Last updated 07-24-2026
Category:
Reviews:
Join thousands of AI enthusiasts in the World of AI!
InsightFace
InsightFace is an open-source face analysis platform that ships research-grade recognition, detection, and swapping models alongside enterprise licensing, SDKs, and APIs. Founded in 2018, it grew from the ArcFace paper into a full stack used by developers on GitHub and by companies that need commercial deployment rights.
The platform covers face recognition model packs like buffalo_l and antelopev2, InSwapper-128 face swapping, deepfake detection, the InspireFace C/C++ SDK with liveness checks, and Live Identity for authorized real-time streaming workflows. You can start with the Python package on PyPI, integrate via REST API, or license models for on-premise and edge deployment.
InsightFace targets ML engineers building prototypes, enterprises running KYC and access control, and creative teams that need reviewed face swap or live identity tooling. Its models rank among NIST FRVT top performers, and the project reports 29,000+ GitHub stars with deployments across 50+ countries.
Commercial licensing for face recognition packs including buffalo_l, antelopev2, buffalo_s, and buffalo_m
InSwapper-128 face swapping with API access and higher-fidelity proprietary models for approved use cases
Deepfake detection via browser demo, cloud API, on-premise SDK, and model licensing
InspireFace cross-platform C/C++ SDK with built-in anti-spoofing and liveness detection under 5MB
Live Identity runs authorized real-time identity transformation locally on NVIDIA GPUs for streaming workflows
Python package installs with pip and supports face recognition, detection, and swapping in a few lines of code
Open-source Python library with widely cited research models and 29,000+ GitHub stars.
Broad enterprise stack spanning recognition, swapping, deepfake detection, SDK, and API deployment.
Flexible deployment options including cloud API, on-premise, mobile edge, and local GPU streaming.
Commercial use of pre-trained models requires a separate license beyond the free MIT codebase.
Enterprise and API pricing is custom and requires contacting the sales team.
Live Identity and several advanced products are limited to reviewed organization-level licensing.
Is InsightFace free to use?
The InsightFace codebase is released under the MIT License and is free for commercial use. Pre-trained models are provided for non-commercial academic research; commercial use of those models requires a separate license from InsightFace.
How do I install InsightFace?
Install the Python package with pip install insightface. The site also documents ONNX runtime dependencies and sample code for face recognition and face swapping workflows.
What enterprise services does InsightFace offer?
InsightFace offers face recognition model licensing, face swap API and model licensing, deepfake detection SDK and API access, InspireFace SDK commercial licenses, Live Identity for streaming teams, and private model evaluation for enterprise benchmarks.
Does InsightFace provide deepfake detection?
Yes. InsightFace offers an online browser demo for single-image evaluation plus commercial cloud API, on-premise SDK, Docker server SDK, and model licensing options for production deepfake and face-forgery detection.
What is InsightFace Live Identity?
Live Identity is a Windows desktop application for organization-level live streaming workflows. It processes camera or video input locally on a supported NVIDIA GPU while using cloud services for activation and authorized identity embedding delivery.
Where can I find InsightFace source code and models?
The open-source project is hosted on GitHub at github.com/deepinsight/insightface, with model releases in the repository model zoo and a Python package published on PyPI.