Android

Faceplugin Face Recognition Android SDK. Fully on-premise face matching, 1:N identify, and 2D liveness. AAR, setActivation, init, faceDetection, VideoWorker.

Fully on-premise Face Recognition SDK for Android. Detect faces, extract templates you store yourself, match 1:1 or live 1:N, and check passive 2D liveness on the device. Images never leave the phone.

Public class: com.faceplugin.facerecognitionsdk.FaceRecognitionSDK. Demo tiles: Enroll, Identify, Capture, Attribute, Settings, About.

Code

GitHub - Faceplugin-ltd/FaceRecognition-Android: Face Recognition, Face Liveness Detection, Face Anti-Spoofing, Face Detection, Face Landmarks, Face Compare, Face Matching, Face Pose, Face Expression, Face Attributes, Face Templates Extraction, Face LandmarksGitHub

Setup

1. Copy the SDK (libfacesdk folder) to the root folder in your project. Place facerecognitionsdk.aar from Google Drive inside libfacesdk/ (next to libfacesdk/build.gradle).

2. Add SDK to the project in settings.gradle

rootProject.name = "YourProjectName"
include ':app'
include ':libfacesdk'

1. Add dependency to your app/build.gradle

android {
    defaultConfig {
        minSdk 24
        ndk { abiFilters 'arm64-v8a', 'armeabi-v7a' }
    }
    packaging {
        jniLibs { useLegacyPackaging = true }
    }
}
dependencies {
    implementation project(path: ':libfacesdk')
}

1. Add camera / gallery permissions in AndroidManifest.xml. Call setActivation → init on a background thread. Keep the demo package name com.faceplugin.facerecognitionsdk. Request a new license key for your own app.

Serialize native calls on one thread. The engine is not concurrent. First init unpacks on-device models (a few seconds).

Status codes: 0 Success (SDK_SUCCESS), 1 Invalid license, 2 Expired, 3 Not activated, 4 Init failed.

APIs

setActivation: This API is used to activate the SDK

public static int setActivation(Context context, String license);

Input	context (Context): Android Context license (String): The license string (license key)
Return value	The SDK activation status code. 0: Success 1: Invalid license 2: Expired 3: Not activated 4: Init failed

init: This API is used to initialize the SDK

public static int init(Context context) throws IOException;

Input	context (Context): Android Context
Return value	The SDK initialization status code. 0: Success 1–4: see setActivation

getMachineCode: This API is used to retrieve the machine code

public static String getMachineCode(Context context);

Input	context (Context): Android Context
Return value	Machine code string (for license requests).

getLicenseStatus: This API is used to read license status

public static String getLicenseStatus();
public static boolean allowsRecognition();
public static boolean allowsLiveness();
public static String lastLicenseError();
public static boolean isActivated();

Input	None
Return value	JSON status string.

faceDetection: This API is used to detect faces

public static List<FaceBox> faceDetection(Bitmap bitmap, FaceDetectionParam param);

Input	bitmap (Bitmap): The Bitmap image param (FaceDetectionParam): Parameters for face detection
Return value	A list of FaceBox objects representing the detected faces.

FaceDetectionParam

public class FaceDetectionParam {
    public boolean check_liveness = false;
    public int check_liveness_level = 0; // 0: High Accuracy, 1: Light
    public boolean check_eye_closeness = false;
    public boolean check_face_occlusion = false;
    public boolean estimate_age_gender = false;
    public boolean check_pose = true;
    public boolean check_landmarks = true;
    public boolean check_quality = false;
    public boolean check_emotion = false;
    public boolean check_mask = false;
    public boolean check_glasses = false;
}

Use FaceDetectionParam.allAttributes() when you want liveness, age, gender, emotion, and quality in one call.

FaceBox

public class FaceBox {
    public int x1, y1, x2, y2;
    public float yaw, roll, pitch, liveness;
    public float face_quality, face_luminance;
    public float left_eye_closed, right_eye_closed, face_occlusion, mouth_opened;
    public int age, gender, landmarkCount;
    public float[] landmarks_68;
    public String livenessLabel, genderLabel, emotionLabel, maskLabel;
    public String qualityLabel, eyesLeftLabel, eyesRightLabel, rawJson;
}

The liveness score ranges from 0.0 to 1.0. If it is greater than the threshold, it is a real face. Demo default threshold is 0.5.

templateExtraction: This API is used to extract face template

public static byte[] templateExtraction(Bitmap bitmap, FaceBox faceBox);

Input	bitmap (Bitmap): The Bitmap image faceBox (FaceBox): The bounding box of the detected face
Return value	A byte array representing the extracted template from the face. Store these in your database.

similarityCalculation: This API is used to calculate the similarity between two face templates

public static float similarityCalculation(byte[] feature1, byte[] feature2);

Input	feature1 (byte[]): The first face template feature2 (byte[]): The second face template
Return value	A float similarity score. Demo Identify default threshold is 0.67.

similarity(byte[] feature1, byte[] feature2) is an alias of similarityCalculation.

detect: This API is used to detect faces and return engine JSON

public static String detect(Bitmap bitmap, boolean crop);
public static String detect(Bitmap bitmap, boolean crop, int flags);

