您好,登錄后才能下訂單哦!
這是我的xml文件結構
<?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>
現在想實現的是修改圖像的size和目標
__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)
最關鍵語句是:
node[4][3].text=str(int(ymax*h_ratio)),注意xml節點的操作是字符型!!!
以上這篇對python修改xml文件的節點值方法詳解就是小編分享給大家的全部內容了,希望能給大家一個參考,也希望大家多多支持億速云。
免責聲明:本站發布的內容(圖片、視頻和文字)以原創、轉載和分享為主,文章觀點不代表本網站立場,如果涉及侵權請聯系站長郵箱:is@yisu.com進行舉報,并提供相關證據,一經查實,將立刻刪除涉嫌侵權內容。