Real-Time Face Re-Identification with FAISS, ArcFace & SCRFD

demo.mp4

This repository implements face re-identification using SCRFD for face detection and ArcFace for face recognition. It supports inference from webcam or video sources.

Features

• FAISS Vector Database Integration: Enables fast and scalable face re-identification using a FAISS index built from facial embeddings. Faces must be placed in the assets/faces/ directory.
• Smaller versions of SCFRD face detection model has been added
• Face Detection: Utilizes Sample and Computation Redistribution for Efficient Face Detection (SCRFD) for efficient and accurate face detection.
  • Added models: SCRFD 500M (2.41 MB), SCRFD 2.5G (3.14 MB)
• Face Recognition: Employs ArcFace: Additive Angular Margin Loss for Deep Face Recognition for robust face recognition.
  • Added models: ArcFace MobileFace (12.99 MB)
• Real-Time Inference: Supports both webcam and video file input for real-time processing.

Project folder structure:

├── assets/
│   ├── demo.mp4
│   |── in_video.mp4
|   └── faces/
│     ├── face1.jpg
│     ├── face2.jpg
│     └── ...
├── database/
|   ├── __init__.py
|   └── face_db.py
├── models/
│   ├── __init__.py
│   ├── scrfd.py
│   └── arcface.py
├── weights/
│   ├── det_10g.onnx
│   ├── det_2.5g.onnx
│   ├── det_500m.onnx
│   ├── w600k_r50.onnx
│   └── w600k_mbf.onnx
├── utils/
|   ├── logging.py
│   └── helpers.py
├── main.py
├── README.md
└── requirements.txt

Installation

1. Clone the repository:

git clone https://github.com/yakyo/face-reidentification.git
cd face-reidentification

2. Install the required dependencies:

pip install -r requirements.txt

3. Download weight files:

   a) Download weights from following links:

   Model	Weights	Size	Type
   SCRFD 500M	det_500m.onnx	2.41 MB	Face Detection
   SCRFD 2.5G	det_2.5g.onnx	3.14 MB	Face Detection
   SCRFD 10G	det_10g.onnx	16.1 MB	Face Detection
   ArcFace MobileFace	w600k_mbf.onnx	12.99 MB	Face Recognition
   ArcFace ResNet-50	w600k_r50.onnx	166 MB	Face Recognition

   b) Run below command to download weights to weights directory (linux):

   sh download.sh

4. Put target faces into assets/faces folder

faces/
    ├── name1.jpg
    ├── name2.jpg

Those file names will be displayed while real-time inference.

Usage

python main.py --source assets/in_video.mp4

main.py arguments:

usage: main.py [-h] [--det-weight DET_WEIGHT] [--rec-weight REC_WEIGHT] [--similarity-thresh SIMILARITY_THRESH] [--confidence-thresh CONFIDENCE_THRESH]
               [--faces-dir FACES_DIR] [--source SOURCE] [--max-num MAX_NUM]

Face Detection-and-Recognition

options:
  -h, --help            show this help message and exit
  --det-weight DET_WEIGHT
                        Path to detection model
  --rec-weight REC_WEIGHT
                        Path to recognition model
  --similarity-thresh SIMILARITY_THRESH
                        Similarity threshold between faces
  --confidence-thresh CONFIDENCE_THRESH
                        Confidence threshold for face detection
  --faces-dir FACES_DIR
                        Path to faces stored dir
  --source SOURCE       Video file or video camera source. i.e 0 - webcam
  --max-num MAX_NUM     Maximum number of face detections from a frame

Reference

1. https://github.com/deepinsight/insightface/tree/master/detection/scrfd
2. https://github.com/deepinsight/insightface/tree/master/recognition/arcface_torch