博主在制作目标检测数据集时,需要将之前的标签进行替换,将person替换为rider,使用python批量处理xml格式文件如下:
import os, sys
import glob
from xml.etree import ElementTree as ET
#批量读取Annotations下的xml文件
#per=ET.parse(r'C:\Users\rockhuang\Desktop\Annotations\000003.xml')
xml_dir=r'C:\Users\rockhuang\Desktop\Annotations'
xml_list= glob.glob(xml_dir + '/*.xml')
for xml in xml_list:
print(xml)
per=ET.parse(xml)
p=per.findall('/object')
for oneper in p: #找出person节点
child=oneper.getchildren()[0] #找出person节点的子节点
if child.text=='person':
child.text='rider'
per.write(xml)
print(child.tag,':',child.text)
修改之前的xml文件为:
修改之后:
需要将 xmin,ymin,xmax,ymax坐标和width,height统一缩小
# coding=utf-8
# 读xml文件中的一个rect
import xml.etree.ElementTree as ET
import os, sys
import glob
xml_dir=r'C:\Users\rockhuang\Desktop\rider_xml'
xml_list= glob.glob(xml_dir + '\*.xml')
for xml in xml_list:
tree = ET.parse(xml)
root = tree.getroot()
for size in root.findall('size'):
for width in size.iter('width'):
width.text=str(round(int(width.text)/4))
# print(width.text)
for height in size.iter('height'):
height.text=str(round(int(height.text)/4))
# print(height.text)
for ob in root.findall('object'):
for bndbox in ob.iter('bndbox'):
# for l in bndbox:
# print(l.text)
for xmin in bndbox.iter('xmin'):
xmin.text= str(round(int(xmin.text)/4))
for ymin in bndbox.iter('ymin'):
ymin.text =str(round(int(ymin.text)/4))
for xmax in bndbox.iter('xmax'):
xmax.text =str(round(int(xmax.text)/4))
for ymax in bndbox.iter('ymax'):
ymax.text = str(round(int(ymax.text)/4))
#print(xmin.text,ymin.text,xmax.text,ymax.text)
tree.write(xml)