def find_xmls_with_minimal_size_of_this_cate(xmlDir, cate):
allXmls = FILES.get_sorted_files(xmlDir)
MinW = 9999
MinH = 9999
MinA = MinW * MinH
for xml in allXmls:
xmlPath = os.path.join(xmlDir, xml)
wMinPath = os.path.join(tarDir, '{}_wMin.xml'.format(cate))
hMinPath = os.path.join(tarDir, '{}_hMin.xml'.format(cate))
aMinPath = os.path.join(tarDir, '{}_aMin.xml'.format(cate))
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text()
if name == cate:
xmin = int(obj.find('xmin').get_text())
xmax = int(obj.find('xmax').get_text())
ymin = int(obj.find('ymin').get_text())
ymax = int(obj.find('ymax').get_text())
width = xmax - xmin
height = ymax - ymin
area = width * height
if width < MinW:
MinW = width
shutil.copy(xmlPath, wMinPath)
if height < MinH:
MinH = height
shutil.copy(xmlPath, hMinPath)
if area < MinA:
MinA = area
shutil.copy(xmlPath, aMinPath)
print("cate: {} MinW={}, MinH={}, MinA={}".format(cate, MinW, MinH, MinA))
def find_most_cates_with_cate(xmlDir, cate, savedCates):
maxCates = 0
allXmls = FILES.get_sorted_files(xmlDir)
for xml in allXmls:
labelDict = {}
xmlPath = os.path.join(xmlDir, xml)
tarPath = os.path.join(tarDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text() # find函数只找到第一个相关的内容
if name not in labelDict.keys():
labelDict[name] = 1
else:
labelDict[name] += 1
if cate in labelDict.keys():
if len(labelDict) > maxCates:
maxCates = len(labelDict)
maxXml = xmlPath
maxTar = tarPath
maxAddCates = labelDict.keys()
print("maxXml={}, maxTar={}".format(maxXml, maxTar))
shutil.copy(maxXml, maxTar)
for maxAddCate in maxAddCates:
savedCates.append(maxAddCate)
def remove_img():
imgs = [x for x in FILES.get_sorted_files(imgDir)]
xmls = [x for x in FILES.get_sorted_files(xmlDir)]
# print(imgs)
# remove img
count = 0
for img in tqdm(imgs):
find_xml_name = img.replace(imgType, xmlType)
# print(find_img_name)
if find_xml_name in xmls:
pass
# print('save %s'%img)
else:
print('del %s' % img)
os.remove(os.path.join(imgDir, img))
count += 1
print("remove {} imgs.\n".format(count))
def sample_random(sampleDir,tarDir,remainNum):
allFiles=FILES.get_sorted_files(sampleDir)
random.shuffle(allFiles)
count=0
for file in allFiles:
if count<remainNum:
filePath=os.path.join(sampleDir,file)
tarPath=os.path.join(tarDir,file)
shutil.move(filePath,tarPath)
count+=1
def remove_xml():
imgs = [x for x in FILES.get_sorted_files(imgDir)]
xmls = [x for x in FILES.get_sorted_files(xmlDir)]
# print(imgs)
# remove xml
count = 0
for xml in tqdm(xmls):
find_img_name = xml.replace(xmlType, imgType)
# print(find_img_name)
if find_img_name in imgs:
pass
# print('save %s'%xml)
else:
print('del %s' % xml)
os.remove(os.path.join(xmlDir, xml))
count += 1
print("remove {} xmls.\n".format(count))
def copy_files_refer_dir(FileDir, FileForm, ReferDir, xmlTar):
allRefers = FILES.get_sorted_files(ReferDir)
for refer in allRefers:
file = refer.split('.')[0] + FileForm
sourPath = os.path.join(FileDir, file)
tarPath = os.path.join(xmlTar, file)
try:
shutil.copy(sourPath, tarPath)
except:
pass
def find_all_cates(xmlDir):
labelDict = {}
allXmls = FILES.get_sorted_files(xmlDir)
for xml in allXmls:
