如何利用u版YOLOv3源码(Pytorch)训练自己的数据集

最新推荐文章于 2023-03-22 17:56:49 发布

xyzjy123

最新推荐文章于 2023-03-22 17:56:49 发布

阅读量1k

点赞数 4

文章标签：深度学习 pytorch

本文链接：https://blog.csdn.net/xyzjy123/article/details/128176863

版权

1、在github官网上下载YOLOv3的源码，我下载的是YOLOv3-4.0版本的

2、对YOLOv3-4.0压缩包解压，解压完就是这样的

3、利用Pycharm打开YOLOv3-4.0文件

新建一个目录，命名为mydata。

在mydata目录中，一共创建5个子目录，分别为Annotations(用于存放.xml标注文件)、JPEGImages(用于存放训练图片)、images(用于存放训练图片)、labels(空子目录)、ImageSets(空子目录)。

4、创建make_txt.py文件，放到和mydata同级目录下

# -*- coding:utf-8 -*
import os
import random

trainval_percent = 0.1
train_percent = 0.9
xmlfilepath = 'mydata/Annotations'
txtsavepath = 'mydata/ImageSets'
total_xml = os.listdir(xmlfilepath)

num = len(total_xml)  #统计所有的标注文件
list = range(num)
tv = int(num * trainval_percent)  # 设置训练验证集的数目
tr = int(tv * train_percent)      # 设置训练集的数目
trainval = random.sample(list, tv)
train = random.sample(trainval, tr)

# txt 文件写入的只是xml 文件的文件名（数字），没有后缀，如下图。
ftrainval = open('mydata/ImageSets/trainval.txt', 'w')
ftest = open('mydata/ImageSets/test.txt', 'w')
ftrain = open('mydata/ImageSets/train.txt', 'w')
fval = open('mydata/ImageSets/val.txt', 'w')

for i in list:
    name = total_xml[i][:-4] + '\n'
    if i in trainval:
        ftrainval.write(name)
        if i in train:
            ftest.write(name)
        else:
            fval.write(name)
    else:
        ftrain.write(name)

ftrainval.close()
ftrain.close()
fval.close()
ftest.close()

运行make_txt.py文件，在ImageSets子目录中生成4个.txt文件，分别是test.txt、train.txt、trainval.txt、val.txt。

5、创建voc_label.py文件，放到和mydata同级目录下

# -*- coding:utf-8 -*
import xml.etree.ElementTree as ET
import pickle
import os
from os import listdir, getcwd
from os.path import join

sets = ['train', 'test', 'val']
classes = ['CW1S', 'CW2S','CW3S','CW4S','CW5S','CW6S','CW7S','CW8S','CW9S','CW10S',
           'CW10S','CW2M','CW3M','CW4M','CW5M','CW6M','CW7M','CW8M','CW9M','CW10M',
           'CW1H','CW1H','CW3H','CW4H','CW5H','CW6H','CW7H','CW8H','CW9H','CW10H']

def convert(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_annotation(image_id):
    in_file = open('mydata/Annotations/%s.xml' % (image_id),encoding='gb18030')
    out_file = open('mydata/labels/%s.txt' % (image_id), 'w')
    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'):
        difficult = obj.find('difficult').text
        cls = obj.find('name').text
        if cls not in classes or int(difficult) == 1:
            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()
print(wd)
for image_set in sets:
    if not os.path.exists('mydata/labels/'):
        os.makedirs('mydata/labels/')
    image_ids = open('mydata/ImageSets/%s.txt' % (image_set)).read().strip().split()
    list_file = open('mydata/%s.txt' % (image_set), 'w')
    for image_id in image_ids:
        list_file.write('mydata/JPEGImages/%s.jpg\n' % (image_id))
        convert_annotation(image_id)
    list_file.close()

修改第9行的classes，存放自己数据集的类别