dlib==19.20.0
opencv_python==4.2.0.34
onnxruntime==1.2.0
fastapi==0.61.0
numpy==1.19.1

Put all target pictures in one directory (data/mask_nomask in this project), one person with two picture masked and none-masked is recommended, the file naming is name.index.jpg

you can download the all weight files from baiduyun passwd：60ri, and put them in data directory

python demo_video.py  --face_db_root data/mask_nomask --input_video_path 0.mp4 --output_video_path output.mp4

uvicorn main:app --reload --host 0.0.0.0

you can upload a picture from http://127.0.0.1:8000/docs,and retrive the face recognition information [name,box,mask or not]

https://blog.csdn.net/jiangpeng59/article/details/107986046

https://github.com/sgrvinod/a-PyTorch-Tutorial-to-Object-Detection https://github.com/AIZOOTech/FaceMaskDetection https://github.com/davisking/dlib