python实现tsinghua-tencent 100000数据集的json文件转为xml文件

算法解决问题：

1，从json文件中抽取数据集中的小目标pne、p11、i5、w57四种交通标志数据集

2.小目标判断标注像素范围在[32,32]

3.转为对应的xml文件

4.标出选出的类别的个数

import os
import time
import json
import cv2xml_head = '''<annotation><folder>traffic-sign</folder><!--文件名--><filename>{}</filename>  <size><width>2048</width><height>2048</height><depth>3</depth></size><segmented>0</segmented><category>{}</category>'''
xml_obj = '''<object>     <!--图片中目标类别-->   <name>{}</name><!--是否被裁减，0表示完整，1表示不完整--><truncated>0</truncated><!--是否容易识别，0表示容易，1表示困难--><difficult>0</difficult><bndbox><xmin>{}</xmin><ymin>{}</ymin><xmax>{}</xmax><ymax>{}</ymax></bndbox></object>'''
xml_end = '''
</annotation>'''
# json文件路径和用于存放xml文件的路径
anno = '/home/xiaobumidm/Yolo_mark-master/annotations.json'
xml_dir = '/home/xiaobumidm/Yolo_mark-master/jsonxml/'
labels = ['pne','p11','i5','w57']  # label for datasets
print("labels类型：",type(labels))
# 读取json文件内容,返回字典格式
with open(anno,'r',encoding='utf8')as fp:json_data = json.load(fp)#imgs数据，字典类型json_imgs=json_data['imgs']#记录小目标总数目,共2908count=0#count_pne的小目标数目count_pne=0#count_p11的小目标数目count_p11=0#count_i5的小目标数目count_i5=0#count_w57的小目标数目count_w57=0#count_pne_trian训练文件夹中pne数量count_pne_trian=0# count_p11_trian训练文件夹中p11数量count_p11_trian = 0# count_i5_trian训练文件夹中i5数量count_i5_trian = 0# count_w57_trian训练文件夹中w57数量count_w57_trian = 0#训练pne数量count_img_pne_train=0#训练p11数量count_img_p11_train=0#训练i5数量count_img_i5_train=0#训练w57数量count_img_w57_train=0# 测试pne数量count_img_pne_test=0# 测试p11数量count_img_p11_test=0# 测试i5数量count_img_i5_test=0# 测试w57数量count_img_w57_test=0#图片总数count_image=0;table=['pne']for item in json_imgs.items():#item是tuple类型 ,item元组中共有两个元素，一开始的ID和图片详细信息#detail_info为图片详细信息，是字典类型，含有path，objects，id三个key值detail_info=item[1]#path图片路径image_path=detail_info['path']if image_path.find('train')!=-1 or image_path.find('test')!=-1:#图片含有的目标,list类型image_objects=detail_info['objects']# 图片的idimage_id = detail_info['id']# flag_category记录category的标记flag_category = ''#obj表示一个图片中目标信息obj=''for info in range(len(image_objects)):# object_info是object的详细信息,数据类型为字典，包含category，objectsobject_info = image_objects[info]#object_category是object的category,字符串object_category = object_info['category']# object_bbox为字典类型，存放目标信息xmin,ymin,xmax,ymaxobject_bbox = object_info['bbox']#x_rang目标的x间的大小，xmax-xminx_rang=object_bbox['xmax']-object_bbox['xmin']#目标的y间的大小，ymax-yminy_rang=object_bbox['ymax']-object_bbox['ymin']#flag=0if x_rang <= 32 and y_rang <= 32:if object_category == 'pne':if flag_category == '':flag_category = object_categoryobj+=xml_obj.format(labels[0],str(object_bbox['xmin']),str(object_bbox['ymin']),str(object_bbox['xmax']),str(object_bbox['ymax']))if image_path.find('train') != -1:count_pne_trian+=1count_pne += 1count += 1if object_category == 'p11':if flag_category == '':flag_category = object_categoryobj += xml_obj.format(labels[1], str(object_bbox['xmin']),str(object_bbox['ymin']),str(object_bbox['xmax']), str(object_bbox['ymax']))if image_path.find('train') != -1:count_p11_trian+=1count_p11 += 1count += 1if object_category == 'i5':if flag_category == '':flag_category = object_categoryobj += xml_obj.format(labels[2], str(object_bbox['xmin']),str(object_bbox['ymin']),str(object_bbox['xmax']), str(object_bbox['ymax']))if image_path.find('train') != -1:count_i5_trian+=1count_i5 += 1count += 1if object_category == 'w57':if flag_category == '':flag_category = object_categoryobj += xml_obj.format(labels[3], str(object_bbox['xmin']),str(object_bbox['ymin']),str(object_bbox['xmax']), str(object_bbox['ymax']))if image_path.find('train') != -1:count_w57_trian+=1count_w57 += 1count += 1print('x_rang:', x_rang)print('y_range:', y_rang)#表示有合适类别的小目标的时候开始创建xml文件,存放在了test和train文件夹下面if flag_category!='' :if image_path.find('train') != -1:if flag_category=='pne':count_img_pne_train+=1xml_name = os.path.join(xml_dir + 'trains/pne/', str(image_id) + '.xml')if flag_category=='p11':count_img_p11_train+=1xml_name = os.path.join(xml_dir + 'trains/p11/', str(image_id) + '.xml')if flag_category=='i5':count_img_i5_train+=1xml_name = os.path.join(xml_dir + 'trains/i5/', str(image_id) + '.xml')if flag_category=='w57':count_img_w57_train+=1xml_name = os.path.join(xml_dir + 'trains/w57/', str(image_id) + '.xml')else :if flag_category=='pne':count_img_pne_test+=1xml_name = os.path.join(xml_dir + 'tests/pne/', str(image_id) + '.xml')if flag_category=='p11':count_img_p11_test+=1xml_name = os.path.join(xml_dir + 'tests/p11/', str(image_id) + '.xml')if flag_category=='i5':count_img_i5_test+=1xml_name = os.path.join(xml_dir + 'tests/i5/', str(image_id) + '.xml')if flag_category=='w57':count_img_w57_test+=1xml_name = os.path.join(xml_dir + 'tests/w57/', str(image_id) + '.xml')count_image+=1print("xml_name:"+xml_name)print("image_id:",image_id)#xml的头部信息传入参数值，文件名字和类别head=xml_head.format(str(xml_name),str(flag_category))#xml的尾部信息end=xml_end#向xml文件中写入内容with open(xml_name, 'w') as f:f.write(head+obj+end)print('count:', count)print('count_pne:', count_pne)print('count_pne_trian:',count_pne_trian)print('count_pne_test:',count_pne - count_pne_trian)print('count_p11:', count_p11)print('count_p11_trian:', count_p11_trian)print('count_p11_test:', count_p11 - count_p11_trian)print('count_i5:', count_i5)print('count_i5_trian:', count_i5_trian)print('count_i5_test:', count_i5 - count_i5_trian)print('count_w57:', count_w57)print('count_w57_trian:', count_w57_trian)print('count_w57e_test:', count_w57 - count_w57_trian)print('count_img_pne_train:',count_img_pne_train)print('count_img_p11_train:',count_img_p11_train)print('count_img_i5_train:',count_img_i5_train)print('count_img_w57_train:',count_img_w57_train)print('count_img_pne_test:',count_img_pne_test)print('count_img_p11_test:',count_img_p11_test)print('count_img_i5_test:',count_img_i5_test)print('count_img_w57_test:',count_img_w57_test)print('count_image:', count_image)

