- 1,200+ Stars | A lightweight, production-grade Android SDK for offline face detection, 1:1 verification, 1:N / M:N vector searching, and liveness anti-spoofing.
- Who should use this: Developers building smart access control, Android-based IoT kiosks, passwordless login systems, or privacy-conscious mobile apps that cannot transmit biometric data off the device.
What is FaceAISDK_Android & Why It Matters
Transmitting biometric data to remote servers introduces latency, requires internet bandwidth, and creates major data privacy liabilities. FaceAISDK_Android addresses these challenges by executing full face processing on-device across Android versions 8 through 16+.
The SDK operates entirely offline. Biometric extraction, feature matching, and vector searches happen on local hardware without sending network packets or storing external logs. It supports standard Android system cameras as well as external UVC-protocol USB cameras—making it directly deployable on embedded hardware, smart door locks, turnstiles, and enterprise access terminals.
graph TD
A[Camera Feed: SysCamera / UVCCamera] --> B[FaceAISDK Runtime]
B --> C{Liveness Engine}
C -->|Pass| D[Feature Vector Extraction]
C -->|Spoof Detected| E[Reject Frame]
D --> F{Search Mode}
F -->|1:1 Mode| G[Static Biometric Verification]
F -->|1:N / M:N Mode| H[Local Database Vector Match]
G --> I[Match Output / Authorization]
H --> I[Match Output / Authorization]
Key Features & Capabilities
1. Offline Vector Search (1:1, 1:N, and M:N)
The core engine computes face feature vectors directly on-device. For single-user authentication (1:1), it compares a live capture against stored template data. For multi-user systems (1:N / M:N), it executes rapid vector matches against local databases, suitable for high-throughput enterprise attendance and residential access control.
2. Multi-Modal Liveness Anti-Spoofing
To block photo attacks, screen replays, and 3D mask spoofs, the SDK implements dual liveness strategies:
- Silent Liveness: Passive optical analysis executed automatically during frame capture without requiring user intervention.
- Action Liveness: Active challenge-response workflows tracking dynamic actions like blinking, smiling, mouth opening, head shaking, and head nodding.
3. Broad Hardware and Camera Support
Beyond standard mobile camera sensors, FaceAISDK_Android explicitly supports external USB cameras running the UVC protocol (optimized for Wide Dynamic Range cameras with WDR > 105dB). This compatibility eliminates driver workarounds when deploying on custom Android boards, tablets, or industrial kiosks.
SDK Architecture & Module Breakdown
| Module Name | Functional Role |
|---|---|
verify |
Handles 1:1 facial verification, static photo matching, and active/silent anti-spoofing checks. |
search |
Drives 1:N and M:N facial search algorithms alongside local CRUD operations for vector databases. |
addFace |
Shared ingestion component for frame sampling, facial extraction, and vector registration. |
SysCamera / UVCCamera |
Abstraction layers for native mobile camera pipelines and external USB UVC video capture streams. |
// Ensure mavenCentral() is included in your settings.gradle or root build.gradle:
repositories {
mavenCentral()
}
// Add the dependency to your app module's build.gradle:
dependencies {
// Replace 'Version' with the target SDK release (e.g., V2026.07.30)
api 'io.github.FaceAISDK:Android:Version'
// Note: If using external USB cameras, include the corresponding UVCAndroid dependency
}
📂 Explore the open-source repository on GitHub: https://github.com/FaceAISDK/FaceAISDK_Android
Leave a Reply