⬇ Download
🤖 AI-Powered Face Detection

RetinaFace — Precise AI Face Detection & Analysis

RetinaFace is a single-stage deep learning face detector delivering sub-millisecond facial landmark localization, multi-face detection, and 3D mesh reconstruction for production computer vision pipelines.

⬇ Download RetinaFace — Free 📖 Read the Guide
✓ Free & Open Source
✓ MIT Licensed
✓ Python Ready
✓ GPU & CPU
📦 RetinaFace Package ✓ Stable
Version
1.0.0
File Size
48 MB
License
MIT
Python
3.7+
⬇ Download Now — Free
1 pip install retinaface
2 Load pre-trained weights
3 Detect faces in any image
🆓
Free Download
⚡
Real-Time Inference
🔒
Safe & Verified
💻
Cross-Platform

RetinaFace at a Glance

91.4%
AP on WIDER FACE Hard Subset
100+
FPS on GPU with MobileNet Backbone
5
Facial Landmarks Detected per Face
1
Single Forward Pass — No Multi-Stage Pipeline
🔍 About RetinaFace

The Gold Standard in Single-Stage Face Detection

RetinaFace is a state-of-the-art face detection model developed by researchers at Imperial College London and the Insightface project. Unlike older multi-stage pipelines, RetinaFace processes an entire image in a single forward pass through a deep Feature Pyramid Network, simultaneously localizing faces and their five key landmarks.

Whether you are building a face recognition pipeline, a crowd analytics system, an AR filter, or a security access control application, RetinaFace provides the reliable detection foundation your project needs.

  • Designed for real-world unconstrained environments
  • Handles small faces, occlusion, and extreme poses
  • Flexible backbone options from MobileNet to ResNet-50
  • ONNX export for cross-platform deployment
  • Active open-source community and regular updates
Learn More About RetinaFace
⚙ Features

Why Developers Choose RetinaFace

A comprehensive set of capabilities that make RetinaFace the preferred face detection solution for production-grade computer vision applications.

🎯

Single-Stage Architecture

Detect faces in one forward pass with no multi-stage cascade, delivering fast inference without sacrificing accuracy.

📐

Five-Point Landmark Detection

Outputs precise coordinates for both eye centers, nose tip, and both mouth corners for accurate face alignment.

🔭

Multi-Scale Detection

Feature Pyramid Network enables detection of faces from 16×16 pixels to full-frame close-ups in a single inference.

🧠

Multi-Task Learning

Simultaneously trained on face classification, bounding box regression, and landmark localization for richer representations.

🌐

3D Mesh Reconstruction

Optional dense 3D face mesh branch provides head pose, depth, and structural face data for AR and animation.

📱

Lightweight Backbone Options

Choose MobileNet for CPU and edge deployment or ResNet-50 for maximum accuracy on GPU servers.

⚡

Real-Time Performance

Achieves 100+ FPS on modern GPUs with the MobileNet backbone — suitable for live video and streaming.

🔄

ONNX & Framework Portable

Export to ONNX and run on TensorRT, OpenVINO, or ONNX Runtime for maximum deployment flexibility.

📊

Benchmark-Leading Accuracy

Achieves 91.4% AP on the WIDER FACE hard subset, outperforming MTCNN, PyramidBox, and many two-stage detectors.

📋 How To Use

Get Started with RetinaFace in Minutes

Follow these steps to integrate RetinaFace into your Python project and start detecting faces immediately.

1
⬇

Download the Package

Click the download button to get the RetinaFace model package including pre-trained weights and Python scripts.

2
🐍

Install Python Dependencies

Run pip install retinaface in your terminal. Dependencies including NumPy and OpenCV install automatically.

3
📂

Extract the Files

Unzip the downloaded package into your project directory. The folder structure is pre-organized for immediate use.

4
⚙

Load the Model

Import RetinaFace in your Python script and call the detect_faces method with your image path or NumPy array.

5
🖼

Run Detection

Pass any image to RetinaFace and receive bounding boxes, confidence scores, and five-point landmark coordinates for every detected face.

6
🔗

Integrate Your Pipeline

Feed the detection output into your downstream model — face recognition, alignment, analysis, or any computer vision task.

⬇ Download

Download RetinaFace

Get the complete RetinaFace model package with pre-trained weights, Python integration examples, and comprehensive documentation.

