This repository contains the implementation of our research paper, check the paper on Arxiv

IdentiFace

A Multimodal Facial Biometric System for Recognition, Gender Classification, Emotion Recognition and Face-Shape Prediciton

Project structure

├── main.py [Main file: Contains the welcome window]
├── Backend
|    ├── functions.py [contains all the used functions]
|    ├── model_manager.py [manages the models across windows]
|    ├── offline.py [offline window layout]
|    ├── online.py [online window layout]
├── utilities [Face-Detection: the used Dlib files for facial detection]
├── assets [Directory for project assets]
├── Models [a drive link for all the used models]
├── snapshots [contains all the notebooks and the codes for the different modalities]
├── test_examples [Test images]
├── snapshots [Snaps taken from the app]
└── requirements.txt [List of all required Python modules]

Getting started :

Clone the repository
Install the required dependencies by running pip install -r requirements.txt

dlib==19.24.2
keras==3.0.2
matplotlib==3.8.2
numpy==1.26.2
PySide6==6.6.1
tensorflow==2.15.0.post1

run main.py to start the application

Models

check the Paper for more detailed information about the data used / preprocessing / methodology or any other aspect of the project

The final used models in the GUI were as follows:

I. Face Recogniton Model trained on a subset of the color FERET database

II. Gender Classification Model trained on a Public Gender dataset

III. Face-Shape Prediciton Model trained on the Celebrity face-shape dataset

IV. Emotion Recognition Model trained on the FER2013 dataset

`Model`	`Train Accuracy`	`Test Accuracy`
Face Recognition	99.7%	99.2%
Gender Classification	96.48%	95.15%
Face-Shape Prediction	99.79%	88.03%
Emotion Recognition	81.26%	66.13%

GUI

We developed a Pyside desktop application called IdentiFace

The app mainly consists of:

I. A welcome window

II. An offline window

III. An online window

Note that because of the recognizer require high quality images , it was added only to the offline mode.

`window`	`screenshot`
welcome window
offline window
offline window
online window

Team Members

Note that this project was part of the Biometrics in the Senior SBME year at Cairo University under the supervision of DR. Ahmed.M.Badawi and the guidance of TA Laila Abbas

MahmoudRabea13/IdentiFace