Yolov5 creates and trains its own data set to realize mask wearing detection

Be careful

• Need to be ahead of time Anaconda Set up well yolov5 Environment , There is no explanation here .

Data sets

Mask identification data set download
Download this file

After downloading, you can extract . Although mixed together , But that's okay .

Data set preprocessing

Go to Roboflow Create an account

Add workspace

Select public

Create a new project

Fill in the name casually

Upload the data set we downloaded



Don't worry about him

Wait for the upload to finish


End upload







export
Note that export TXT Under the YOLO v5 PyTorch This option , download zip Format .

Get ready to train

These three files are required

train and valid Folders are placed in and yolov5 In the same directory

modify data.yaml

0 Change it to no-mask（ Don't wear a mask ）,1 Change it to mask（ Wear a mask ）


hold data.yaml Put it in yolov5 Of data Under the folder

Start training

stay anaconda Execute... In a virtual environment , Notice to change to your corresponding path .

python train.py --data data.yaml

End of training

Can be in yolo Find the model file after training under the path of

Test model file

add --weights Parameters can be detected by selecting the corresponding model file .

python detect.py  --weights runs/exp/weights/best.pt

Detection effect

Reference material

https://blog.csdn.net/lynxzong/article/details/86647805
https://blog.csdn.net/sinat_28371057/article/details/120598220
https://github.com/ultralytics/yolov5
https://zhuanlan.zhihu.com/p/269587479
https://github.com/ultralytics/yolov5/wiki/Train-Custom-Data
https://public.roboflow.com/
https://codeantenna.com/a/FtLN7QmRtk
http://www.jishudaxue.com/cblog/python/9489.html
https://github.com/ultralytics/yolov5/issues/5086‘
https://www.jb51.net/article/211043.htm
https://blog.csdn.net/qq_36756866/article/details/109111065
https://blog.csdn.net/HJZ11/article/details/109838775