研究背景

在深度学习中常用的数据集进行归纳和总结

语义分割的数据集

1、COCO 数据集

COCO(Common Objects in Context)是一个新的图像识别、分割和图像语义数据集，是一个大规模的图像识别、分割、标注数据集。它可以用于多种竞赛，与本领域最相关的是检测部分，因为其一部分是致力于解决分割问题的。

COCO2014数据集类别汇总

coco目标检测数据集标注目标信息采用的是数据格式是json，其内容本质是一种字典结构，字典堆栈和列表信息内容维护。
coco里面的id和类名字对应：总共80类，但id号到90

    person  # 1
    vehicle 交通工具 #8
        {bicycle
         car
         motorcycle
         airplane
         bus
         train
         truck
         boat}
    outdoor  #5
        {traffic light
        fire hydrant
        stop sign
        parking meter
        bench}
    animal  #10
        {bird
        cat
        dog
        horse
        sheep
        cow
        elephant
        bear
        zebra
        giraffe}
    accessory 饰品 #5
        {backpack 背包
        umbrella 雨伞
        handbag 手提包
        tie 领带
        suitcase 手提箱
        }
    sports  #10
        {frisbee
        skis
        snowboard
        sports ball
        kite
        baseball bat
        baseball glove
        skateboard
        surfboard
        tennis racket
        }
    kitchen  #7
        {bottle
        wine glass
        cup
        fork
        knife
        spoon
        bowl
        }
    food  #10
        {banana
        apple
        sandwich
        orange
        broccoli
        carrot
        hot dog
        pizza
        donut
        cake
        }
    furniture 家具 #6
        {chair
        couch
        potted plant
        bed
        dining table
        toilet
        }
    electronic 电子产品 #6
        {tv
        laptop
        mouse
        remote
        keyboard
        cell phone
        }
    appliance 家用电器 #5
        {microwave
        oven
        toaster
        sink
        refrigerator
        }
    indoor  #7
        {book
        clock
        vase
        scissors
        teddy bear
        hair drier
        toothbrush
        }

coco_id_name_map={1: 'person', 2: 'bicycle', 3: 'car', 4: 'motorcycle', 5: 'airplane',
                   6: 'bus', 7: 'train', 8: 'truck', 9: 'boat', 10: 'traffic light',
                   11: 'fire hydrant', 13: 'stop sign', 14: 'parking meter', 15: 'bench',
                   16: 'bird', 17: 'cat', 18: 'dog', 19: 'horse', 20: 'sheep', 21: 'cow',
                   22: 'elephant', 23: 'bear', 24: 'zebra', 25: 'giraffe', 27: 'backpack',
                   28: 'umbrella', 31: 'handbag', 32: 'tie', 33: 'suitcase', 34: 'frisbee',
                   35: 'skis', 36: 'snowboard', 37: 'sports ball', 38: 'kite', 39: 'baseball bat',
                   40: 'baseball glove', 41: 'skateboard', 42: 'surfboard', 43: 'tennis racket',
                   44: 'bottle', 46: 'wine glass', 47: 'cup', 48: 'fork', 49: 'knife', 50: 'spoon',
                   51: 'bowl', 52: 'banana', 53: 'apple', 54: 'sandwich', 55: 'orange',
                   56: 'broccoli', 57: 'carrot', 58: 'hot dog', 59: 'pizza', 60: 'donut',
                   61: 'cake', 62: 'chair', 63: 'couch', 64: 'potted plant', 65: 'bed', 67: 'dining table',
                   70: 'toilet', 72: 'tv', 73: 'laptop', 74: 'mouse', 75: 'remote', 76: 'keyboard',
                   77: 'cell phone', 78: 'microwave', 79: 'oven', 80: 'toaster', 81: 'sink',
                   82: 'refrigerator', 84: 'book', 85: 'clock', 86: 'vase', 87: 'scissors',
                   88: 'teddy bear', 89: 'hair drier', 90: 'toothbrush'}

