深度学习和目标检测系列教程 18-300：关于yolovoc格式标签转化问题

Posted 2022-01-21 刘润森！

@Author：Runsen

PASCAL VOC(The PASCAL Visual Object Classes)是一个世界级的计算机视觉挑战赛，第一届比赛在2005年举办，随后一年举办一次，直到2012年最后一次。PASCAL的全称是Pattern Analysis, Statistical modelling and ComputAtional Learning，官方地址是 http://host.robots.ox.ac.uk/pascal/VOC/

一个典型的xml标签文件如下

<annotation>
	<folder>images</folder>
	<filename>003707.jpg</filename>
	<path>G:\\sz210628\\images\\003707.jpg</path>
	<source>
		<database>Unknown</database>
	</source>
	<size>
		<width>1920</width>
		<height>1080</height>
		<depth>3</depth>
	</size>
	<segmented>0</segmented>
	<object>
		<name>航空器头</name>
		<pose>Unspecified</pose>
		<truncated>0</truncated>
		<difficult>0</difficult>
		<bndbox>
			<xmin>827</xmin>
			<ymin>224</ymin>
			<xmax>916</xmax>
			<ymax>333</ymax>
		</bndbox>
	</object>
</annotation>

• folder：所属文件夹
• filename：文件名
• database：数据库名
• annotation：标记文件格式
• size：图像尺寸，width宽、height高，depth通道数目
• segmented：分割
• object: 表示一个目标，name标签名、pose拍摄角度，有front、rear、left、right和unspecified、truncated是否被截断也就是图片中是否包含完整目标、
• difficult检测难易程度，1表示是、0表示否
• bndbox：目标所在的位置，用xmin、ymin、xmax、ymax来表示

yolo要求每张图片对应一个txt标注文件。

每行一个标注对象，分别是：class_id x y width height。
class_id: 从0到（classes-1）的整数。

x y width height: 相对于图像的宽度和高度的浮点值，它可以等于（0.0到1.0]。例如：x=pixel_x/image_width, height=pixel_height/image_height

x y: 是标注对象的中心点。

0 0.750000 0.501111 0.315000 0.993333

下面脚本读取 PascalVOC xml 文件，并将它们转换为 YOLO txt 文件。

#coding=gbk
import os
import xml.etree.ElementTree as ET


# xml VOC标注的类   
classes=['person','rider','car','bus','truck','bike','motor','tl_green','tl_red','tl_yellow','tl_none','t_sign','train']

def convert(size,box):
	#  VOC to YOLO format
	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 converting_annotation(ann_file,yolo_out_str):
	# ann_file 标注的所有xml文件
	# out_str yolo格式的txt 文件夹
	for ann in ann_file:
		try:
			txt_file= ann.split('.')[0]+'.txt'
			in_file=open(ann_dir + "/" + ann,encoding="utf-8")
			out_file =open(yolo_out_str + "\\\\"+txt_file, 'w',encoding="utf-8")
			tree=ET.parse(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'):
				cls=obj.find('name').text
				if cls not in classes:
					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))
				data = convert((w,h), b)
				out_file.write( yolo_out_str + str(cls_id) + " " + " ".join([str(a) for a in data]) + '\\n')
		except  Exception as e:
			print(ann)
			continue


if __name__ == '__main__':
	ann_dir= "Annotations"
	ann_file=os.listdir(ann_dir)
	ann_file.sort()
	# "label" yolo txt 存放的位置
	converting_annotation(ann_file,"label")

下面脚本读取 YOLO txt 文件，并将它们转换为PascalVOC xml 文件。

#coding=gbk
import csv
import os
from lxml import etree
from PIL import Image

IMG_PATH = r"images\\trains"
fw = os.listdir(IMG_PATH)

save_path = r'labels\\xml' # keep it blank

# yolo txt_folder
txt_folder = r'labels\\trains'


labels=['person','rider','car','bus','truck','bike','motor','tl_green','tl_red','tl_yellow','tl_none','t_sign','train']



def csvread(fn):
    with open(fn, 'r') as csvfile:
        list_arr = []
        reader = csv.reader(csvfile, delimiter=' ')

        for row in reader:
            list_arr.append(row)
    return list_arr


def convert_label(txt_file):
    global label
    for i in range(len(labels)):
        if txt_file[0] == str(i):
            label = labels[i]
            return label

    return label



def extract_coor(txt_file, img_width, img_height):
    x_rect_mid = float(txt_file[1])
    y_rect_mid = float(txt_file[2])
    width_rect = float(txt_file[3])
    height_rect = float(txt_file[4])

    x_min_rect = ((2 * x_rect_mid * img_width) - (width_rect * img_width)) / 2
    x_max_rect = ((2 * x_rect_mid * img_width) + (width_rect * img_width)) / 2
    y_min_rect = ((2 * y_rect_mid * img_height) -
                  (height_rect * img_height)) / 2
    y_max_rect = ((2 * y_rect_mid * img_height) +
                  (height_rect * img_height)) / 2

    return x_min_rect, x_max_rect, y_min_rect, y_max_rect


for line in fw:
    root = etree.Element("annotation")

