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

February 16, 2026 · View on GitHub

Tip

The models and functionality in this repository are integrated into UniFace — an all-in-one face analysis library.

Key Features

Real-Time Face Recognition: Process webcam or video files with SCRFD detection and ArcFace embeddings
FAISS Similarity Search: Batch cosine-similarity lookup using a FAISS inner-product index
Multiple Model Sizes: Choose from lightweight to high-accuracy detection and recognition models
Minimal Dependencies: Built on ONNX Runtime, OpenCV, NumPy, and FAISS with no extra frameworks

Note

Place your target face images in the assets/faces/ directory. The filenames will be used as identity labels during recognition.

Components

SCRFD — Sample and Computation Redistribution for Efficient Face Detection
ArcFace — Additive Angular Margin Loss for Deep Face Recognition
FAISS — Facebook AI Similarity Search

Available Models

Category	Model	Size	Description
Detection	SCRFD 500M	2.41 MB	Lightweight face detection
Detection	SCRFD 2.5G	3.14 MB	Balanced performance
Detection	SCRFD 10G	16.1 MB	High accuracy
Recognition	ArcFace MobileFace	12.99 MB	Mobile-friendly recognition
Recognition	ArcFace ResNet-50	166 MB	High-accuracy recognition

Project Structure

├── assets/
│   ├── demo.mp4
│   ├── in_video.mp4
│   └── faces/              # Place target face images here
│       ├── face1.jpg
│       ├── face2.jpg
│       └── ...
├── database/               # FAISS database implementation
├── models/                 # Neural network models
├── weights/                # Model weights (download required)
├── utils/                  # Helper functions
├── main.py                 # Main application entry
└── requirements.txt        # Dependencies

Getting Started

Prerequisites

Important

Make sure you have Python 3.10+ installed on your system.

Installation

Clone the repository:

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

Install dependencies:

pip install -r requirements.txt

Download model weights:

Click to see download links 📥

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

Quick download (Linux/Mac):

sh download.sh

Add target faces: Place face images in assets/faces/ directory. The filename will be used as the person's identity.

Usage

Basic Usage

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

Command Line Arguments

Tip

Use these arguments to customize the recognition behavior:

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]

Argument	Description	Default
`--det-weight`	Detection model path	`./weights/det_10g.onnx`
`--rec-weight`	Recognition model path	`./weights/w600k_mbf.onnx`
`--similarity-thresh`	Face similarity threshold	`0.4`
`--confidence-thresh`	Detection confidence threshold	`0.5`
`--faces-dir`	Target faces directory	`./assets/faces`
`--source`	Video source (file or camera index)	`./assets/in_video.mp4`
`--max-num`	Max faces per frame (0 = unlimited)	`0`
`--db-path`	Custom database storage location	`./database/face_database`
`--update-db`	Force rebuild face database	`False`
`--output`	Specify output video path	`output_video.mp4`

Technical Notes

Face database is saved to and loaded from disk automatically; no rebuild needed on restart
All detected faces in a frame are queried in a single FAISS index.search() call
For GPU-accelerated inference, install onnxruntime-gpu instead of onnxruntime

References

Note

This project builds upon the following research: