@nobegroom/react-native-tflite
v1.0.0
Published
<p align="center"> <img src="https://raw.githubusercontent.com/nobecomp/assets-/main/nobegloom_fill.svg" width="280" /> <br/> <img src="https://developer.android.com/static/codelabs/recognize-flowers-with-tensorflow-on-android/img/657431be3173fa8
Maintainers
Readme
NobeGroom | Headless Face Mesh
The Philosophy
NobeGroom is a headless, UI-free React Native library designed for high-performance facial analysis.
It prioritizes raw data over pre-built components, giving you the freedom to render facial meshes using Skia, SVG, or Canvas without being locked into a specific UI engine.
Features
Detection Scores
Direct mapping of detection_scores for precise filtering and confidence thresholds.
Zero-Bridge JSI
Built on react-native-fast-tflite for near-native inference speeds.
Headless Control
Pure logic for landmark extraction. No mandatory camera components or styles.
Installation
npm install @nobegroom/react-native-tflite react-native-fast-tflite react-native-image-resizerUsage
import { FaceDetector, FaceMeshMapper } from '@nobegroom/react-native-tflite';
const detector = new FaceDetector();
const mapper = new FaceMeshMapper();
const processFace = async (imageUri) => {
const box = await detector.detect(imageUri, 0.85);
if (box) {
const mesh = await mapper.extractFeatures(imageUri, box);
console.log("Score:", box.detection_scores);
console.log("Left Eye:", mesh.features.leftEye);
}
};Face Indices
| Key | Description | Points |
|------------|-------------------------------------|--------|
| faceOval | Outline of the face | 36 |
| lips | Upper, Lower, and Inner paths | 80 |
| irises | Left and Right iris boundaries | 10 |
Built by NobeGroom • 2026
A simple, high-performance facial toolkit for React Native.