    # try debug to check your path
    img_style = IMG_PATH.split('/')[-1]
    img_name = line
    image_info = IMG_PATH + "\\\\" + line
    img_txt_root = txt_folder + "\\\\" + line[:-4]
    txt = ".txt"

    txt_path = img_txt_root + txt
    txt_file = csvread(txt_path)
    ######################################
    # read the image  information
    img_size = Image.open(image_info).size

    img_width = img_size[0]
    img_height = img_size[1]
    img_depth = Image.open(image_info).layers
    ######################################

    folder = etree.Element("folder")
    folder.text = "%s" % (img_style)

    filename = etree.Element("filename")
    filename.text = "%s" % (img_name)

    path = etree.Element("path")
    path.text = "%s" % (IMG_PATH)

    source = etree.Element("source")
    ##################source - element##################
    source_database = etree.SubElement(source, "database")
    source_database.text = "Unknown"
    ####################################################

    size = etree.Element("size")
    ####################size - element##################
    image_width = etree.SubElement(size, "width")
    image_width.text = "%d" % (img_width)

    image_height = etree.SubElement(size, "height")
    image_height.text = "%d" % (img_height)

    image_depth = etree.SubElement(size, "depth")
    image_depth.text = "%d" % (img_depth)
    ####################################################

    segmented = etree.Element("segmented")
    segmented.text = "0"

    root.append(folder)
    root.append(filename)
    root.append(path)
    root.append(source)
    root.append(size)
    root.append(segmented)

    for ii in range(len(txt_file)):
        label = convert_label(txt_file[ii][0])
        x_min_rect, x_max_rect, y_min_rect, y_max_rect = extract_coor(
            txt_file[ii], img_width, img_height)

        object = etree.Element("object")
        ####################object - element##################
        name = etree.SubElement(object, "name")
        name.text = "%s" % (label)

        pose = etree.SubElement(object, "pose")
        pose.text = "Unspecified"

        truncated = etree.SubElement(object, "truncated")
        truncated.text = "0"

        difficult = etree.SubElement(object, "difficult")
        difficult.text = "0"

        bndbox = etree.SubElement(object, "bndbox")
        #####sub_sub########
        xmin = etree.SubElement(bndbox, "xmin")
        xmin.text = "%d" % (x_min_rect)
        ymin = etree.SubElement(bndbox, "ymin")
        ymin.text = "%d" % (y_min_rect)
        xmax = etree.SubElement(bndbox, "xmax")
        xmax.text = "%d" % (x_max_rect)
        ymax = etree.SubElement(bndbox, "ymax")
        ymax.text = "%d" % (y_max_rect)
        #####sub_sub########

        root.append(object)
        ####################################################

    file_output = etree.tostring(root, pretty_print=True, encoding='UTF-8')
    xml_dir = save_path  + "\\\\" + '%s.xml' % (img_name[:-4])
    print(xml_dir)
    ff = open(xml_dir, 'w', encoding="utf-8")
    ff.write(file_output.decode('utf-8'))

在处理 VOC数据集，往往需要从voc数据集获取所有标签的所有类别数，具体的脚本如下所示：

#coding=gbk

import xml.dom.minidom as xmldom
import os

#voc数据集获取所有标签的所有类别数
annotation_path="Annotations"

annotation_names=[os.path.join(annotation_path,i) for i in os.listdir(annotation_path)]

labels = list()
for names in annotation_names:
    xmlfilepath = names
    domobj = xmldom.parse(xmlfilepath)
    # 得到元素对象
    elementobj = domobj.documentElement
    #获得子标签
    subElementObj = elementobj.getElementsByTagName("object")
    for s in subElementObj:
        label=s.getElementsByTagName("name")[0].firstChild.data
        if label not in labels:
            labels.append(label)

print(labels)

在进行yolo模型训练，需要统计每一个类别的数量，来寻找哪些数据标签的样本样例比较少，怎么处理样本不平衡问题？

下面代码是样本类别统计计算的代码。

import os
import xml.etree.ElementTree as ET
from PIL import Image

def parse_obj(xml_path, filename):
    tree = ET.parse(xml_path + filename)
    objects = []
    for obj in tree.findall('object'):
        obj_struct = 
        obj_struct['name'] = obj.find('name').text
        objects.append(obj_struct)
    return objects

def read_image(image_path, filename):
    im = Image.open(image_path + filename)
    W = im.size[0]
    H = im.size[1]
    area = W * H
    im_info = [W, H, area]
    return im_info

if __name__ == '__main__':
    xml_path = 'Annotations\\\\'
    filenamess = os.listdir(xml_path)
    filenames = []
    for name in filenamess:
        name = name.replace('.xml', '')
        filenames.append(name)
    recs = 
    obs_shape = 
    classnames = []
    num_objs = 
    obj_avg = 
    for i, name in enumerate(filenames):
        recs[name] = parse_obj(xml_path, name + '.xml')
    for name in filenames:
        for object in recs[name]:
            if object['name'] not in num_objs.keys():
                num_objs[object['name']] = 1
            else:
                num_objs[object['name']] += 1
            if object['name'] not in classnames:
                classnames.append(object['name'])
    for name in classnames:
        print(':个'.format(name, num_objs[name]))
    print('信息统计算完毕。')