A React Native library for face detection, recognition, and analysis, built on top of the InspireFace SDK with Nitro Modules to bring native performance to your React Native face recognition features.
All features are powered by the underlying InspireFace C++ SDK:
- Face detection and tracking
- Face recognition and comparison
- Facial landmarks detection (106-point dense + 5-point key)
- Face quality assessment
- Mask detection
- Liveness detection (both silent and cooperative)
- Face attribute analysis (age, gender, race)
- Face emotion recognition
- Face pose estimation
- Face embedding management (FeatureHub with search, insert, update, remove)
- Zero-copy camera frame processing via
createEmptyImageStream()+setBuffer() - Configurable landmark engines (HypLMv2, InsightFace)
- Apple CoreML / CUDA hardware acceleration
- SDK log level control
For real-time camera feeds, avoid per-frame allocations by reusing a single stream:
// Create once
const stream = InspireFace.createEmptyImageStream();
stream.setFormat(ImageFormat.YUV_NV21);
stream.setRotation(CameraRotation.ROTATION_90);
// Each frame — just swaps the pointer, zero allocations
stream.setBuffer(frameBuffer, width, height);
const faces = session.executeFaceTrack(stream);
// Clean up when done
stream.dispose();-
For more technical details, visit the InspireFace repository.
See the contributing guide to learn how to contribute to the repository and the development workflow.
- InspireFace - The underlying C++ SDK
- Marc Rousavy - Nitro modules
MIT