count值：

python实现tsinghua-tencent 100000数据集的json文件转为xml文件相关推荐

将xls表格文件转为xml文件以及json格式的文件，支持2010xls解析
项目需求,要通过数据库将数据以xls文件格式导出.直接解析xls的文件是没法解析的,所以需要通过解析成xml或者json格式文件后进一步处理.2003版本的Excel有65536的限制.搞了个java ...
python批量json文件转xml文件脚本(附代码)
场景:在使用了mask rcnn跑实验后标注了大量地json格式文件,现在打算使用yolo和faster rcnn 跑实验所以需要将之前地json文件转为xml 但是找了很久,没发现有批量处 ...
python转csv_python脚本如何将Excel文件转为csv文件（代码）
本篇文章给大家带来的内容是关于python脚本如何将Excel文件转为csv文件(代码),有一定的参考价值,有需要的朋友可以参考一下,希望对你有所帮助.#!/usr/bin/env python __ ...
python csv转excel_将Excel文件转为csv文件的python脚本
将Excel文件转为csv文件的python脚本 #!/usr/bin/env python __author__ = "lrtao2010" ''' Excel文件转csv文件脚 ...
比赛中学习（2）-voc数据集制作中将.txt文件变为.xml文件
描述和voc数据集制作有一点不同的是,因为参加的比赛中坐标给了4个,也就是说目标的形状可能是是平行四边形.所以和voc数据集有点不同啊.放出一张.txt文件的形式最后的结果大致是这样的(和上面的图 ...
Python 批量将.xlsx文件转为.xls文件
工作需要,需要向xls文件里写入数据.由于xlrd\xlwt写入的数据不带格式,只好把xls模板先转为xlsx格式,用openpyxl批量写入完后再批量转为xls. 百度找的转换实例,实际应用时生成的 ...
BDD100K数据集的Jason文件转xml文件
针对交通灯数据集,因为其包含其它标签,而我这里只需要标签为traffic light的一共是70000个jason文件对应的路径文件应该这样,每个人存放的位置不一致,所以自行设置就行文件结构如图 ...
python中dom模块_Python中使用dom模块生成XML文件示例
在Python中解析XML文件也有Dom和Sax两种方式,这里先介绍如何是使用Dom解析XML,这一篇文章是Dom生成XML文件,下一篇文章再继续介绍Dom解析XML文件. 在生成XML文件中,我们主 ...
python改文件名_python批量修改文件名、批量修改xml文件的path和filename
1. python批量修改文件名(控制文件名的起始编号) import os path = "E:/Project/label/fire/" # 目标路径 filename_lis ...

python实现tsinghua-tencent 100000数据集的json文件转为xml文件

python实现tsinghua-tencent 100000数据集的json文件转为xml文件相关推荐

最新文章

热门文章