xmlPath = os.path.join(xmlDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text() # find函数只找到第一个相关的内容
if name not in labelDict.keys():
labelDict[name] = 1
else:
labelDict[name] += 1
return labelDict.keys()
def move_xmls_with_cates(xmlDir, cates):
allXmls = FILES.get_sorted_files(xmlDir)
ii = 0
for xml in allXmls:
xmlPath = os.path.join(xmlDir, xml)
tarPath = os.path.join(tarDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text()
if name in cates:
shutil.move(xmlPath, tarPath)
ii += 1
break
def sample_files(sampleDir, tarDir, remainNum):
allFiles = FILES.get_sorted_files(sampleDir)
filesCount = len(allFiles)
REMRATE = filesCount / remainNum
remRate = REMRATE
count = int(REMRATE) + 1
fileNum = 0
for file in allFiles:
if count - remRate > 0:
filePath = os.path.join(sampleDir, file)
tarPath = os.path.join(tarDir, file)
shutil.copy(filePath, tarPath)
fileNum += 1
remRate += REMRATE
count += 1
def filter_xmls_without_cates(xmlDir, cates):
allXmls = FILES.get_sorted_files(xmlDir)
for xml in allXmls:
xmlPath = os.path.join(xmlDir, xml)
tarPath = os.path.join(tarDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
noTruck = 1
for obj in objs:
name = obj.find('name').get_text()
if name in cates:
noTruck = 0
break
if noTruck == 1:
shutil.copy(xmlPath, tarPath)
def filter_short_cates(xmlDir, cates):
allXmls = FILES.get_sorted_files(xmlDir)
for xml in allXmls:
xmlPath = os.path.join(xmlDir, xml)
tarPath = os.path.join(tarDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text()
if name in cates:
bndb = obj.find("bndbox")
xsize = int(bndb.find('xmax').text) - int(
bndb.find('xmin').text)
ysize = int(bndb.find('ymax').text) - int(
bndb.find('ymin').text)
if xsize > ysize * 0.9:
shutil.copy(xmlPath, tarPath)
break
def filter_lack_helmet(xmlDir):
personCates = ['person']
helmetCates = ['red', 'yellow', 'blue', 'white', 'orange']
allXmls = FILES.get_sorted_files(xmlDir)
for xml in allXmls:
numPer = 0
numHel = 0
xmlPath = os.path.join(xmlDir, xml)
lackPath = os.path.join(tarDir, xml)
txts = TXT.read_txt(xmlPath)
soup = BeautifulSoup(txts, "xml")
objs = soup.find_all('object')
for obj in objs:
name = obj.find('name').get_text()
if name in personCates:
numPer += 1
elif name in helmetCates:
numHel += 1
else:
print("Lack of category {} set in function filter_lack_helmet".
format(name))
if numPer > numHel:
shutil.copy(xmlPath, lackPath)
# encoding=utf8
import os
from basicFun import FILES
if __name__ == "__main__":
count = 0
rmvDir = r"/DATACENTER3/yh/detectron/caffe2/detect/factory/people_longVideoAll/"
referDirRoot = r"/DATACENTER3/yh/detectron/caffe2/detect/factory"
# referDirNames=["tankClose","tankNoPipe","tankPipe","terribleCover"]
referDirNames = ["outXml"]
referType = ".xml"
rmvType = ".jpg"
# referDirNames=["1","2","3"]
allRmvs = [x for x in FILES.get_sorted_files(rmvDir)]
for referDirName in referDirNames:
referDir = referDirRoot + '/' + referDirName
allRefers = [
x for x in FILES.get_sorted_files(referDir) if referType in x
]
# 审核为.jpg形成列表
for rmv in allRmvs:
if (rmv.split('.')[0] + referType) not in allRefers:
removePath = rmvDir + '/' + rmv.split('.')[0] + rmvType
