[GreenLabs] - Face Recognition

This library written in MXNET framework.

Author: HungLV

Last Update: July 13, 2020

It features:

• Face Detection
• Extract Face Embedding
• Search Face in database

Code:

Documentation:

User-Guide:

Model zoo:

Tech report:

What's new

• [Jul 9] Initial version
• [Jul 13] Make the models can run on GPU/CPU. Integrated rcnn library.

Installation

Make sure conda is installed

pip install green-face-recognition

Getting started

Prepare the config file as yaml type. Take a look this example.

Show the list of models

from face_recognition import models
models.show_avai_models()

Load the models

retina_model = models.build_model('retina-r50',config_path)
arcface = models.build_model('arc-face',config_path)

Make the prediction

# get faces and landmarks
retina_model.detect_fast(img,img.shape,0.8,[1],do_flip=False)

Appendix

Generate the distributions for uploading to server

python3 setup.py sdist bdist_wheel 

Upload to server

python3 -m twine upload --repository pypi dist/*

Reference

1. RetinaFaceModel