Input	bitmap (Bitmap): The Bitmap image crop (boolean): Whether to crop faces flags (int): Bit flags (DETECT_ALL, DETECT_LANDMARKS, DETECT_LIVENESS, DETECT_LIVENESS_ACCURATE, …)
Return value	JSON string (same idea as HTTP POST /api/detect).

quality: This API is used to run face image quality checks

public static String quality(Bitmap bitmap, boolean crop);

match: This API is used to compare two photos (1:1)

public static String match(Bitmap image1, Bitmap image2, boolean crop);

extractFeature: This API is used to extract a template from a cropped face image

public static String extractFeature(Bitmap bitmap);

Also: cropFace(Bitmap, FaceBox), parseFeatureBytes(String json), yuv2Bitmap(...), estimatorStatusJSON().

setLandmarkMode: This API is used to set the landmark model

public static int setLandmarkMode(int mode); // 14, 68, or 468
public static int getLandmarkMode();

VideoWorker: This API is used for live 1:N identify

Call in this order: startVideoWorker → syncVideoWorkerDatabase(templates) → addVideoWorkerFrame → stopVideoWorker. Listen with setVideoWorkerEventHandler. Identify in the demo is passive 2D liveness plus VideoWorker.

FaceRecognitionSDK.startVideoWorker(config);
FaceRecognitionSDK.syncVideoWorkerDatabase(templates, 0.67f);
FaceRecognitionSDK.addVideoWorkerFrame(bitmap);
FaceRecognitionSDK.stopVideoWorker();

Also: startVideoWorker(float matchThreshold), addVideoWorkerFrameRgb, addVideoWorkerFrameBgra, VideoWorkerConfig / ActiveLivenessConfig.

Input	Match threshold, enrolled templates, and camera frames.
Return value	Match events via the VideoWorker listener (JSON).

deinit: This API is used to unload the engine

public static int deinit();

Input	None
Return value	Status code. 0: Success.

Run the demo

1. Place facerecognitionsdk.aar in libfacesdk/ (see Setup).

2. Open the cloned repo in Android Studio.

3. Run on a physical phone (emulator is not recommended for camera / liveness).

4. Wait until the home status bar disappears. Then Enroll / Identify / Capture / Attribute unlock.

Keep the demo package name com.faceplugin.facerecognitionsdk so the sample key works.

Demo tile	What it does
Enroll	Enroll a person from a gallery photo (exactly one face) into the on-device database
Identify	Live 1:N camera match (stop on first hit) with 2D liveness
Capture	Oval coach capture → still with attributes → optional enroll
Attribute	Gallery analysis: landmarks, liveness, pose, quality, age, gender, emotion
Settings	Camera lens, identify / liveness / pose / eye-close thresholds
About	SDK name and license label

Screenshots

Home	Identify	Capture

System requirements

Item	Minimum	Recommended
Android	API 24 (7.0)	API 29 (10) or newer
ABI	arm64-v8a, armeabi-v7a	arm64-v8a
RAM	4 GB	6 GB or more
Camera	Front camera	720p or 1080p
Device	Physical device	Same

License

Licenses are offline. Keep the demo package name so the sample key works. For your own app, request a license.

Try it (after the demo compiles)

Thread {
    FaceRecognitionSDK.getMachineCode(context)
    var code = FaceRecognitionSDK.setActivation(context, DEMO_LICENSE)
    if (code == FaceRecognitionSDK.SDK_SUCCESS) {
        code = FaceRecognitionSDK.init(context)
    }
    val faces = FaceRecognitionSDK.faceDetection(bitmap, FaceDetectionParam())
    if (faces.isNotEmpty()) {
        val t = FaceRecognitionSDK.templateExtraction(bitmap, faces[0])
        val score = FaceRecognitionSDK.similarityCalculation(t, t)
    }
}.start()

Call setActivation, init, and all process methods off the UI thread. The engine is not concurrent — serialize calls on one thread.

Integrate into your own app

You need libfacesdk/ (the AAR) and FaceRecognitionSDK. You do not need the demo Activities.

1. Copy libfacesdk into your project root and put facerecognitionsdk.aar inside it.

2. Wire Gradle as in Setup (minSdk 24, abiFilters, useLegacyPackaging).

3. Add CAMERA and photo-library permissions.

4. Request license key for your own app.

5. Optional: copy app/.../kit/ (FaceRecognitionClient) so you do not rewrite threading, CameraX, or VideoWorker.

Typical call order: setActivation → init → faceDetection → templateExtraction → store templates in your database → similarityCalculation or VideoWorker for live 1:N.

Request a License & Support · Contact us

FAQ

What APIs does this page ship? Java/Kotlin FaceRecognitionSDK.setActivation, init, faceDetection, templateExtraction, similarityCalculation, and VideoWorker Identify.

Offline / on-premise? Yes, after a license key bound to the demo app id (or your own id).

Identify threshold? Default 0.67. Mobile Identify includes passive 2D liveness.

Related documentation

• Faceplugin Face Recognition Mobile SDK · iOS · Glossary

• Face Liveness Detection Android SDK

• ID Document Recognition Android SDK

• Request a License · Status codes

• Combining products (eKYC) · SDK comparison