A pure python face recognition system

Introduction

This repo contructs a simple face recognition system based on the famous MTCNN and ArcFace model. The system consists of two parts, i.e., a database management interface and a real-time face recognition program.

Dependencies

• Python 3.6+
• TensorFlow 1.12.0 (TF 1.x or TF 2.x should also works)
• PyQt5==5.9.2
• scikit-image==0.16.2
• opencv-python

Get started

Clone the project

git clone https://github.com/huangh12/py_face_recognition.git

Add your face to database

Simply run

python create_database.py

It will open a graphical interface. Try to add your target face into database as follows

Open the camera

Enable face detection

Capture the frame when you feel the face quality is fine. Also, you can release the frame and re-capture until the desired frame is obtained.

Enter the information according to the detection results and import to database finally.

Run the face recognition program

Simply run

python run.py

It will open your webcam and execute real-time face recognition. Let's take the other Stephen Curry image for testing.

BINGO!

End

Have fun with the project, and please star it if you like it.
Thank you.