Citypersons数据集转VOC标准格式（YOLO 目标检测txt格式）

CItyscapes城市数据集包含一组不同的立体视频序列中记录来自50个不同城市的街景,高质量的进行像素级的注释。数据集下载地址（需要申请注册，通过申请才能下载）[https://www.cityscapes-dataset.com]
该数据集用于常语义分割，包含以下类别：

提供百度云链接地址：链接: https://pan.baidu.com/s/108_NgFheDIpnQRrwz5uhmw 提取码: dhr8

#cityperson数据集是cityscapes数据集的子集，cityperson的标注文件只标注了其中Human的类别，如上图。该标注文件下载地址为：

下载地址：https://bitbucket.org/shanshanzhang/citypersons/get/ae6814faa761.zip

提取cityscapes中标注好的类别为VOC的标准格式（JPEGImages和Annotations，txt的代码就不写了，参考博主其他博客）

#! /usr/bin/python
# -*- coding:UTF-8 -*-
import os, sys
import glob
from PIL import Image
import shutil
from scipy.io import loadmat
#img_Lists = glob.glob(src_img_dir + '\*.png')
 
# citypersons图像的标注位置
src_anno_dir = loadmat(r'c:\Users\rockhuang\Desktop\anno_train.mat')

# cityscapes图像的存储位置
src_img_dir = r"g:\dataset\cityscapes\leftImg8bit\train\\"

#保存为VOC 数据集的原图和xml标注路径
new_img= r"g:\dataset\cityscapes\JPEGImages"
new_xml=r"g:\dataset\cityscapes\Annotations"

if not os.path.isdir(new_img):
    os.makedirs(new_img)
    
if not os.path.isdir(new_xml):
    os.makedirs(new_xml)   
    
a=src_anno_dir['anno_train_aligned'][0]
 
    #处理标注文件

for i in range(len(a)):
    img_name=a[i][0][0][1][0]   #frankfurt_000000_000294_leftImg8bit.png     
    dir_name=img_name.split('_')[0]
    img=src_img_dir+dir_name+"\\"+img_name
     
    shutil.copy(img, new_img+"\\"+img_name)
    img=Image.open(img)
    width, height = img.size
    
    position=a[i][0][0][2]
    print(position)
    #sys.exit()
    xml_name=img_name.split('.')[0]
    xml_file = open((new_xml + '\\' + xml_name + '.xml'), 'w')
          
    xml_file.write('<annotation>\n')
    xml_file.write('    <folder>citysperson</folder>\n')
    xml_file.write('    <filename>' + str(img_name)+ '</filename>\n')
    xml_file.write('    <size>\n')
    xml_file.write('        <width>' + str(width) + '</width>\n')
    xml_file.write('        <height>' + str(height) + '</height>\n')
    xml_file.write('        <depth>3</depth>\n')
    xml_file.write('    </size>\n')
    
    for j in range(len(position)):
        category_location=position[j]  #[    1   947   406    17    40 24000   950   407    14    39]
        category=category_location[0]  # class_label =0: ignore regions 1: pedestrians 2: riders 3: sitting persons 4: other persons 5: group of people
       

        if category == 0:
            continue
#             if 
            #if category == 1 or category ==2 or category ==3 category ==4 or category ==5:
        else:
            x=category_location[1]   #class_label==1 or 2: x1，y1，w，h是与全身对齐的边界框；
            y=category_location[2]
            w=category_location[3]
            h=category_location[4]
               
       
            xml_file.write('    <object>\n')
            xml_file.write('        <name>' + 'person' + '</name>\n')
            xml_file.write('        <pose>Unspecified</pose>\n')
            xml_file.write('        <truncated>0</truncated>\n')
            xml_file.write('        <difficult>0</difficult>\n')
            xml_file.write('        <bndbox>\n')
            xml_file.write('            <xmin>' + str(x) + '</xmin>\n')
            xml_file.write('            <ymin>' + str(y) + '</ymin>\n')
            xml_file.write('            <xmax>' + str(x+w) + '</xmax>\n')
            xml_file.write('            <ymax>' + str(y+h) + '</ymax>\n')
            xml_file.write('        </bndbox>\n')
            xml_file.write('    </object>\n')
        xml_file.write('</annotation>\n')

YOLO训练VOC格式数据集，源代码中附有voc_labels.py如下，自己转下归一化的txt就OK了

import xml.etree.ElementTree as ET
import pickle
import os
from os import listdir, getcwd
from os.path import join

sets=[ ('2007', 'train'), ('2007', 'val'), ('2007', 'test')]

classes = ["car", "person","rider"]

#parser = ET.XMLParser("utf-8")
#tree = ET.fromstring(xmlstring, parser=parser)

def convert(size, box):
    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
    return (x,y,w,h)

def convert_annotation(year, image_id):
    in_file = open('VOCdevkit/VOC%s/Annotations/%s.xml'%(year, image_id))
    out_file = open('VOCdevkit/VOC%s/labels/%s.txt'%(year, image_id), 'w')
    
    #parser = ET.XMLParser(encoding="utf-8")
   # tree = ET.fromstring(in_file, parser=True)
    print in_file

    tree=ET.parse(in_file)
   # print in_file
    root = tree.getroot()
    size = root.find('size')
    w = int(size.find('width').text)
    h = int(size.find('height').text)

    for obj in root.iter('object'):
        difficult = obj.find('difficult').text
        cls = obj.find('name').text
        if cls not in classes or int(difficult)==1:
            continue
        cls_id = classes.index(cls)
        xmlbox = obj.find('bndbox')
        b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text), float(xmlbox.find('ymax').text))
        bb = convert((w,h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')

wd = getcwd()

for year, image_set in sets:
    if not os.path.exists('VOCdevkit/VOC%s/labels/'%(year)):
        os.makedirs('VOCdevkit/VOC%s/labels/'%(year))
    image_ids = open('VOCdevkit/VOC%s/ImageSets/Main/%s.txt'%(year, image_set)).read().strip().split()
    list_file = open('%s_%s.txt'%(year, image_set), 'w')
    for image_id in image_ids:
        list_file.write('%s/VOCdevkit/VOC%s/JPEGImages/%s.jpg\n'%(wd, year, image_id))
        convert_annotation(year, image_id)
    list_file.close()

#os.system("cat 2007_train.txt 2007_val.txt  > train.txt")
#os.system("cat 2007_train.txt 2007_val.txt 2007_test.txt 2012_train.txt 2012_val.txt > train.all.txt")