Faceplugin-ltd/FaceRecognition-Android

What it solves

This project provides a fully on-device biometric engine for Android, enabling developers to integrate face recognition and liveness detection into mobile apps without sending biometric data to the cloud. It is designed for privacy-first applications such as banking, KYC (Know Your Customer), and access control.

How it works

The SDK operates as an on-premise biometric engine that processes all data locally on the device. It uses a native library (facerecognitionsdk.aar) to perform face detection, template extraction for enrollment, and similarity calculations for 1:N identification. It includes a VideoWorker for live camera matching and a "demo kit" of Kotlin helpers to manage threading, CameraX integration, and local database storage.

Who it’s for

Android developers building identity verification, mobile onboarding, or security apps that require offline, on-device facial biometric processing.

Highlights

• On-Device Processing: No biometric data is sent to the cloud, ensuring high privacy.
• Comprehensive Biometrics: Supports 1:N identification, 2D liveness/anti-spoofing, and face capture with an oval coach.
• Attribute Analysis: Extracts 14-point landmarks, age, gender, emotion, and image quality (blur, noise, pose).
• Developer Tooling: Includes a demo kit with FaceRecognitionClient to simplify the implementation of threading and camera plumbing.