# removePath=rmvDir+'/'+rmv
os.remove(removePath)
count += 1
print("cancel {} files={}\n".format(rmvType, count))
import os, shutil
from basicFun import FILES
dir1 = r"/disk2/hao.yang/project/Qin/data/preProcess/tolabelImg/0501/safe_cashbox_FengShi_simulate/"
des1 = r"/disk2/hao.yang/project/Qin/data/preProcess/tolabelImg/0501/safe_cashbox_FengShi_simulate_img/"
dir2 = '/disk2/hao.yang/project/Qin/data/preProcess/tolabelImg/0501/xml/'
des2 = '/disk2/hao.yang/project/Qin/data/preProcess/tolabelImg/0501/labeled_tube_px/'
FILES.mkdir(des1)
FILES.mkdir(des2)
preTitle = 'safe_cashbox_FengShi_simulate_'
endTitle = ''
i = 0
filelist = FILES.get_sorted_files(dir1)
for file in filelist:
if ".jpg" in file:
newfileName = '%06d' % i
# newfileName=file
# shutil.copy(os.path.join(dir1,file),os.path.join(des1,preTitle+newfileName+endTitle+".jpg"))
shutil.copy(os.path.join(dir1, file),
os.path.join(des1, preTitle + file))
if os.path.exists(
os.path.join(
dir2, 'unload_FengShi_tube_simulate_' + newfileName +
endTitle + ".xml")):
shutil.copy(
os.path.join(
dir2, 'unload_FengShi_tube_simulate_' + newfileName +
endTitle + ".xml"),
os.path.join(des2, preTitle + file.replace('.jpg', ".xml")))
else:
print(file, "no xml")
i += 1
#encoding=utf-8
import os
import shutil
from basicFun import FILES
if __name__ == "__main__":
kind = 'door_close'
referXml = "/DATACENTER5/hao.yang/dataRoom/Qin/safe/caotan/referXml/{}.xml".format(
kind)
xmlDir = "/DATACENTER5/hao.yang/dataRoom/Qin/safe/caotan/labels_door/"
imgDir = "/DATACENTER5/hao.yang/dataRoom/Qin/safe/caotan/safe_door_classified_all/{}".format(
kind)
allfiles = [x for x in FILES.get_sorted_files(imgDir)]
for file in allfiles:
newPath = os.path.join(xmlDir, file.split('.')[0] + '.xml')
shutil.copy(referXml, newPath)
#encoding=utf-8
import os
import shutil
import time
from basicFun import FILES
jpgRoot=r"/DATACENTER2/yh/dataRoom/gasStation/tank/train/sj18JunFrames"
if __name__=="__main__":
remainNum=500
start = time.time()
jpgDirs=FILES.get_sub_dirs(jpgRoot)
for dirName in jpgDirs:
jpgDir=jpgRoot+'/'+dirName
allJpgs=FILES.get_sorted_files(jpgDir)
jpgCount=len(allJpgs)
if jpgCount>3000:
for jpg in allJpgs[remainNum:]:
jpgPath=os.path.join(jpgDir,jpg)
os.remove(jpgPath)
# def removeRemain(rmdir,remainNum)：
# allFiles=FILES.get_sorted_files(rmdir)
# filesCount=len(rmdir)
# for
# remRate=REMRATE
# # print(rmRate)
# count=int(REMRATE)+1
# fileNum=0
# for file in allFiles:
# if count-remRate>0:
# filePath=dirDir+'/'+file
# tarPath=tarDir+'/'+file
if os.path.exists(dirpath):
difdirpath = dirpath.replace('extractFrame',
'removedSimilarityFrame')
# if difdirpath[-1]=='/':
# difdirpath=difdirpath[:-1]+"Unique"+str(int(SIMIBOUND*1000))
# else:
# difdirpath=difdirpath+"Unique"+str(int(SIMIBOUND*1000))
if dirCount < startdir:
dirCount += 1
continue
if dirCount > enddir:
break
begin = 0
restart = 1
FILES.rm_mkdir(difdirpath)
allFILES = FILES.get_sorted_files(dirpath)
jpgsCount = len(allFILES)
for file in allFILES:
if ".jpg" in file:
# basePath='/DATACENTER5/hao.yang/dataRoom/Qin/trade/zhangbabeilu/29AD13E0_1552790204_1_000061.jpg'
checkPath = os.path.join(dirpath, file)
