image_id = None
  204.         current_category_id = None
  205.         file_name = None
  206.         size['width'] = None
  207.         size['height'] = None
  208.         size['depth'] = None
  209.  
  210.         xml_file = os.path.join(xml_path, f)
  211.         print(xml_file)
  212.  
  213.         tree = ET.parse(xml_file)
  214.         root = tree.getroot()
  215.         if root.tag != 'annotation':
  216.             raise Exception('pascal voc xml root element should be annotation, rather than {}'.format(root.tag))
  217.  
  218.         # elem is <folder>, <filename>, <size>, <object>
  219.         for elem in root:
  220.             current_parent = elem.tag
  221.             current_sub = None
  222.             object_name = None
  223.  
  224.             if elem.tag == 'folder':
  225.                 continue
  226.  
  227.             if elem.tag == 'filename':
  228.                 file_name = elem.text
  229.                 if file_name in category_set:
  230.                     raise Exception('file_name duplicated')
  231.  
  232.             # add img item only after parse <size> tag
  233.             elif current_image_id is None and file_name is not None and size['width'] is not None:
  234.                 if file_name not in image_set:
  235.                     current_image_id = addImgItem(file_name, size)
  236.                     print('add image with {} and {}'.format(file_name, size))
  237.                 else:
  238.                     raise Exception('duplicated image: {}'.format(file_name))
  239.                     # subelem is <width>, <height>, <depth>, <name>, <bndbox>
  240.             for subelem in elem:
  241.                 bndbox['xmin'] = None
  242.                 bndbox['xmax'] = None
  243.                 bndbox['ymin'] = None
  244.                 bndbox['ymax'] = None
  245.  
  246.                 current_sub = subelem.tag
  247.                 if current_parent == 'object' and subelem.tag == 'name':
  248.                     object_name = subelem.text
  249.                     if object_name not in category_set:
  250.                         current_category_id = addCatItem(object_name)
  251.                     else:
  252.                         current_category_id = category_set[object_name]
  253.  
  254.                 elif current_parent == 'size':
  255.                     if size[subelem.tag] is not None:
  256.                         raise Exception('xml structure broken at size tag.')
  257.                     size[subelem.tag] = int(subelem.text)
  258.  
  259.                 # option is <xmin>, <ymin>, <xmax>, <ymax>, when subelem is <bndbox>
  260.                 for option in subelem:
  261.                     if current_sub == 'bndbox':
  262.                         if bndbox[option.tag] is not None:
  263.                             raise Exception('xml structure corrupted at bndbox tag.')
  264.                         bndbox[option.tag] = int(option.text)
  265.  
  266.                 # only after parse the <object> tag
  267.                 if bndbox['xmin'] is not None:
  268.                     if object_name is None:
  269.                         raise Exception('xml structure broken at bndbox tag')
  270.                     if current_image_id is None:
  271.                         raise Exception('xml structure broken at bndbox tag')
  272.                     if current_category_id is None:
  273.                         raise Exception('xml structure broken at bndbox tag')
  274.                     bbox = []
  275.                     # x
  276.                     bbox.append(bndbox['xmin'])
  277.                     # y
  278.                     bbox.append(bndbox['ymin'])
  279.                     # w
  280.                     bbox.append(bndbox['xmax'] - bndbox['xmin'])
  281.                     # h
  282.                     bbox.append(bndbox['ymax'] - bndbox['ymin'])
  283.                     print('add annotation with {},{},{},{}'.format(object_name, current_image_id, current_category_id,
  284.                                                                    bbox))
  285.                     addAnnoItem(object_name, current_image_id, current_category_id, bbox)
  286.     json.dump(coco, open(json_save_path, 'w'))
  287.  
  288.  
  289. if __name__ == '__main__':
  290.     # 下面三个test.json、train.json、val.json只能一个一个运行
  291.     voc_data_dir = "./defect4"
  292.     json_save_path = "./defect4/test.json"
  293.     parseXmlFiles_by_txt(voc_data_dir, json_save_path, "test")
  294.  
  295.     # json_save_path = "./defect4/train.json"
  296.     # parseXmlFiles_by_txt(voc_data_dir, json_save_path, "train")
  297.  
  298.     # json_save_path = "./defect4/val.json"
  299.     # parseXmlFiles_by_txt(voc_data_dir, json_save_path, "val")
  300.  
  301.     # ann_path = "E:/VOCdevkit/VOC2007/Annotations"
  302.     # json_save_path = "E:/VOCdevkit/test.json"
  303.     # parseXmlFiles(ann_path, json_save_path)

