Open source 2D and 3D face analysis toolbox with local evaluation, recognition, detection, and face swap trials
- Stars29k
- Forks6k
- Open Issues1.3k
Other
- Python
- C++
- C

About InsightFace
InsightFace is an open source 2D and 3D face analysis toolbox for face recognition, face detection, and face alignment. It is built for training and deployment on image inputs, and it includes local evaluation and reporting for desktop use.
It supports ArcFace, SubCenter ArcFace, PartialFC, VPL, and implementations across PyTorch, MXNet, OneFlow, and PaddlePaddle. The package also includes face recognition, face detection, face alignment, and face swap trials, plus the InsightFace Evaluation Studio cross-platform GUI demo for Windows, macOS, and Linux.
InsightFace is maintained by Jia Guo and Jiankang Deng and is released under the MIT License for code. The project also includes licensing notes for training data and models, and it links to a cross-platform C/C++ face recognition SDK called InspireFace.
Key features
- ArcFace, SubCenter ArcFace, PartialFC, VPL implementations
- Face recognition, face detection, and face alignment
- InsightFace Evaluation Studio for local evaluation and reporting
- Basic face swap trials in the desktop GUI demo
- Works with PyTorch, MXNet, OneFlow, and PaddlePaddle
Details
- First released
- 2017
- Platforms
- Windows · macOS · Linux
- Languages
- Python · C/C++
- Frameworks
- PyTorch · MXNet · OneFlow · PaddlePaddle
- License
- MIT
- Deployment
- Offline-first
