Video Face Recognition vid Face Object

This repository is the official implementation of https://www.dbpia.co.kr/Journal/articleDetail?nodeId=NODE09302060.

Requirments

Code was tested in anaconda virtual environment with Python 2.7, CUDA ≥ 8.0

make

conda env create -f videoFR_requrements.yaml

Face Recognition on video

This snipplet will infer your video to recognize registered face.

python videoFR_v4.py \
--display
--back r100
--db dohee_face
--gpu 0
--simal_gpu

Arguments discription

• display : display face recognition results
• back : facerecognition backbone network. [r50/ r100]
• record : store face recognition results
• db : registered face emb vector. if file exists, code will automatically load the db file. if not, code will generate db file
• gpu : gpu index
• simcal_gpu : parameter whether calculate cosine similarity using gpu or not

Demo video

demo.mp4