Skip to content

aangfanboy/deepface

Repository files navigation

liyana: face analysis project

Twitter Twitch YouTube

For supporters, My Bitcoin/coin.space: 1LUFWnzrGVLdsZ7gnfee87iX6QqSn24Tvr

Thank you for donations, i am grateful.

🌟🌟🌟 You can also support me by starring this project 🌟🌟🌟

Face Recognition Models 🤓

You can use one of my models or train your own face recognition model with the help of Google Colab page linked below. For more information of the title face recognition, please visit this folder

Open In Colab

Model Architecture Epochs LFW Acc AgeDB Acc CFP Acc
A InceptionResNetV1 9 %99.53 %95.11 %93.97
B ResNet50V2 11 %99.51 %94.53 %93.60
C L_Resnet50_E_IR 7 %99.70 %96.75 %97.34

Features ✔️

  • Facial recognition with database
  • Facial recognition on webcam
  • Facial recognition on video
  • Displaying 2D space which database lies on
  • Age, sex and ethnicity detection
  • DeepFake detection

Features that expected to come in next versions 📝

  • Online database that every user can add face to common pool
  • Face re-generation with extracted features
  • Advanced video analysis

liyana 1.1.0

User gives the face at the right side of page to program, program extract features through machine learning model, features compares with those already saved in database and result prints on the screen. Face on the left side of page is Emma Watson's face in the database. Values such age, sex and ethnicity are loaded from database too, you can re-analyze those values by clicking re-analyze with models

Usage 📗

Python Side 🐍

  • Install libraries with pip install -r requirements.txt
  • Download a model from face recognition folder, extract it, copy arcface_final.h5 to main_app/python_server/arcface_final.h5
  • Download Age-Sex-Ethnicity classification models, details and last models can be found here
  • Download DeepFake detection models, details and last model can be found here
  • Run main_app/python_server/server.py
  • If you don't need a GUI, run main_app/python_server/client.py. Commands are listed in here.

I highly recommend to use a GPU, you can follow those steps if you have one.

Electron Side 🔌

  • Install electron with npm. Check this page for help.
  • You may need to install dependencies for electron, all can be found in here
  • Go /main_app/electron_scripts and run electron . while python server is running.

I will add a video to YouTube to explain how you can install and how it works, stay tuned.

I will make a version of app that works with TensorFlow JS so it can be work with just electron, stay tuned.