vscode调试方法&&跑yolov5心得_vscode中使用yolov5

• vscode终端，训练运行train.py.检测验证模型运行detect.py.
• 注意vscode运行环境是下面环境相当于终端的，上面相当于vscode自己的。
• 终端直接输入命令运行：好处可以输入参数，但设断点不会中断。vscode点击上面直接调试，会中断，但是不能输入参数。方法：设置，在默认default地方写。
• yolov5s预训练模型轻于yolov5。选用小模型多数据 要好于 大模型大数据和大模型小数据。 大模型小数据收敛很快但是过拟合，结果往往差相比小模型。
• 训练前要对label提取，划分训练集等预处理。

# coding:utf-8
 
import os
import random
import argparse
 """
 根据label的所有xml划分训练，测试，验证集合
 输入所有xml路径，输出三个txt（里面是图片名，每一行是一个图片）
 """
parser = argparse.ArgumentParser()
# xml文件的地址，根据自己的数据进行修改 xml一般存放在Annotations下
parser.add_argument('--xml_path', default='data/mydata/Annotations', type=str, help='input xml label path')
# 数据集的划分，地址选择自己数据下的ImageSets/Main
parser.add_argument('--txt_path', default='data/mydata/dataSet', type=str, help='output txt label path')
opt = parser.parse_args()

def split_data_indices(total_indices):
    random.shuffle(total_indices)
    total_samples = len(total_indices)
    train_ratio, test_ratio = 0.7, 0.2

    train_end = int(total_samples * train_ratio)
    test_end = int(total_samples * (train_ratio + test_ratio))

    train_indices = total_indices[:train_end]
    test_indices = total_indices[train_end:test_end]
    val_indices = total_indices[test_end:]

    return train_indices, test_indices, val_indices


xmlfilepath = opt.xml_path
txtsavepath = opt.txt_path
total_xml = os.listdir(xmlfilepath)
if not os.path.exists(txtsavepath):
    os.makedirs(txtsavepath)
 
num = len(total_xml)
list_index = list(range(num))
train_indices, test_indices, val_indices = split_data_indices(list_index)
 #'data/mydata/dataSet下创建划分数据集的图片名字
file_trainval = open(txtsavepath + '/trainval.txt', 'w')
file_train = open(txtsavepath + '/train.txt', 'w')
file_val = open(txtsavepath + '/val.txt', 'w')
 
for i in range(num):
    name = total_xml[i][:-4] + '\n'
    if i in train_indices:
        file_trainval.write(name)
    elif i in test_indices:
        file_train.write(name)
    else:
        file_val.write(name)

 
file_trainval.close()
file_train.close()
file_val.close()

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59

# -*- coding: utf-8 -*-
import xml.etree.ElementTree as ET
import os
from os import getcwd
 """
 作用 1：把xml每一个图片对象转换成txt形式（每个txt表示一张图片，
 txt里面每行表示图片里的一个object。
 2：把之前训练，测试，验证集合三个txt里面图片名称改成全局路径
 """
sets = ['train', 'val', 'trainval']
classes = ["red", "green", "yellow", "off"]  # 改成自己的类别
abs_path = os.getcwd()
print(abs_path)

 
def convert(size, box):
    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 convert_annotation(image_id):
    if image_id == '': return
    in_file = open('data/mydata/Annotations/%s.xml' % (image_id), encoding='UTF-8')
    out_file = open('data/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
        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))
        b1, b2, b3, b4 = b
        # 标注越界修正
        if b2 > w:
            b2 = w
        if b4 > h:
            b4 = h
        b = (b1, b2, b3, b4)
        bb = convert((w, h), b)
        out_file.write(str(cls_id) + " " + " ".join([str(a) for a in bb]) + '\n')
 
wd = getcwd()
for image_set in sets:
    if not os.path.exists('data/mydata/labels/'):
        os.makedirs('data/mydata/labels/')
    image_ids = open('data/mydata/dataSet/%s.txt' % (image_set)).read().strip().split("\n")
    list_file = open('data/mydata/%s.txt' % (image_set), 'w')
    for image_id in image_ids:
        list_file.write(abs_path + '/data/mydata/JPEGImages/%s.jpg\n' % (image_id))
        convert_annotation(image_id)
    list_file.close()
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67

list_file.write(abs_path + ‘/data/mydata/JPEGImages/%s.jpg\n’ % (image_id))注意jpg格式写死了，可能会有图片为png,但这里为jpg的bug。
全修改为jpg脚本：

# import os

# def rename_files_in_folder(folder_path):
#     for filename in os.listdir(folder_path):
#         if filename.endswith(".png"):
#             base = os.path.splitext(filename)[0]
#             os.rename(os.path.join(folder_path, filename), os.path.join(folder_path, base + ".jpg"))

# # 使用方法
# # replace 'your_folder_path' with the path of your folder
# rename_files_in_folder(r'E:\panda_ws-main\traffic_light\data\mydata\JPEGImages')
1
2
3
4
5
6
7
8
9
10
11

有时候txt里记录了1.jpg，但JPEGIMG里却缺少这张图片，报错。修改代码

# 复制图像到另一个文件夹
# 文件所在文件夹
import os
import shutil
with open(r"E:\panda_ws-main\traffic_light\data\mydata\dataSet\val.txt", 'r') as f:
    for line in f.readlines():
        line = line.strip('\n')
        print(line)
        if os.path.exists(line):
            shutil.copy(line, r'E:\panda_ws-main\traffic_light\data\images')
1
2
3
4
5
6
7
8
9
10

其余按照网上流程都可以运行。