训练检测口罩模型

下载项目

https://github.com/Megvii-BaseDetection/YOLOX

打开项目

安装环境

pip install -r requirements.txt

如果报错

pip install -c esri pycocotools

第一步下载数据集

在datasets文件夹下新建VOCdevkit文件夹并将下载的data文件夹放进去，并改名为VOC2020。
在VOC2020文件夹下新建ImageSets文件夹，在建Main文件夹。

下载make_train_and_test.py文件并运行

import os
import random
 
train_percent = 0.8
test_percent = 0.2
xmlfilepath = 'VOCdevkit/VOC2020/Annotations'
txtsavepath = 'VOCdevkit/VOC2020/ImageSets'
total_xml = os.listdir(xmlfilepath)
 
num = len(total_xml)
list = range(num)
tr = int(num * train_percent)
te = int(num * test_percent)
train = random.sample(list, tr)
test = random.sample(list, te)
 
ftest = open('VOCdevkit/VOC2020/ImageSets/Main/test.txt', 'w')
ftrain = open('VOCdevkit/VOC2020/ImageSets/Main/train.txt', 'w')

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

生成训练样本和测试样本的下标

第二步修改代码

yolox_voc_s.py相关代码改为

self.num_classes = 2

然后

voc_classea.pu改为

VOC_CLASSES = (
    "no_masked",
    "masked",
)

voc.py修改

训练模型

python tools/train.py -f ./exps/example/yolox_voc/yolox_voc_s.py -d 1 -b 4 --fp 16 -o

会报错
Traceback (most recent call last):
File "./tools/train.py", line 13, in

from yolox.core import launch

ModuleNotFoundError: No module named 'yolox'

路径错误
在tools/train.py开头加上开头路径

import sys
sys.path.append(r"D:\PycharmProjects\YOLOX-main")

报错 DataLoader worker

yolox_voc_s.py添加

 self.data_num_workers = 0

报错OMP: Error #15: Initializing libiomp5md.dll, but found libiomp5md.dll already initialized.

环境冲突
tools/train.py添加代码

import os
os.environ['KMP_DUPLICATE_LIB_OK'] = 'True'

报错 ERROR | yolox.core.launch:98 - An error has been caught in function 'launch', process 'MainProcess' (22360), thread 'MainThread' (18700):

修改
voc.py