# print(checkPath)
if 'shgjihsjkghj' in file:
restart = 1
if restart == 0:
continue
if begin == 0:
difpath = os.path.join(difdirpath, file)
shutil.copy(checkPath, difpath)
basePath = checkPath
begin = 1
#encoding=utf-8
'assign img task to members as sequence'
import os
import shutil
import random
# personal lib
from basicFun import FILES
if __name__ == "__main__":
namei = 0
members = ['day_1', 'day_2', 'day_3', 'day_4']
taskDir = '/DATACENTER2/ke.cao/oil_video_Data/unload_images_dif'
outRoot = taskDir + '_day'
FILES.mkdir(outRoot)
allImgs = FILES.get_sorted_files(taskDir)
# 每人大概分配的数量
perAmount = int(len(allImgs) / len(members))
# 可能会有分不匀的情况
redundant = len(allImgs) % len(members)
print('{}*{}+{}=?{}'.format(perAmount, len(members), redundant,
len(allImgs)))
assignCount = 0
for member in members:
outDir = os.path.join(outRoot, member)
FILES.mkdir(outDir)
memberAmount = perAmount
for img in allImgs:
if assignCount < memberAmount:
# print(members[namei])
desDir = os.path.join(outRoot, members[namei])
# jpg move
srcPath = os.path.join(taskDir, img)
sourPath = os.path.join(FileDir, file)
tarPath = os.path.join(xmlTar, file)
try:
shutil.copy(sourPath, tarPath)
except:
pass
sites = ['FengShi', 'MingDeMen', 'XiWan']
for site in sites:
for isle in range(1, 6):
outXmlDir = "/disk2/hao.yang/project/Qin/data/imgs/isle/{}/isle{}_{}_img/".format(
site, isle, site)
volume = 1000
partition = 1
allFiles = FILES.get_sorted_files(outXmlDir)
random.shuffle(allFiles)
count = 0
for file in allFiles:
if count < volume + volume * (partition - 1) * 0.2:
count += 1
else:
# exit()
# break
partition += 1
count = 0
# tar_outXmlDir='{}_partition{}'.format(outXmlDir,partition)
pDir = '/disk2/hao.yang/project/Qin/data/labelTask/isles/p{}'.format(
partition)
FILES.mkdir(pDir)
tar_outXmlDir = '/disk2/hao.yang/project/Qin/data/labelTask/isles/p{}/{}_isle{}'.format(
print(iou)
if iou > 0.6:
labelObj['name'] = modelObj['name']
return labelObj
modelXmlDir = '/disk2/hao.yang/project/Qin/data/imgs/isle/modelxml_isle_17117_withred/'
labelXmlDir = r'/disk2/hao.yang/project/Qin/data/imgs/isle/modelxml_isle_17117_havered/'
# checkedDir='/disk2/hao.yang/project/Qin/data/xmls/checkout/complete'
attentionNames = ['red']
confuseNames = ['yellow', 'other', 'blue']
count = 0
allChecked = []
# allChecked=[x for x in FILES.get_sorted_files(checkedDir) if ".xml" in x]
allXmls = [
x for x in FILES.get_sorted_files(modelXmlDir)
if ".xml" in x and x not in allChecked
]
for xml in allXmls:
labelXmlPath = os.path.join(labelXmlDir, xml)
modelXmlPath = os.path.join(modelXmlDir, xml)
if os.path.exists(labelXmlPath):
labelObjs = XML.read_objects(labelXmlPath)
modelObjs = XML.read_objects(modelXmlPath)
newObjs = []
for labelObj in labelObjs:
if labelObj['name'] in confuseNames:
newObjs.append(fuse(labelObj, modelObjs))
else:
newObjs.append(labelObj)
tree = ET.ElementTree(file=labelXmlPath)
file = refer.split('.')[0] + FileForm
sourPath = os.path.join(FileDir, file)
tarPath = os.path.join(xmlTar, file)
try:
shutil.copy(sourPath, tarPath)
except:
pass
xmlDir = r"E:\factory\voc\add_data\summary\rawImg\Tank_jl"
