1
GPU=0
2
CUDNN=0
3
OPENCV=0
4
OPENMP=0
5
DEBUG=0

1
classes = ${num_classes}
2
train = ${path-to-trainlist.txt}
3
valid = ${path-to-vallist.txt}
4
backup = ${path-to-backup-dir}
5
names = ${path-to-.names-file}

1
import os
2
import xml.etree.ElementTree as ET
3

4
def convert(size, box):
5
    dw = 1. / size[0]
6
    dh = 1. / size[1]
7
    x = (box[0] + box[1]) / 2.0
8
    y = (box[2] + box[3]) / 2.0
9
    w = box[1] - box[0]
10
    h = box[3] - box[2]
11
    x = x * dw
12
    w = w * dw
13
    y = y * dh
14
    h = h * dh
15
    return (x, y, w, h)
16

17

18
def convert_annotation(input_list, output_path, classes):
19
    for xml_file in input_list:
20
        image_id = os.path.splitext(os.path.basename(xml_file))[0]
21
        out_file = open(os.path.join(output_path, image_id + '.txt'), 'w')
22
        tree = ET.parse(xml_file)
23
        root = tree.getroot()
24
        size = root.find('size')
25
        w = int(size.find('width').text)
26
        h = int(size.find('height').text)
27

28
        for obj in root.iter('object'):
29
            difficult = obj.find('difficult').text
30
            cls = obj.find('name').text
31
            if cls not in classes or int(difficult) == 1:
32
                continue
33
            cls_id = classes.index(cls)
34
            xmlbox = obj.find('bndbox')
35
            b = (float(xmlbox.find('xmin').text), float(xmlbox.find('xmax').text), float(xmlbox.find('ymin').text),
36
                 float(xmlbox.find('ymax').text))
37
            bb = convert((w, h), b)
38
            out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
39

40
classes = []  # the classes name
41
images_path = "../dataset5/images"
42
xml_path = "../dataset5/labels_xml"
43
tar_images_path = "../dataset5/images"
44
tar_label_path = "../dataset5/labels_txt"
45
if not os.path.exists(tar_images_path):
46
    os.makedirs(tar_images_path)
47
if not os.path.exists(tar_label_path):
48
    os.makedirs(tar_label_path)
49

50
xml_ids = [os.path.splitext(f)[0] for f in os.listdir(xml_path) if f.endswith('.xml')]
51
image_ids = [os.path.splitext(f)[0] for f in os.listdir(images_path)
52
                if f.endswith('.jpg') or f.endswith('JPG') or f.endswith('jpeg') or f.endswith('JPEG')]
53
assert xml_ids == image_ids  # if xml_ids != immage_ids, than annotations are not complete!
54

55
xmls_path = [os.path.join(xml_path, f) for f in os.listdir(xml_path) if f.endswith('.xml')]
56

57
convert_annotation(xmls_path, tar_label_path, classes)
58

59
images_path = [os.path.join(images_path, f) for f in os.listdir(images_path)
60
                if f.endswith('.jpg') or f.endswith('JPG') or f.endswith('jpeg') or f.endswith('JPEG')]
61

62
# for image in images_path:
63
#     shutil.copy(image, tar_images_path)

1
import os
2
import shutil
3
import random
4

5
def copy_images_and_labels(src_path, dst_path):
6
    os.makedirs(f'../{dst_path}/images', exist_ok=True)
7
    os.makedirs(f'../{dst_path}/labels', exist_ok=True)
8

9
    for dir in ['train', 'val']:
10
        os.makedirs(f'../{dst_path}/images/' + dir, exist_ok=True)
11
        os.makedirs(f'../{dst_path}/labels/' + dir, exist_ok=True)
12

13
    for image_id in image_ids:
14
        shutil.copy(f"../{src_path}/images/{image_id}.jpg",
15
                    f"../{dst_path}/images/train/{image_id}.jpg")
16
        shutil.copy(f"../{src_path}/labels/{image_id}.txt",
17
                    f"../{dst_path}/labels/train/{image_id}.txt")
18

19
    for image_id in val_ids:
20
        shutil.copy(f"../{src_path}/images/{image_id}.jpg",
21
                    f"../{dst_path}/images/val/{image_id}.jpg")
22
        shutil.copy(f"../{src_path}/labels/{image_id}.txt",
23
                    f"../{dst_path}/labels/val/{image_id}.txt")
24

25
src_path = "dataset_all"
26
dst_path = "dataset6"
27
image_ids = [os.path.splitext(f)[0] for f in os.listdir(f'../{src_path}/labels') if
28
                f.endswith('.txt')]
29

30
# Shuffle the image ids and split into train and val
31
random.shuffle(image_ids)
32
split = int(0.1 * len(image_ids))
33
val_ids = image_ids[:split]
34
train_ids = image_ids[split:]
35
copy_images_and_labels(src_path, dst_path)

1
import os
2

3
# 指定需要遍历的文件夹
4
folder_path = '../dataset_darknet/images/'
5

6
# 打开文件，准备写入
7
with open('trainlist.txt', 'w') as file:
8
    # 遍历文件夹
9
    for root, dirs, files in os.walk(folder_path):
10
        # 遍历所有文件
11
        for name in files:
12
            # 检查文件是否为.jpg文件
13
            if name.endswith(".jpg"):
14
                # 获取文件的绝对路径
15
                abs_path = os.path.abspath(os.path.join(root, name))
16
                # 写入文件
17
                file.write(abs_path + '\n')