COCO2017数据集类别汇总

包含了超过80个物体类别，分别为：['background = 0','person=1', 'bicycle=2', 'car=3', 'motorcycle=4', 'airplane=5', 'bus=6', 'train=7', 'truck=8', 'boat=9', 'traffic light=10', 'fire hydrant=11', 'stop sign=13', 'parking meter=14', 'bench=15', 'bird=16', 'cat=17', 'dog=18', 'horse=19', 'sheep=20', 'cow=21', 'elephant=22', 'bear=23', 'zebra=24', 'giraffe=25', 'backpack=27', 'umbrella=28', 'handbag=31', 'tie=32', 'suitcase=33', 'frisbee=34', 'skis=35', 'snowboard=36', 'sports ball=37', 'kite=38', 'baseball bat=39', 'baseball glove=40', 'skateboard=41', 'surfboard=42', 'tennis racket=43', 'bottle=44', 'wine glass=46', 'cup=47', 'fork=48', 'knife=49', 'spoon=50', 'bowl=51', 'banana=52', 'apple=53', 'sandwich=54', 'orange=55', 'broccoli=56', 'carrot=57', 'hot dog=58', 'pizza=59', 'donut=60', 'cake=61', 'chair=62', 'couch=63', 'potted plant=64', 'bed=65', 'dining table=67', 'toilet=70', 'tv=72', 'laptop=73', 'mouse=74', 'remote=75', 'keyboard=76', 'cell phone=77', 'microwave=78', 'oven=79', 'toaster=80', 'sink=81', 'refrigerator=82', 'book=84', 'clock=85', 'vase=86', 'scissors=87', 'teddy bear=88', 'hair drier=89', 'toothbrush=90']。

91个填充类别，分别为['banner=92', 'blanket=93', 'branch=94', 'bridge=95', 'building-other=96', 'bush=97', 'cabinet=98', 'cage=99', 'cardboard=100', 'carpet=101', 'ceiling-other=102', 'ceiling-tile=103', 'cloth=104', 'clothes=105', 'clouds=106', 'counter=107', 'cupboard=108', 'curtain=109', 'desk-stuff=110', 'dirt=111', 'door-stuff=112', 'fence=113', 'floor-marble=114', 'floor-other=115', 'floor-stone=116', 'floor-tile=117', 'floor-wood=118', 'flower=119', 'fog=120', 'food-other=121', 'fruit=122', 'furniture-other=123', 'grass=124', 'gravel=125', 'ground-other=126', 'hill=127', 'house=128', 'leaves=129', 'light=130', 'mat=131', 'metal=132', 'mirror-stuff=133', 'moss=134', 'mountain=135', 'mud=136', 'napkin=137', 'net=138', 'paper=139', 'pavement=140', 'pillow=141', 'plant-other=142', 'plastic=143', 'platform=144', 'playingfield=145', 'railing=146', 'railroad=147', 'river=148', 'road=149', 'rock=150', 'roof=151', 'rug=152', 'salad=153', 'sand=154', 'sea=155', 'shelf=156', 'sky-other=157', 'skyscraper=158', 'snow=159', 'solid-other=160', 'stairs=161', 'stone=162', 'straw=163', 'structural-other=164', 'table=165', 'tent=166', 'textile-other=167', 'towel=168', 'tree=169', 'vegetable=170', 'wall-brick=171', 'wall-concrete=172', 'wall-other=173', 'wall-panel=174', 'wall-stone=175', 'wall-tile=176', 'wall-wood=177', 'water-other=178', 'waterdrops=179', 'window-blind=180', 'window-other=181', 'wood=182', 'other=183']。提供了118287张训练图片，5000张验证图片，以及超过40670张测试图片。由于其规模巨大，目前已非常常用，对领域发展很重要。实际上，该竞赛的结果每年都会在ECCV的研讨会上与ImageNet数据集的结果一起公布。它有如下特点：
1）Object segmentation：物体分割
2）Recognition in context ：上下文识别
3）Superpixel stuff segmentation：超分辨率的实物分割
4）330K images (>200K labeled)：33万张图片（超过20万有标记）
5）1.5 million object instances：150万个物体实例
6）80 object categories：80个物体类别
9）91 stuff categories ：91个stuff类别
10）5 captions per image：每张图像5个标题
11）250,000 people with keypoints：25万张带关节点的人物图片

COCO数据集对于图像的标注信息不仅有类别、位置信息，还有对图像的语义文本描述，COCO数据集的开源使得近两三年来图像分割语义理解取得了巨大的进展，也几乎成为了图像语义理解算法性能评价的“标准”数据集。详细介绍参考。注意COCO用于语义分割的API要从这里下载：https://github.com/nightrome/cocostuffapi

