1.voc转yolo、 数据集切分
# 缺陷坐标xml转txt
import os
import xml.etree.ElementTree as ET
import os
import random
import cv2
# classes = ["背景","边异常","角异常","白色点瑕疵","浅色块瑕疵","深色点块瑕疵","光圈瑕疵"] # 输入类别名称,必须与xml标注名称一致
# classes1 = ["0","1","2"]
classes2 = ["normal", "abnormal"]
# def convert(size, box):
# print(size, box)
# dw = 1. / size[0]
# dh = 1. / 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)
def convert(size, box):
print(size, box)
dw = 1. / size[0]
dh = 1. / 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, 0.5, w, 1.0)
def convert_annotation(image_id):
print(image_id)
in_file = open(r'./data/Annotations/%s' % (image_id), 'rb') # 读取xml文件路径
out_file = open('./data/labels/%s.txt' % (image_id.split('.')[0]), 'w') # 需要保存的txt格式文件路径
tree = ET.parse(in_file)
root = tree.getroot()
# size = root.find('size')
# w = int(size.find('width').text)
# h = int(size.find('height').text)
img_=cv2.imread(os.path.join('./data/images/', image_id.replace('xml', 'jpg')))
w, h = img_.shape[1], img_.shape[0]
for obj in root.iter('object'):
cls = obj.find('name').text
# if cls not in classes:
# continue
# cls_id = classes.index(cls)
# if cls in classes1:
# cls_id = classes1.index(cls)
if cls in classes2:
cls_id = classes2.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)]
#判断框是否出界
if b[0]>=w or b[2]>=h:
continue
if b[1]>=w:
b[1]=w
if b[3]>=h:
b[3]=h
bb = convert((w, h), b)
out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
image_ids_train = os.listdir('./data/Annotations') # 读取xml文件名索引
for image_id in image_ids_train:
print(image_id)
convert_annotation(image_id)
trainval_percent = 0.2 # 可自行进行调节
train_percent = 1
xmlfilepath = './data/images'
total_xml = os.listdir(xmlfilepath)
num = len(total_xml)
list = range(num)
tv = int(num * trainval_percent)
tr = int(tv * train_percent)
trainval = random.sample(list, tv)
train = random.sample(trainval, tr)
ftest = open('./data/test.txt', 'w')
ftrain = open('./data/train.txt', 'w')
for i in list:
name = total_xml[i] + '\n'
if i in trainval:
if i in train:
# ftest.write('data/images/' + name)
ftest.write('data/images/'+name)
else:
# ftrain.write('data/images/' + name)
ftrain.write('data/images/'+name)
ftrain.close()
ftest.close()
2、自定义yaml
train: ./data/train.txt # 上面我们生成的train,根据自己的路径进行更改
val: ./data/test.txt # 上面我们生成的test
# number of classes
nc: 1 #训练的类别
# class names
names: ["normal"]