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.

Read More →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.

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.

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

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

ZIP archive containing model weights, Python scripts, and README

Python 3.7+ · PyTorch · ONNX · MXNet frameworks supported

Windows, macOS, Linux — GPU & CPU inference modes

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

MIT licensed — free for personal and commercial use

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

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

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

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

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

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

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

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

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

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."

"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."

"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."

"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."

"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."

"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."

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

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

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