代码：获取COCO caption 每张图片有5句文本描述

from pycocotools.coco import COCO
import numpy as np
import skimage.io as io
import matplotlib.pyplot as plt
import pylab
pylab.rcParams['figure.figsize'] = (8.0, 10.0)
dataDir='./coco2017'
dataType='val2017'  # train2017
# initialize COCO api for caption annotations\n",
annFile = '{}/annotations/captions_{}.json'.format(dataDir,dataType)
coco=COCO(annFile)

coco_caps=COCO(annFile)
imgIdsall = coco_caps.getImgIds()
print(imgIdsall)
print(len(imgIdsall))

for i in imgIdsall:

  imgIds = coco.getImgIds(imgIds = [i])
  img = coco.loadImgs(imgIds[np.random.randint(0,len(imgIds))])[0]
  print(img)

  str = img['file_name']
  str1 = str[:-4]
  print(str1)

  path = './val2017/'+str1+'.txt'  # train2017

  with open (path,'w') as f:

  # load and display caption annotations\n",
    annIds = coco.getAnnIds(imgIds=img['id'])
    anns = coco.loadAnns(annIds)
    for ann in anns:
      print(ann['caption'])
      f.write(ann['caption']+'\n')
      print('    ')
  coco.showAnns(anns)

代码：从指定文本中，读取文件名，然后总指定路径将文件复制到指定文件夹中

# -*- coding: utf-8 -*-   
import time     
import os  
import shutil
 
def re_mycopyfile(srcfile,dstfile,num):
    #name_long=16
    l=len(str(num))
    zero='00000000'
    newname = srcfile[-16:-4]
    if not os.path.isfile(srcfile):
        print "%s not exist!"%(srcfile)
    else:
        #fpath,fname=os.path.split(dstfile)    #分离文件名和路径
        if not os.path.exists(dstfile):
            os.makedirs(dstfile) #创建路径
        #dstfile=dstfile+zero[:name_long-l-1]+str(num)+'.txt'
        dstfile = dstfile+str(newname)+'.txt'
        print dstfile             
        shutil.copyfile(srcfile,dstfile)      #复制文件
        print "copy %s -> %s"%(srcfile,dstfile)
 
 
 
if __name__ == '__main__':
    path1="/home/henry/Files/ICCV2019/cocostuffapi/PythonAPI/trainls.txt"  # 待复制文件列表
    path2="/home/henry/Files/ICCV2019/cocostuffapi/PythonAPI/train2017all/"  # 待复制文件目录
    path3="/home/henry/Files/ICCV2019/cocostuffapi/PythonAPI/train2017/"  # 保存目标目录
    path4="/home/henry/Files/ICCV2019/cocostuffapi/PythonAPI/trainnew.txt"  
 
    begin=0
    count=begin
    with open(path1,'r')as f:
        for line in f:
            line=line.split('\n')
            print line[0]
            srcfile = path2+str(line[0])
            print srcfile
            count=count+1
            print count
            dstfile=path3
            re_mycopyfile(srcfile,dstfile,count)
 
    count=begin
    name_long=6
    l=len(str(count+1))
    zero='00000000'

    with open(path1,'r')as f:
        for line in f:
            count=count+1
            out_words=line.split('/')
            #out_words[-1]=zero[:name_long-l-1]+str(count)+'.txt'
            out_words[-1] = zero[:name_long - l - 1] + str(count) + '.txt'
            with open(path4,'a+') as fp:
                fp.write("/".join(out_words)+"\n")

2、VOC2007数据集

类别汇总

    aeroplane
    bicycle
    bird
    boat
    bottle
    bus
    car
    cat
    chair
    cow
    diningtable
    dog
    horse
    motorbike
    person
    pottedplant
    sheep
    sofa
    train
    tvmonitor

• MSCOCO数据集格式转化成VOC数据集格式
  参考链接COCO数据集转化成VOC数据集格式
  首先得到COCO_train.json文件，可以根据实际需要的类别进行修改

