对python修改xml文件的节点值方法详解

这是我的xml文件结构

<?xml version='1.0' encoding='utf-8'?>
<annotation>
<folder>JPEGImages</folder>
<filename>train_2018-05-08_1000.jpg</filename>
<path>D:\all_data\2018-05-08\JPEGImages\train_2018-05-08_1000.jpg</path>
<source>
<database>Unknown</database>
</source>
<size>
<width>4032</width>
<height>3024</height>
<depth>3</depth>
</size>
<segmented>0</segmented>
<object>
<name>yl-ylhzdhmbbz-gz-hm-280g</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>1863</xmin>
 <ymin>355</ymin>
 <xmax>2512</xmax>
 <ymax>902</ymax>
</bndbox>
</object>
<object>
<name>hy-hybfbgz-hz-xcw-200ml</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>1076</xmin>
 <ymin>1602</ymin>
 <xmax>1648</xmax>
 <ymax>2105</ymax>
</bndbox>
</object>
<object>
<name>ys-zzyspyz-gz-yw-245ml</name>
<pose>Unspecified</pose>
<truncated>1</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>2017</xmin>
 <ymin>2475</ymin>
 <xmax>2681</xmax>
 <ymax>3024</ymax>
</bndbox>
</object>
<object>
<name>mn-zgl-hz-cmw-250ml</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>1849</xmin>
 <ymin>1207</ymin>
 <xmax>2242</xmax>
 <ymax>2047</ymax>
</bndbox>
</object>
<object>
<name>qc-qckf-pz-shnt-268ml</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>480</xmin>
 <ymin>1213</ymin>
 <xmax>1308</xmax>
 <ymax>1544</ymax>
</bndbox>
</object>
<object>
<name>wt-wtcyl-gz-nm-310ml</name>
<pose>Unspecified</pose>
<truncated>0</truncated>
<difficult>0</difficult>
<bndbox>
 <xmin>867</xmin>
 <ymin>488</ymin>
 <xmax>1527</xmax>
 <ymax>938</ymax>
</bndbox>
</object>

</annotation>

现在想实现的是修改图像的size和目标

__author__ = 'Sam'

import cv2
import xml.etree.ElementTree as ET
import os
import sys
import lxml
import shutil

#user input files path
path="E:/test_folder"
image_path = path + "/Annotations/" #image path with .jpg ending
label_path = path + "/JPEGImages/" #label path with .xml ending
min_size=800

def search_jpg_xml(image_dir,label_dir):
 #find out all of sepecified file
 image_ext='.jpg'
 img=[fn for fn in os.listdir(image_dir) if fn.endswith(image_ext)]
 label_ext='.xml'
 label=[fn for fn in os.listdir(label_dir) if fn.endswith(label_ext)]
 return img, label

def copyfile():
 if "Annotations_temp" in os.listdir(path):
   shutil.rmtree(path+"/Annotations_temp")
 if "JPEGImages_temp" in os.listdir(path):
   shutil.rmtree(path+"/JPEGImages_temp")
 save_annotation_path=path+"/Annotations_temp/"
 save_jpg_path=path+"/JPEGImages_temp/"
 shutil.copytree(path + "/Annotations",save_annotation_path)
 shutil.copytree(path + "/JPEGImages", save_jpg_path)
 return save_jpg_path ,save_annotation_path

def write_xml_jpg(jpg_path,annotation_path):
 img,label=search_jpg_xml(jpg_path,annotation_path)
 sorted(img)
 sorted(label)
 print(img)
 print(label)
 if "Annotations_1" not in os.listdir(path):
   os.mkdir(path+"/Annotations_1")
 if "JPEGImages_1" not in os.listdir(path):
   os.mkdir(path+"/JPEGImages_1")
 new_image_path=path+"/JPEGImages_1/"
 new_annotation_path=path+"/Annotations_1/"
 for index,file in enumerate(label):
   cur_img = cv2.imread(jpg_path+img[index])
   width=cur_img.shape[1]
   height=cur_img.shape[0]
   if width<height:
     new_width=min_size
     new_height=int(min_size*height/width)
     w_ratio=new_width/width
     h_ratio=new_height/height
   elif width>height:
     new_width=int(min_size*width/height)
     new_height=min_size
     w_ratio=new_width/width
     h_ratio=new_height/height
   elif width==height:
     new_width=min_size
     new_height=min_size
     w_ratio=new_width/width
     h_ratio=new_height/height
   cur_img = cv2.resize(cur_img, (new_width, new_height))
   cv2.imwrite(new_image_path+img[index],cur_img)
   cur_xml = ET.parse(annotation_path+file)
   root = cur_xml.getroot()
   for node in root:
     if node.tag=='size':
       node[0].text=str(new_width)
       node[1].text=str(new_height)
     elif node.tag=='object':
        xmin=int(node[4][0].text)#bbox position
        ymin=int(node[4][1].text)
        xmax=int(node[4][2].text)
        ymax=int(node[4][3].text)
        node[4][0].text=str(int(xmin*w_ratio))
        node[4][1].text=str(int(ymin*h_ratio))
        node[4][2].text=str(int(xmax*w_ratio))
        node[4][3].text=str(int(ymax*h_ratio))
   cur_xml.write(new_annotation_path+file)
 shutil.rmtree(path+"/JPEGImages_temp")
 shutil.rmtree(path+"/Annotations_temp")

if __name__ == "__main__":
 jpg_path,annotation_path=copyfile()
 write_xml_jpg(jpg_path,annotation_path)

最关键语句是：

node[4][3].text=str(int(ymax*h_ratio))，注意xml节点的操作是字符型！！!

来源：https://blog.csdn.net/Touch_Dream/article/details/80435767