xml转matlab目标检测,将Cityscape转换为PASACAL VOC格式的目标检测数据集

1、将Cityscape中的json格式的标注转换为.txt格式的标签

# convert cityscape dataset to pascal voc format dataset

# 1. convert every cityscape image label '.json' to '.txt'

import json

import os

from os import listdir, getcwd

from os.path import join

import os.path

rootdir = 'D:\dataset\cityscapes\leftImg8bit\\train\\zurich' # 写自己存放图片的数据地址

def position(pos):

# 该函数用来找出xmin,ymin,xmax,ymax即bbox包围框

x = []

y = []

nums = len(pos)

for i in range(nums):

x.append(pos[i][0])

y.append(pos[i][1])

x_max = max(x)

x_min = min(x)

y_max = max(y)

y_min = min(y)

# print(x_max,y_max,x_min,y_min)

b = (float(x_min), float(y_min), float(x_max), float(y_max))

# print(b)

return b

# pascal voc 标准格式

# < xmin > 174 < / xmin >

# < ymin > 101 < / ymin >

# < xmax > 349 < / xmax >

# < ymax > 351 < / ymax >

def convert(size, box):

# 该函数将xmin,ymin,xmax,ymax转为x,y,w,h中心点坐标和宽高

dw = 1. / (size[0])

dh = 1. / (size[1])

x = (box[0] + box[1]) / 2.0 - 1

y = (box[2] + box[3]) / 2.0 - 1

w = box[1] - box[0]

h = box[3] - box[2]

x = x * dw

w = w * dw

y = y * dh

h = h * dh

print((x, y, w, h))

return (x, y, w, h)

def convert_annotation(image_id):

# load_f = open("/home/ubuntu/PycharmProjects/city2pascal/source/train/tubingen/%s_gtFine_polygons.json" % (image_id), 'r') # 导入json标签的地址

load_f = open("D:\dataset\cityscapes\gtFine\\train\\zurich\%s_gtFine_polygons.json" % (image_id), 'r') # 导入json标签的地址

load_dict = json.load(load_f)

out_file = open('D:\dataset\cityscapes\gtFine\\train\\zurich\%s_leftImg8bit.txt' % (image_id), 'w') # 输出标签的地址

# keys=tuple(load_dict.keys())

w = load_dict['imgWidth'] # 原图的宽，用于归一化

h = load_dict['imgHeight']

# print(h)

objects = load_dict['objects']

nums = len(objects)

# print(nums)

# object_key=tuple(objects.keys()

cls_id = ''

for i in range(0, nums):

labels = objects[i]['label']

# print(i)

if (labels in ['person', 'rider']):

# print(labels)

pos = objects[i]['polygon']

bb = position(pos)

# bb = convert((w, h), b)

cls_id = 'pedestrian' # 我这里把行人和骑自行车的人都设为类别pedestrian

out_file.write(cls_id + " " + " ".join([str(a) for a in bb]) + '\n')

# print(type(pos))

elif (labels in ['car', 'truck', 'bus', 'caravan', 'trailer']):

# print(labels)

pos = objects[i]['polygon']

bb = position(pos)

# bb = convert((w, h), b)

cls_id = 'car' # 我这里把各种类型的车都设为类别car

out_file.write(cls_id + " " + " ".join([str(a) for a in bb]) + '\n')

if cls_id == '':

print('no label json:',"%s_gtFine_polygons.json" % (image_id))

def image_id(rootdir):

a = []

for parent, dirnames, filenames in os.walk(rootdir):

for filename in filenames:

# print(filename)

filename = filename[:-16]

# filename = filename.strip('_leftImg8bit.png')

a.append(filename)

return a

if __name__ == '__main__':

names = image_id(rootdir)

for image_id in names:

print(image_id)

convert_annotation(image_id)

2、将.txt转换为.xml的标签

#! /usr/bin/python

# -*- coding:UTF-8 -*-

# Convert cityscape dataset to pascal voc format dataset

# 2. convert '.txt' to '.xml'

import os, sys

import glob

from PIL import Image

# VEDAI 图像存储位置

src_img_dir = "D:\dataset\cityscapes\leftImg8bit\\train\\zurich\\"

# VEDAI 图像的 ground truth 的 txt 文件存放位置

src_txt_dir = "D:\dataset\cityscapes\gtFine\\train\\zurich\\"

src_xml_dir = "D:\dataset\cityscapes\gtFine\\train\\zurich\\"

img_Lists = glob.glob(src_img_dir + '/*.png')

img_basenames = [] # e.g. 100.jpg

for item in img_Lists:

img_basenames.append(os.path.basename(item))

img_names = [] # e.g. 100

for item in img_basenames:

temp1, temp2 = os.path.splitext(item)

img_names.append(temp1)

for img in img_names:

im = Image.open((src_img_dir + img + '.png'))

width, height = im.size

# open the crospronding txt file

gt = open(src_txt_dir + '/' + img + '.txt').read().splitlines()

# gt = open(src_txt_dir + '/gt_' + img + '.txt').read().splitlines()

# write in xml file

# os.mknod(src_xml_dir + '/' + img + '.xml')

xml_file = open((src_xml_dir + '/' + img + '.xml'), 'w')

xml_file.write('\n')

xml_file.write(' CITYSCAPE\n')

xml_file.write(' ' + str(img) + '.png' + '\n')

xml_file.write(' \n')

xml_file.write(' ' + str(width) + '\n')

xml_file.write(' ' + str(height) + '\n')

xml_file.write(' 3\n')

xml_file.write(' \n')

# write the region of image on xml file

for img_each_label in gt:

spt = img_each_label.split(' ') # 这里如果txt里面是以逗号‘，’隔开的，那么就改为spt = img_each_label.split(',')。

xml_file.write(' \n')

xml_file.write(' ' + str(spt[0]) + '\n')

xml_file.write(' Unspecified\n')

xml_file.write(' 0\n')

xml_file.write(' 0\n')

xml_file.write(' \n')

xml_file.write(' ' + str(spt[1]) + '\n')

xml_file.write(' ' + str(spt[2]) + '\n')

xml_file.write(' ' + str(spt[3]) + '\n')

xml_file.write(' ' + str(spt[4]) + '\n')

xml_file.write(' \n')

xml_file.write(' \n')

xml_file.write('')

承接Matlab、Python和C++的编程，机器学习、计算机视觉的理论实现及辅导，本科和硕士的均可，咸鱼交易，专业回答请走知乎，详谈请联系QQ号757160542，非诚勿扰。

本文同步分享在 博客“于小勇”(CSDN)。

如有侵权，请联系 support@oschina.cn 删除。

本文参与“OSC源创计划”，欢迎正在阅读的你也加入，一起分享。