很多小伙伴可能在找到数据集后发现他的label文件是xml格式的,对于xml格式的文件是不能在YOLO中进行识别的,需要转化为txt格式文件。笔者在尝试了一些办法后,找到了一个比较好用的,分享如下:
一.xml格式转txt
注意需要提前建立好存放txt标注的label文件。
import xml.etree.ElementTree as ET
from os import listdir, getcwd
import glob
import cv2
# xml格式文件注释
#
# <object>
# <name>person</name>
# <pose>Unspecified</pose>
# <truncated>0</truncated>
# <difficult>0</difficult>
# <bndbox>
# <xmin>476</xmin>
# <ymin>137</ymin>
# <xmax>498</xmax>
# <ymax>156</ymax>
# </bndbox>
# </object>
classes = ["person"] #<name>person</name>
def convert(size, box):#新增对框框范围的判断,防止0作为被除数
if size[0]==0:
dw=size[0]
else:
dw = 1.0 / size[0]
if size[1] == 0:
dw = size[1]
else:
dh = 1.0 / size[1]
x = (box[0] + box[1]) / 2.0
y = (box[2] + box[3]) / 2.0
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)
return ('%.6f'%x,'%.6f'%y,'%.6f'%w,'%.6f'%h)
def convert_annotation(image_name,image_path):
print(f"Processing {image_name}")
f = open('./Annotations/' + image_name[:-3] + 'xml', encoding="utf8") # xml文件存放文件夹路径(Annoatations文件夹下)
out_file = open('./label/' + image_name[:-3] + 'txt', 'w') # 这里表示的是存放转换后的txt文件路径,所以需要我们自己先建立好label文件
xml_text = f.read()
root = ET.fromstring(xml_text)
f.close()
size = root.find('size')
#填补图片高宽缺失
img = cv2.imread(image_path)
sz = img.shape
w=int(sz[1])
h=int(sz[0])
# 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:
print(cls)
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()
if __name__ == '__main__':
for image_path in glob.glob("./JPEGImages/*.jpg"): # 放图片的文件夹,根据自己图片的类型替换JPG
image_name = image_path.split('\\')[-1]
convert_annotation(image_name,image_path)
二.数据集划分
只需要修改最后面main的部分即可。
import os
import shutil
import random
random.seed(0)
def split_data(file_path, new_file_path, train_rate, val_rate, test_rate):
eachclass_image = []
for image in os.listdir(file_path):
eachclass_image.append(image)
total = len(eachclass_image)
random.shuffle(eachclass_image)
train_images = eachclass_image[0:int(train_rate * total)]
val_images = eachclass_image[int(train_rate * total):int((train_rate + val_rate) * total)]
test_images = eachclass_image[int((train_rate + val_rate) * total):]
for image in train_images:
print(image)
old_path = file_path + '/' + image
new_path1 = new_file_path + '/' + 'train' + '/' + 'images'
if not os.path.exists(new_path1):
os.makedirs(new_path1)
new_path = new_path1 + '/' + image
shutil.copy(old_path, new_path)
new_name = os.listdir(new_file_path + '/' + 'train' + '/' + 'images')
for im in new_name:
old_xmlpath = txtpath + '/' + im[:-3] + 'txt'
new_xmlpath1 = new_file_path + '/' + 'train' + '/' + 'labels'
if not os.path.exists(new_xmlpath1):
os.makedirs(new_xmlpath1)
new_xmlpath = new_xmlpath1 + '/' + im[:-3] + 'txt'
shutil.copy(old_xmlpath, new_xmlpath)
for image in val_images:
old_path = file_path + '/' + image
new_path1 = new_file_path + '/' + 'val' + '/' + 'images'
if not os.path.exists(new_path1):
os.makedirs(new_path1)
new_path = new_path1 + '/' + image
shutil.copy(old_path, new_path)
new_name = os.listdir(new_file_path + '/' + 'val' + '/' + 'images')
for im in new_name:
old_xmlpath = txtpath + '/' + im[:-3] + 'txt'
new_xmlpath1 = new_file_path + '/' + 'val' + '/' + 'labels'
if not os.path.exists(new_xmlpath1):
os.makedirs(new_xmlpath1)
new_xmlpath = new_xmlpath1 + '/' + im[:-3] + 'txt'
shutil.copy(old_xmlpath, new_xmlpath)
for image in test_images:
old_path = file_path + '/' + image
new_path1 = new_file_path + '/' + 'test' + '/' + 'images'
if not os.path.exists(new_path1):
os.makedirs(new_path1)
new_path = new_path1 + '/' + image
shutil.copy(old_path, new_path)
new_name = os.listdir(new_file_path + '/' + 'test' + '/' + 'images')
for im in new_name:
old_xmlpath = txtpath + '/' + im[:-3] + 'txt'
new_xmlpath1 = new_file_path + '/' + 'test' + '/' + 'labels'
if not os.path.exists(new_xmlpath1):
os.makedirs(new_xmlpath1)
new_xmlpath = new_xmlpath1 + '/' + im[:-3] + 'txt'
shutil.copy(old_xmlpath, new_xmlpath)
if __name__ == '__main__':
file_path = "JPEGImages" #这里放你存JPG照片的地址
txtpath = 'labels' #这里放你存label的地址
new_file_path = "datases" #这里填你打算新建立的文件夹的名字,在下面的函数中可以更改你对数据集中train,val and test的划分。
split_data(file_path, new_file_path, train_rate=0.7, val_rate=0.1, test_rate=0.2)
按照上述步骤,就可以得到自己训练的数据集了。