📦

Package Format

ZIP archive containing model weights, Python scripts, and README

🐍

Python Compatible

Python 3.7+ · PyTorch · ONNX · MXNet frameworks supported

💻

Platform Support

Windows, macOS, Linux — GPU & CPU inference modes

🔒

Safe & Clean

Virus-free, no adware, no bundled software — clean download

📜

Open License

MIT licensed — free for personal and commercial use

📚

Documentation Included

Step-by-step installation guide, API reference, and usage examples

🤖

RetinaFace Model Package

Pre-trained weights, Python implementation, and usage examples in one clean download.

Version
1.0.0
File Size
48 MB
⬇ Download Now — Free
✓ Free Forever ✓ No Sign-up ✓ MIT License
💻 Compatibility

RetinaFace Platform Support

RetinaFace runs across all major operating systems and supports multiple Python-based deep learning frameworks.

Platform / Browser CPU Inference GPU Inference ONNX Runtime Notes
🪟 Windows 10/11✓ Supported✓ CUDA✓Recommended for desktop development
🍎 macOS 12+✓ Supported✓ MPS✓Apple Silicon (M1/M2) supported via MPS
🐧 Linux (Ubuntu 20+)✓ Supported✓ CUDA✓Best for server & cloud deployment
🤖 Android (via ONNX)✓ Partial— N/A✓ONNX Runtime Android required
📱 iOS (via CoreML)✓ Partial— N/A✓CoreML export via ONNX converter
🟢 PyTorch Backend✓ Full✓ Full✓Primary recommended backend
🔵 MXNet Backend✓ Full✓ Full—Original Insightface implementation
🟠 ONNX Runtime✓ Full✓ TRT✓Best for cross-platform production
💡 Benefits

Who Benefits from RetinaFace?

RetinaFace serves a wide community of professionals from machine learning engineers to researchers and app developers.

👨‍💻

ML Engineers

Integrate a production-ready face detection layer into your recognition or analysis pipeline without reinventing the wheel.

🎓

Researchers

Use RetinaFace as a reliable baseline or benchmark starting point for computer vision and face analysis research.

📸

App Developers

Add automatic face detection and tagging to photo apps, social platforms, and media management tools.

🔐

Security Teams

Deploy RetinaFace in access control, surveillance, and identity verification systems requiring reliable multi-face detection.

🎮

AR/VR Developers

Use the 3D mesh output to drive avatar animation, virtual try-on, and AR filter applications accurately.

🏥

Healthcare AI

Apply RetinaFace in dermatology screening, ophthalmology tools, and patient monitoring systems needing precise facial localization.

⭐ Reviews

What Developers Say About RetinaFace

Trusted by computer vision engineers and researchers building real-world face analysis systems.

★★★★★

"RetinaFace replaced MTCNN in our entire production pipeline and improved our end-to-end recognition accuracy by over 4%. The landmark predictions are noticeably more stable on challenging angles."

AM
Arjun Mehta
ML Engineer, Computer Vision Startup
★★★★★

"We deployed RetinaFace with the MobileNet backbone on Jetson Nano devices across 12 retail locations. It handles real crowds reliably and has been running without issues for six months."

LF
Lena Fischer
Embedded Systems Developer
★★★★★

"The ONNX export workflow is smooth and well-documented. Moving from PyTorch training to TensorRT inference cut our detection latency from 18ms to 6ms per frame."

CR
Carlos Rivera
Senior CV Engineer, FinTech Company
★★★★

"Excellent detection on the WIDER FACE hard set cases. I tried several detectors for my research on crowd analysis and RetinaFace gave the best recall on densely packed scenes by a clear margin."

PN
Dr. Priya Nair
Computer Vision Researcher
★★★★★

"Integration took less than an hour. The Python API is clean, the docs are clear, and the performance exceeded what I expected from an open-source model. Highly recommended for any face pipeline."

TA
Tom Andersen
Full Stack Developer, Photo App
★★★★★

"The five-point landmarks are accurately calibrated even on faces with mild occlusion. Our face alignment quality improved significantly after switching from a two-stage detector to RetinaFace."

ML
Mei Liu
AI Product Engineer
📖 Guide

Learn More About RetinaFace

In-depth articles covering RetinaFace architecture, usage, benchmarks, and integration for every skill level.