#-*- coding:utf-8-*-
import json
className = {  # 84  total
    1:'person',
    2:'bicycle',
    3:'car',
    4:'motorcycle',
    5:'airplane',
    6:'bus',
    7:'train',
    8:'truck',
    9:'boat',
    10:'traffic light',
    11:'fire hydrant',
    13:'stop sign',
    14:'parking meter',
    15:'bench',
    16:'bird',
    17:'cat',
    18:'dog',
    19:'horse',
    20:'sheep',
    21:'cow',
    22:'elephant',
    23:'bear',
    24:'zebra',
    25:'giraffe',
    27:'backpack',
    28:'umbrella',
    31:'handbag',
    32:'tie',
    33:'suitcase',
    34:'frisbee',
    35:'skis',
    36:'snowboard',
    37:'sports ball',
    38:'kite',
    39:'baseball bat',
    40:'baseball glove',
    41:'skateboard',
    42:'surfboard',
    43:'tennis racket',
    44:'bottle',
    46:'wine glass',
    47:'cup',
    48:'fork',
    49:'knife',
    50:'spoon',
    51:'bowl',
    52:'banana',
    53:'apple',
    54:'sandwich',
    55:'orange',
    56:'broccoli',
    57:'carrot',
    58:'hot dog',
    59:'pizza',
    60:'donut',
    61:'cake',
    62:'chair',
    63:'couch',
    64:'potted plant',
    65:'bed',
    67:'dining table',
    70:'toilet',
    71:'truck',
    72:'tv',
    73:'laptop',
    74:'mouse',
    75:'remote',
    76:'keyboard',
    77:'cell phone',
    78:'microwave',
    79:'oven',
    80:'toaster',
    81:'sink',
    82:'refrigerator',
    84:'book',
    85:'clock',
    86:'vase',
    87:'scissors',
    88:'teddy bear',
    89:'hair drier',
    90:'toothbrush',
}
classNum = [1,2,3,4,5,6,7,8,9,10,
11,12,13,14,15,16,17,18,19,20,
21,22,23,24,25,26,27,28,29,30,
31,32,33,34,35,36,37,38,39,40,
41,42,43,44,45,46,47,48,49,50,
51,52,53,54,55,56,57,58,59,60,
61,62,63,64,65,66,67,68,69,70,
71,72,73,74,75,76,77,78,79,80,
81,82,83,84,85,86,87,88,89,90]

cocojson="/home/ouc/data1/liuhongzhi/AttnGAN/dataset/coco2014/annotations/instances_train2014.json"
def writeNum(Num):
    with open("COCO_train.json", "a+") as f:
        f.write(str(Num))
inputfile = []
inner = {}
cnt = 0
with open(cocojson, "r+") as f:
    allData = json.load(f)
    data =allData["annotations"]
    print(data[1])
    print("read ready")
for i in data:
    if (i['category_id'] in classNum):
        inner = {
            "filename":str(i["image_id"]).zfill(12),
            "name":className[i["category_id"]],
            "bndbox":i["bbox"]
        }
        inputfile.append(inner)
        cnt = cnt + 1
        if cnt%10000 == 0:
           print("id : " + str(cnt))
inputfile = json.dumps(inputfile)
writeNum(inputfile)

其次根据选取出来的类别中的图片筛选需要的图片到指定目录存放，得到训练集图片

# -*- coding: utf-8 -*-
# @Time    : 2018/03/09 10:46
# @Author  : SyGoing
# @Site    :
# @File    : getimagesbyID.py
# @Software: PyCharm
import json
import os
import cv2
#from utils.timer import Timer

nameStr = []
with open("COCO_train.json", "r+") as f:
    data = json.load(f)
    print("read ready")
for i in data:
    imgName = "COCO_train2014_"+ str(i["filename"]) + ".jpg"
    nameStr.append(imgName)
nameStr = set(nameStr)
print(nameStr)
print(len(nameStr))

#t_total = Timer()
#total_time = t_total.toc()
#wait_time = max(int(60 - total_time * 1000), 1)
#cv2.waitKey(0)

path = "/home/ouc/data1/liuhongzhi/AttnGAN/dataset/coco2014/images/train2014/"
savePath="/home/ouc/data1/liuhongzhi/yolo2-pytorch/datasets/COCO/VOC2007/JPEGImages/"
count=0
for file in nameStr:
    print(path+file)
    img=cv2.imread(path+file)
    '''
    print(str(img))
    cv2.imshow('test',img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()
    '''
    cv2.imwrite(savePath+file,img)
    count=count+1
    print('num: '+count.__str__()+'     '+file)

然后根据筛选出来的图片ID生成VOC数据集的XML文件到Annotations文件夹

#-*- coding:utf-8-*-

import xml.dom
import xml.dom.minidom
import os
# from PIL import Image
import cv2
import json