jpgDir = r"E:\research\lackLabel\voc\img4206"
xmlTar = r"E:\factory\voc\add_data\summary\rawImg\Tank_jl_lite"
jpgTar = r"E:\research\lackLabel\voc\img2103_1"
remainNum = 70
FILES.rm_mkdir(xmlTar)
# FILES.rm_mkdir(jpgTar)
allFiles = FILES.get_sorted_files(xmlDir)
filesCount = len(allFiles)
REMRATE = filesCount / remainNum
remRate = REMRATE
count = int(REMRATE) + 1
fileNum = 0
for file in allFiles:
if count - remRate > 0:
filePath = xmlDir + '/' + file
tarPath = xmlTar + '/' + file
shutil.copy(filePath, tarPath)
fileNum += 1
remRate += REMRATE
count += 1
# copy_files_refer_dir(jpgDir,'.jpg',xmlTar,jpgTar)
if imgDir[-1] == '/':
errorDir = imgDir[:-1] + "_error"
difDir = imgDir[:-1] + "_dif"
else:
errorDir = imgDir + "_error"
difDir = imgDir + "_dif"
if xmlDir[-1] == '/':
difXmlDir = xmlDir[:-1] + "_dif"
else:
difXmlDir = xmlDir + "_dif"
print(xmlDir)
FILES.rm_mkdir(difXmlDir)
FILES.rm_mkdir(difDir)
# allXmls=[x for x in FILES.get_sorted_files(xmlDir) if '.xml' in x]
allXmls = [
x.replace('.jpg', '.xml') for x in FILES.get_sorted_files(imgDir)
if '.jpg' in x
]
begin = 0
labelNumber = 0
difNumber = 0
resume = 0
t = 0
baseTitle = 'baseTitle'
print(len(allXmls))
for xmlName in tqdm(allXmls):
curTitle = xmlName.split('_000')[0]
if '.' not in curTitle:
checkTitle = curTitle
else:
checkTitle = baseTitle
#encoding=utf8
#所有xml文件内含box的数目
import os, sys
from bs4 import BeautifulSoup
import shutil
from basicFun import FILES
from basicFun import TXT
from tqdm import tqdm
if __name__ == "__main__":
labelDict = {}
xmlpath = sys.argv[1]
index = 0
safeboxCount = 0
allxml = [x for x in FILES.get_sorted_files(xmlpath) if ".xml" in x]
for xmlName in tqdm(allxml):
xmlpath3 = os.path.join(xmlpath, xmlName)
txts = TXT.read_txt(xmlpath3)
soup = BeautifulSoup(txts, "xml") #解析xml文
section = soup.find_all('object') #通过object标签匹配所有相关的标签中的内容
for sec in section:
name = sec.find('name').get_text() # find函数只找到第一个相关的内容
if name not in labelDict.keys():
labelDict[name] = 1
else:
labelDict[name] += 1
print(xmlpath, '\nFILES=', len(allxml), "classes=", len(labelDict))
print(labelDict.items())
XML.write_xml(tree,xmlPath)
def add_new_cates(xmlPath,referPath,addCates):
addBoxes=XML.read_objects(referPath)
tree = ET.ElementTree(file=xmlPath)
root = tree.getroot()
root=XML.del_tag(root,addCates)
# print(addBoxes)
for box in addBoxes:
if box['name'] in addCates:
root=XML.add_tag(root,box)
XML.write_xml(tree,xmlPath)
if __name__=="__main__":
# imgDir=r"F:\gasStation\tank\regular_dataset\SanJiang_tankbox\checked\no_pipe"
xmlDir=r'/disk2/hao.yang/project/Qin/data/xmls/guide/complete/guide_FMXX_8000_complete/'
referRoot='/disk2/hao.yang/project/Qin/data/xmls/guide/complete/oilstation/'
referDirs=FILES.get_sub_dirs(referRoot)
for dirName in referDirs:
referDir=os.path.join(referRoot,dirName)
# referPath=r'F:\gasStation\tank\regular_dataset\SanJiang_tankbox\tank\SanJiang_refer_9.xml'
# addCates=['car','motorcycle','truck','bus']
# addCates=['blue','yellow','other','security','red']
# addCates=['tank_close','tank_open']
# addCates=['cashbox_close','cashbox_open']
addCates=['gun_in','gun_out']
