face-recognition

Real-Time Face Recognition

Face Recognition

Table of Contents

• Architecture
• How to use
  • Create Environment and Install Packages
  • Add new persons to datasets
• Technology
  • Face Detection
  • Face Recognition
  • Face Tracking
  • Matching Algorithm
• Reference

Architecture

Sequence Diagram

How to use

Create Environment and Install Packages

conda create -n face-dev python=3.9

conda activate face-dev

pip install torch==1.9.1+cpu torchvision==0.10.1+cpu torchaudio==0.9.1 -f https://download.pytorch.org/whl/torch_stable.html
pip install -r requirements.txt

Add new persons to datasets

1. Create a folder with the folder name being the name of the person

   datasets/
   ├── backup
   ├── data
   ├── face_features
   └── new_persons
       ├── name-person1
       └── name-person2
   

2. Add the person’s photo in the folder

   datasets/
   ├── backup
   ├── data
   ├── face_features
   └── new_persons
       ├── name-person1
       │   └── image1.jpg
       │   └── image2.jpg
       └── name-person2
           └── image1.jpg
           └── image2.jpg
   

3. Run to add new persons

   python add_persons.py
   

4. Run to recognize

   python recognize.py
   

Technology

Face Detection

1. Retinaface

   • Retinaface is a powerful face detection algorithm known for its accuracy and speed. It utilizes a single deep convolutional network to detect faces in an image with high precision.

2. Yolov5-face

   • Yolov5-face is based on the YOLO (You Only Look Once) architecture, specializing in face detection. It provides real-time face detection with a focus on efficiency and accuracy.

3. SCRFD

   • SCRFD (Single-Shot Scale-Aware Face Detector) is designed for real-time face detection across various scales. It is particularly effective in detecting faces at different resolutions within the same image.

Face Recognition

1. ArcFace

   • ArcFace is a state-of-the-art face recognition algorithm that focuses on learning highly discriminative features for face verification and identification. It is known for its robustness to variations in lighting, pose, and facial expressions.

   
   ArcFace

Face Tracking

1. ByteTrack

   
   ByteTrack is a simple, fast and strong multi-object tracker.

Matching Algorithm

1. Cosine Similarity Algorithm

   • The Cosine Similarity Algorithm is employed for matching faces based on the cosine of the angle between their feature vectors. It measures the similarity between two faces’ feature representations, providing an effective approach for face recognition.

   
   Cosine Similarity Algorithm

Reference

• ByteTrack
• Yolov5-face
• InsightFace - ArcFace
• InsightFace-REST