Please read my article here, which has an explanation about how an attendance system using Face Recognition, given a video feed works and read this article here to know how to implement it using this repository.
- Clone this repo.
- Download retinaface and insightface models from here
- Make a directory "models" inside Face_detection folder and extract "retinaface-R50.zip" in a folder inside "models" with name "retinaface" and extract "insightface.zip" in the "models" folder only. So now you will have two directories inside models, namely insightface and retinaface.
- In the root folder there is a file environment.yml which we will use to create a conda environment byt executing :
conda env create -n face_recog -f environment.yml
- Start the redis server in which we will store the vectors of registered faces by running :
redis-server
- Activate the conda environment created in previous step by running :
conda activate face_recog
- Register all the different person using register.py. To know about how to structure the data for registering please go through my implementation article
python register.py -p path/to/folder/ -db 0
- Run the following command to infer on a video feed after registering all the people :
python infer.py -in path/to/video -out to/save/path -db 0