What is RetinaFace?
Overview
What is RetinaFace?

RetinaFace is a single-stage face detection model built on deep learning that simultaneously detects faces and localizes five key facial landmarks with exceptional accuracy.

📅 June 10, 2025
Read More →
How Does RetinaFace Work?
Technical
How Does RetinaFace Work?

A deep dive into the internal mechanics of RetinaFace, from feature pyramid extraction to multi-task loss functions and bounding box regression.

📅 June 12, 2025
Read More →
What Are the Main Features of RetinaFace?
Features
What Are the Main Features of RetinaFace?

Explore the standout features of RetinaFace including multi-task learning, five-point landmark detection, 3D mesh estimation, and its performance on WIDER FACE.

📅 June 14, 2025
Read More →
Is RetinaFace Suitable for Real-Time Applications?
Performance
Is RetinaFace Suitable for Real-Time Applications?

Analyze RetinaFace performance benchmarks, FPS rates, hardware requirements, and optimization strategies for deploying it in real-time video and streaming systems.

📅 June 16, 2025
Read More →
What Facial Landmarks Does RetinaFace Detect?
Technical
What Facial Landmarks Does RetinaFace Detect?

Understand the five facial landmarks that RetinaFace detects, how they are used for face alignment, and their role in downstream recognition and analysis tasks.

📅 June 18, 2025
Read More →
How Accurate is RetinaFace?
Performance
How Accurate is RetinaFace?

A detailed look at RetinaFace accuracy metrics on WIDER FACE, IJB-C, and other benchmarks, including precision, recall, and AP scores across difficulty levels.

📅 June 20, 2025
Read More →
View All Guides →
❓ FAQ

Frequently Asked Questions about RetinaFace

Answers to the most common questions about using, installing, and deploying RetinaFace.

🛠
Technical Questions
What Python version does RetinaFace require? ▾
RetinaFace supports Python 3.7 and above. It is compatible with Python 3.8, 3.9, 3.10, and 3.11. Earlier Python versions are not officially supported due to dependency requirements.
Can RetinaFace run on CPU without a GPU? ▾
Yes. The MobileNet backbone variant runs adequately on CPU for non-real-time applications. On a modern CPU, you can expect 10 to 20 FPS at standard resolution. For real-time applications a GPU is strongly recommended.
What output does RetinaFace provide? ▾
RetinaFace returns a dictionary for each detected face containing a confidence score, a bounding box in [x1, y1, x2, y2] format, and five facial landmark coordinates (left eye, right eye, nose tip, left mouth corner, right mouth corner).
Is RetinaFace compatible with ONNX? ▾
Yes. The PyTorch RetinaFace model can be exported to ONNX format and run with ONNX Runtime, TensorRT, or OpenVINO for flexible cross-platform production deployment.
How do I detect multiple faces in one image? ▾
RetinaFace automatically detects all faces in an image in a single forward pass. Call detect_faces() on your image and the returned dictionary will contain one entry per detected face, regardless of how many appear in the frame.
📋
General Questions
Is RetinaFace free to use commercially? ▾
Yes. RetinaFace is licensed under the MIT License, which allows free use for both personal and commercial projects. You are free to integrate it into products, services, and commercial applications.
How does RetinaFace compare to MTCNN? ▾
RetinaFace consistently outperforms MTCNN on benchmark datasets, particularly on the WIDER FACE hard subset. It achieves higher AP scores, better small face detection, and more accurate landmark predictions, while being faster on GPU hardware.
What is the minimum detectable face size? ▾
Under standard settings, RetinaFace reliably detects faces as small as 16×16 pixels. For smaller faces, processing the image at higher resolution or applying preprocessing can improve detection recall further.
Does RetinaFace support video processing? ▾
RetinaFace does not include a built-in video reader, but it integrates seamlessly with OpenCV or PyAV for frame-by-frame video processing. For real-time performance, pair it with a lightweight tracker between detection frames.
Where can I report bugs or get community support? ▾
RetinaFace is maintained under the Insightface project on GitHub. You can open issues, view existing bug reports, and follow discussions in the project repository. For deployment questions, contact our support team via the contact page.

Start Detecting Faces with RetinaFace Today

Join thousands of developers and researchers using RetinaFace for production computer vision applications. Free download, no sign-up required.

⬇ Download RetinaFace — Free 📖 Browse the Guide