# xml文件规范定义


_IMAGE_PATH = '/home/ouc/data1/liuhongzhi/yolo2-pytorch/datasets/COCO/VOC2007/JPEGImages/'

_INDENT = '' * 4
_NEW_LINE = '\n'
_FOLDER_NODE = 'COCO2014'
_ROOT_NODE = 'annotation'
_DATABASE_NAME = 'LOGODection'
_ANNOTATION = 'COCO2014'
_AUTHOR = 'SyGoing_CSDN'
_SEGMENTED = '0'
_DIFFICULT = '0'
_TRUNCATED = '0'
_POSE = 'Unspecified'

# _IMAGE_COPY_PATH= 'JPEGImages'
_ANNOTATION_SAVE_PATH = '/home/ouc/data1/liuhongzhi/yolo2-pytorch/datasets/COCO/VOC2007/Annotations/'


# _IMAGE_CHANNEL= 3

# 封装创建节点的过程
def createElementNode(doc, tag, attr):  #创建一个元素节点
    element_node = doc.createElement(tag)

    # 创建一个文本节点
    text_node = doc.createTextNode(attr)

    # 将文本节点作为元素节点的子节点
    element_node.appendChild(text_node)

    return element_node


# 封装添加一个子节点
def createChildNode(doc, tag, attr, parent_node):
    child_node = createElementNode(doc,tag, attr)

    parent_node.appendChild(child_node)


# object节点比较特殊
def createObjectNode(doc, attrs):
    object_node =doc.createElement('object')

    midname=attrs['name']


    #if midname !='person':   # 注释后可以得到所有类别
    #    midname='car'

    createChildNode(doc, 'name', midname,
                    object_node)

    #createChildNode(doc, 'name',attrs['name'],
    #                object_node)

    createChildNode(doc, 'pose',
                    _POSE, object_node)

    createChildNode(doc, 'truncated',
                    _TRUNCATED,object_node)

    createChildNode(doc, 'difficult',
                    _DIFFICULT,object_node)

    bndbox_node = doc.createElement('bndbox')

    createChildNode(doc, 'xmin',str(int(attrs['bndbox'][0])),
                    bndbox_node)

    createChildNode(doc, 'ymin',str(int(attrs['bndbox'][1])),
                    bndbox_node)

    createChildNode(doc, 'xmax',str(int(attrs['bndbox'][0] + attrs['bndbox'][2])),
                    bndbox_node)

    createChildNode(doc, 'ymax',str(int(attrs['bndbox'][1] + attrs['bndbox'][3])),
                    bndbox_node)

    object_node.appendChild(bndbox_node)

    return object_node


# 将documentElement写入XML文件
def writeXMLFile(doc, filename):
    tmpfile = open('tmp.xml', 'w')

    doc.writexml(tmpfile, addindent='' *4, newl='\n', encoding='utf-8')


    tmpfile.close()

    # 删除第一行默认添加的标记

    fin = open('tmp.xml')
    # print(filename)
    fout = open(filename, 'w')
    # print(os.path.dirname(fout))

    lines = fin.readlines()

    for line in lines[1:]:

        if line.split():
            fout.writelines(line)

            # new_lines =''.join(lines[1:])

        # fout.write(new_lines)

    fin.close()

    fout.close()


if __name__ == "__main__":
    ##读取图片列表
    img_path ="/home/ouc/data1/liuhongzhi/yolo2-pytorch/datasets/COCO/VOC2007/JPEGImages/"
    fileList = os.listdir(img_path)
    if fileList == 0:
        os._exit(-1)

    with open("COCO_train.json", "r") as f:
        ann_data = json.load(f)

    current_dirpath =os.path.dirname(os.path.abspath('__file__'))

    if not os.path.exists(_ANNOTATION_SAVE_PATH):
        os.mkdir(_ANNOTATION_SAVE_PATH)

        # if not os.path.exists(_IMAGE_COPY_PATH):
    #    os.mkdir(_IMAGE_COPY_PATH)

    for imageName in fileList:

        saveName =imageName.strip(".jpg")
        print(saveName)
        # pos =fileList[xText].rfind(".")
        # textName =fileList[xText][:pos]