voc数据删除指定类别或更改类别名称

  1. #!/usr/bin/env python
  2. # -*- coding: utf-8 -*-
  3. # @Time        :2023/8/8 9:08
  4. # @Author      :weiz
  5. # @ProjectName :jiangling_new
  6. # @File        :xml_process.py
  7. # @Description :
  8. # Copyright (C) 2021-2025 Jiangxi Institute Of Intelligent Industry Technology Innovation
  9. import os
  10. import xml.dom.minidom
  11. import xml.etree.cElementTree as ET
  12.  
  13.  
  14. def delete_specified_cls(xmlFolderPath, xmlSavePath, delete_cls):
  15.     xml_list = os.listdir(xmlFolderPath)
  16.  
  17.     for xml_name in xml_list:
  18.         xml_tree = ET.parse(os.path.join(xmlFolderPath, xml_name))
  19.         xml_root = xml_tree.getroot()
  20.  
  21.         total_class_num = 0
  22.         class_del_num = 0
  23.  
  24.         for xml_child in xml_root.findall("object"):
  25.  
  26.             total_class_num = total_class_num + 1
  27.  
  28.             old_cls = xml_child.find("name").text
  29.             if old_cls in delete_cls:
  30.                 class_del_num = class_del_num + 1
  31.                 xml_root.remove(xml_child)
  32.         if total_class_num == class_del_num:
  33.             print("类别全部被删除了，请删除该文件及对应的图片：", xml_name)
  34.             xml_tree.write(os.path.join(xmlSavePath, xml_name))
  35.         else:
  36.             xml_tree.write(os.path.join(xmlSavePath, xml_name))
  37.             print(xml_name, "删除指定类别成功")
  38.  
  39.  
  40. def change_class_name(xmlFolderPath, xmlSavePath, old_class, new_class):
  41.     """
  42.     voc 数据类别替换
  43.     :param xmlFolderPath:
  44.     :param xmlSavePath:
  45.     :param old_class:
  46.     :param new_class:
  47.     :return:
  48.     """
  49.     xml_list = os.listdir(xmlFolderPath)
  50.  
  51.     for xml_name in xml_list:
  52.         xml_dom = xml.dom.minidom.parse(os.path.join(xmlFolderPath, xml_name))
  53.         root = xml_dom.documentElement
  54.         newfilename = root.getElementsByTagName('name')
  55.         for i, node in enumerate(newfilename):
  56.             for j, cls in enumerate(old_class):
  57.                 if node.firstChild.data == cls:
  58.                     newfilename[i].firstChild.data = new_class[j]
  59.  
  60.         with open(os.path.join(xmlSavePath, xml_name), 'w') as fh:
  61.             xml_dom.writexml(fh)
  62.             print(xml_name, "文件类别替换成功")
  63.  
  64.  
  65. g_delete_cls = ["spare_tire"]
  66. # g_old_class = ["EKS_18_19", "EKS_20寸", "743出口", "743国内", "743BEV", "756"]
  67. # g_new_class = ["NS1-1532-EA", "NS1-1352-DB", "MS1-1532-BA", "FS1-1532-AC", "KS2-1352-AA", "KS1-1532-AA"]
  68. # g_old_class = ["fuel_92", "fuel_91", "fuel_87", "gasolina", "fuel_91_ron"]
  69. # g_new_class = ["FS1-9A095-AB", "LK29-9A095-BA", "NS1-9A095-CA", "MS1-9A095-AA", "NS1-9A095-AA"]
  70. g_old_class = ["strchromium", "nonchromium"]
  71. g_new_class = ["KS1-D13210-BFD1", "KS1-D13210-ADD1"]
  72. g_xmlFolderPath = r"E:\江铃工业相机及手机采集数据\FL09\燃油标签\Annotations"
  73. g_xmlSavePath = r"E:\江铃工业相机及手机采集数据\FL09\燃油标签\Annotations"
  74. if __name__ == "__main__":
  75.     change_class_name(g_xmlFolderPath, g_xmlSavePath, g_old_class, g_new_class)
  76.  
  77.     #delete_specified_cls(g_xmlFolderPath, g_xmlSavePath, g_delete_cls)
  78.

声明：本文内容由网友自发贡献，不代表【wpsshop博客】立场，版权归原作者所有，本站不承担相应法律责任。如您发现有侵权的内容，请联系我们。转载请注明出处：https://www.wpsshop.cn/w/知新_RL/article/detail/869552
推荐阅读
相关标签

Copyright © 2003-2013 www.wpsshop.cn 版权所有，并保留所有权利。

  

闽ICP备14008679号