allXmls=[x for x in FILES.get_sorted_files(xmlDir) if ".xml" in x]
for xml in allXmls:
xmlPath=os.path.join(xmlDir,xml)
referPath=os.path.join(referDir,xml)
if os.path.exists(referPath):
add_cates(xmlPath,referPath,addCates)
# sys.stdout.write('\r>> Detecting {name} {count:.2f}%'.format(name=videoPath.split('/')[-1],count=(100*float(i)/total_frames)))
sys.stdout.flush()
return filteredXmls
xmlDir = r"/DATACENTER2/hao.yang/project/Qin/checkout/FengShi/xml1383_8kinds/"
if xmlDir[-1] == '/':
xmlTar = xmlDir[:-1] + '_val'
else:
xmlTar = xmlDir + '_val'
FILES.mkdir(xmlTar)
# Get leastNum boxes of the cate at least
leastNum = 40
count = 0
valCount = 0
allXmls = FILES.get_sorted_files(xmlDir)
# Filter xmls having a box of a cate
cateName = 'pos_idle'
print('Start to filter xmls containing cate {}'.format(cateName))
filteredXmls = filter_by_cate(allXmls, cateName)
print('\nTotally {} xmls containing cate {}'.format(len(filteredXmls),
cateName))
valRate = int(len(filteredXmls) / leastNum)
for xml in filteredXmls:
if count % valRate == 0:
xmlPath = os.path.join(xmlDir, xml)
tarPath = os.path.join(xmlTar, xml)
shutil.copy(xmlPath, tarPath)
valCount += 1
count += 1
print('Made validation set with {} xmls'.format(valCount))
newObjs.append(labelObj)
return differ, newObjs
imgDir = r'/DATACENTER4/hao.yang/project/Qin/data/imgs/safe/safe_FMXX_img/'
# modelXmlDir=imgDir[:-1]+'_modelxml_baseline4'
modelXmlDir = '/DATACENTER4/hao.yang/project/Qin/data/xmls/safe/modelXml_safe_FMXX/'
labelXmlDir = r'/DATACENTER4/hao.yang/project/Qin/data/xmls/safe/safe_FMXX_22983_total/'
badDir = '/DATACENTER4/hao.yang/project/Qin/data/xmls/safe/bad_safe_FMXX/'
checkedDir = '/'
FILES.rm_mkdir(badDir)
nocareKinds = ['cashbox_close', 'cashbox_open', 'safe_hide', 'jug']
bigNames = []
# bigNames不可能有漏标，如果漏掉，必然是故意的，所以不算错误
count = 0
allChecked = [x for x in FILES.get_sorted_files(checkedDir) if ".xml" in x]
if allChecked:
allXmls = [
x for x in FILES.get_sorted_files(labelXmlDir)
if ".xml" in x and x not in allChecked
]
else:
allXmls = [x for x in FILES.get_sorted_files(modelXmlDir) if ".xml" in x]
for xml in tqdm(allXmls):
labelXmlPath = os.path.join(labelXmlDir, xml)
modelXmlPath = os.path.join(modelXmlDir, xml)
if os.path.exists(labelXmlPath):
modelObjs = XML.read_objects(modelXmlPath)
labelObjs = XML.read_objects(labelXmlPath)
differ = 0
differ, newObjs = combine(modelObjs, labelObjs)
'ph': 267,
'xgt': 318,
'yxy': 360,
'zt': 259,
'hyd': 276,
'hjc': 413
}
taskImg = r'E:\factory\voc\helmet_img'
taskXml = r'E:\factory\voc\labeled_xml_lackHel\xml'
outRoot = r'E:\factory\voc\labeled_xml_lackHel_assignTask'
for member in members:
outDir = outRoot + '/' + member
FILES.rm_mkdir(outDir)
FILES.mkdir(outDir + '/img')
FILES.mkdir(outDir + '/xml')
allXmls = FILES.get_sorted_files(taskXml)
random.shuffle(allXmls)
memberI = 0
memberF = 0
for xml in allXmls:
print(memberI)
member = members[memberI]
desDir = outRoot + '/' + member
xmlPath = taskXml + '/' + xml
desXml = desDir + '/xml/' + xml
jpg = xml.split('.')[0] + '.jpg'
jpgPath = taskImg + '/' + jpg
desJpg = desDir + '/img/' + jpg
# copy file
shutil.copy(xmlPath, desXml)
shutil.copy(jpgPath, desJpg)