Yolov5 advanced six target tracking environment construction

Be proficient in using labelimg After marking and training identification , You can do some interesting things , For example, identify whether to wear a mask , If you don't bring it, you will be prompted “ Please wear a mask ”.https://download.csdn.net/download/buaaweibin/85802968
Here are the trained mask.pt file , Just use this weight , You can identify whether to wear a mask . You can put mask.pt Put it in weights Under the table of contents , In this way, the wearing of masks can be detected in real time .

python detect.py --weights weights/mask.pt --source 0

Of course, the main content of today is not identification , It is track track .
Use yolov5+deepsort Is the easiest way to implement tracing .
The following are the specific implementation steps ：
Choose a compatible version , Use yolov5-5.0 and
Yolov5_StrongSORT_OSNet-3.0,python3.6
The websites are
https://github.com/ultralytics/yolov5/releases
https://github.com/mikel-brostrom/Yolov5_StrongSORT_OSNet/releases
Pay attention to releases Find the corresponding version . Be careful yolov5-6.0 and 5.0 Of pt Weight files cannot be mixed .
Now let's start to build a familiar environment .

conda create --prefix=D:/PycharmProjects/yoloV5_deepsort/yoloV5deepsort_env python=3.6

If it's too slow , Add the Tsinghua image temporarily when creating the environment , Cancel later

conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pk