python中的txt文件转换为XML

txt文件转换为XML

很多目标检测的模型都是默认需要VOC的文件输入格式

手上数据label是txt文件。

为了避免不必要的bug，还是选择转换下格式

将数据按VOC形式放置

# -*- coding: utf-8 -*-

from xml.dom.minidom import Document
import os
import os.path
from PIL import Image
import importlib
import sys
importlib.reload(sys)

xml_path = "Annotations\\"
img_path = "JPEGImages\\"
ann_path = "Labelss\\"

if not os.path.exists(xml_path):
   os.mkdir(xml_path)

def writeXml(tmp, imgname, w, h, objbud, wxml):
   doc = Document()
   # owner
   annotation = doc.createElement('annotation')
   doc.appendChild(annotation)
   # owner
   folder = doc.createElement('folder')
   annotation.appendChild(folder)
   folder_txt = doc.createTextNode("VOC2007")
   folder.appendChild(folder_txt)

filename = doc.createElement('filename')
   annotation.appendChild(filename)
   filename_txt = doc.createTextNode(imgname)
   filename.appendChild(filename_txt)
   # ones#
   source = doc.createElement('source')
   annotation.appendChild(source)

database = doc.createElement('database')
   source.appendChild(database)
   database_txt = doc.createTextNode("The VOC2007 Database")
   database.appendChild(database_txt)

annotation_new = doc.createElement('annotation')
   source.appendChild(annotation_new)
   annotation_new_txt = doc.createTextNode("PASCAL VOC2007 ")
   annotation_new.appendChild(annotation_new_txt)

image = doc.createElement('image')
   source.appendChild(image)
   image_txt = doc.createTextNode("flickr")
   image.appendChild(image_txt)
   # onee#
   # twos#
   size = doc.createElement('size')
   annotation.appendChild(size)

width = doc.createElement('width')
   size.appendChild(width)
   width_txt = doc.createTextNode(str(w))
   width.appendChild(width_txt)

height = doc.createElement('height')
   size.appendChild(height)
   height_txt = doc.createTextNode(str(h))
   height.appendChild(height_txt)

depth = doc.createElement('depth')
   size.appendChild(depth)
   depth_txt = doc.createTextNode("3")
   depth.appendChild(depth_txt)
   # twoe#
   segmented = doc.createElement('segmented')
   annotation.appendChild(segmented)
   segmented_txt = doc.createTextNode("0")
   segmented.appendChild(segmented_txt)

# threes#
   object_new = doc.createElement("object")
   annotation.appendChild(object_new)

name = doc.createElement('name')
   object_new.appendChild(name)
   name_txt = doc.createTextNode('cancer')
   name.appendChild(name_txt)

pose = doc.createElement('pose')
   object_new.appendChild(pose)
   pose_txt = doc.createTextNode("Unspecified")
   pose.appendChild(pose_txt)

truncated = doc.createElement('truncated')
   object_new.appendChild(truncated)
   truncated_txt = doc.createTextNode("0")
   truncated.appendChild(truncated_txt)

difficult = doc.createElement('difficult')
   object_new.appendChild(difficult)
   difficult_txt = doc.createTextNode("0")
   difficult.appendChild(difficult_txt)
   # threes-1#
   bndbox = doc.createElement('bndbox')
   object_new.appendChild(bndbox)

xmin = doc.createElement('xmin')
   bndbox.appendChild(xmin)

#objbud存放[类别，xmin,ymin,xmax,ymax]
   xmin_txt = doc.createTextNode(objbud[1])
   xmin.appendChild(xmin_txt)

ymin = doc.createElement('ymin')
   bndbox.appendChild(ymin)
   ymin_txt = doc.createTextNode(objbud[2])
   ymin.appendChild(ymin_txt)

xmax = doc.createElement('xmax')
   bndbox.appendChild(xmax)
   xmax_txt = doc.createTextNode(objbud[3])
   xmax.appendChild(xmax_txt)

ymax = doc.createElement('ymax')
   bndbox.appendChild(ymax)
   ymax_txt = doc.createTextNode(objbud[4])
   ymax.appendChild(ymax_txt)
   # threee-1#
   # threee#

tempfile = tmp + "test.xml"
   with open(tempfile, "wb") as f:
       f.write(doc.toprettyxml(indent="\t", newl="\n", encoding="utf-8"))

rewrite = open(tempfile, "r")
   lines = rewrite.read().split('\n')
   newlines = lines[1:len(lines) - 1]

fw = open(wxml, "w")
   for i in range(0, len(newlines)):
       fw.write(newlines[i] + '\n')

fw.close()
   rewrite.close()
   os.remove(tempfile)
   return

for files in os.walk('E:\ssd_pytorch_cancer\data\cancer_or_not\Labels'):
   print(files)
   temp = "/temp/"
   if not os.path.exists(temp):
       os.mkdir(temp)
   for file in files[2]:
       print(file + "-->start!")
       img_name = os.path.splitext(file)[0] + '.jpg'
       fileimgpath = img_path + img_name
       im = Image.open(fileimgpath)
       width = int(im.size[0])
       height = int(im.size[1])

filelabel = open(ann_path + file, "r")
       lines = filelabel.read().split(' ')
       obj = lines[:len(lines)]

filename = xml_path + os.path.splitext(file)[0] + '.xml'
       writeXml(temp, img_name, width, height, obj, filename)
   os.rmdir(temp)

不过代码只使用于每个label文件只有一个标注框，可在生成bndbox节点处加入循环

来源：https://blog.csdn.net/weixin_43289424/article/details/106371995