        # ouput_file = open(_TXT_PATH +'/' + fileList[xText])
        # ouput_file =open(_TXT_PATH)

        # lines = ouput_file.readlines()

        xml_file_name =os.path.join(_ANNOTATION_SAVE_PATH, (saveName + '.xml'))
        # withopen(xml_file_name,"w") as f:
        #     pass

        img =cv2.imread(os.path.join(img_path, imageName))
        print(os.path.join(img_path,imageName))
        # cv2.imshow(img)
        height, width, channel =img.shape
        print(height, width, channel)

        my_dom = xml.dom.getDOMImplementation()

        doc = my_dom.createDocument(None,_ROOT_NODE, None)

        # 获得根节点
        root_node = doc.documentElement

        # folder节点

        createChildNode(doc, 'folder',_FOLDER_NODE, root_node)

        # filename节点

        createChildNode(doc, 'filename',saveName + '.jpg', root_node)

        # source节点

        source_node =doc.createElement('source')

        # source的子节点

        createChildNode(doc, 'database',_DATABASE_NAME, source_node)

        createChildNode(doc, 'annotation',_ANNOTATION, source_node)

        createChildNode(doc, 'image','flickr', source_node)

        createChildNode(doc, 'flickrid','NULL', source_node)

        root_node.appendChild(source_node)

        # owner节点

        owner_node = doc.createElement('owner')

        # owner的子节点

        createChildNode(doc, 'flickrid','NULL', owner_node)

        createChildNode(doc, 'name',_AUTHOR, owner_node)

        root_node.appendChild(owner_node)

        # size节点

        size_node =doc.createElement('size')

        createChildNode(doc, 'width',str(width), size_node)

        createChildNode(doc, 'height',str(height), size_node)

        createChildNode(doc, 'depth',str(channel), size_node)

        root_node.appendChild(size_node)

        # segmented节点

        createChildNode(doc, 'segmented',_SEGMENTED, root_node)

        for ann in ann_data:
            imgName ="COCO_train2014_" + str(ann["filename"])
            cname=saveName;
            if (saveName == imgName ):
                # object节点
                object_node =createObjectNode(doc, ann)
                root_node.appendChild(object_node)

            else:
                continue

                # 构建XML文件名称

        print(xml_file_name)

        # 创建XML文件

        # createXMLFile(attrs, width,height, xml_file_name)

        # # 写入文件
        #
        writeXMLFile(doc, xml_file_name)

最后得到train.txt文件，里面是所有训练图片的名字，需要删除路径和后缀，只保留图片名。

find ./JPEGImages -name '*.jpg'  > train.txt

3、 Cityscapes数据集

Cityscapes数据集则是由奔驰主推，提供无人驾驶环境下的图像分割数据集，用于评估视觉算法在城区场景语义理解方面的性能。图像Translation算法常用，如Pix2pix和CycleGAN。

Cityscapes包含50个欧洲城市不同场景、不同背景、不同季节的街景的33类标注物体，包括：{'unlabeled'=0 , 'ego vehicle'=1 , 'rectification border'=2 , 'out of roi'= 3 , 'static'=4 , 'dynamic'=5 , 'ground'=6 ,'road'=7 ,'sidewalk'=8 ,parking'=9 ,'rail track'=10 ,'building'=11 ,'wall'=12 ,'fence'=13 , 'guard rail'=14 ,'bridge'=15 ,'tunnel'=16 ,'pole'=17 ,'polegroup'=18 , 'traffic light'=19 ,'traffic sign'=20 , 'vegetation'=21 , 'terrain'=22 ,'sky'=23 , 'person'=24 , 'rider'=25 , 'car'=26 ,'truck'=27 , 'bus'=28 ,'caravan'=29 ,'trailer'=30 ,'train'=31 ,'motorcycle'=32 , 'bicycle'=33 }，但是在这33个类中，评估时只用到了19个类别，因此训练时将33个类映射为19个类，评估时需要将19个类又映射回33个类上传评估服务器。这个数据需要注册账号才能下载。

Cityscapes数据集共有fine和coarse两套评测标准，前者提供5000张精细标注的图像，后者提供5000张精细标注外加20000张粗糙标注的图像，用PASCAL VOC标准的 intersection-over-union （IoU）得分来对算法性能进行评价。 5000张精细标注的图片分为训练集2975张图片，验证集有500张图片，而测试集有1525张图片，测试集不对外公布，需要将预测结果上传到评估服务器才能计算mIoU值。