Face landmark detection guide for Android

The MediaPipe Face Landmarker task lets you detect face landmarks and facial expressions in images and videos. You can use this task to identify human facial expressions, apply facial filters and effects, and create virtual avatars. This task uses machine learning (ML) models that can work with single images or a continuous stream of images. The task outputs 3-dimensional face landmarks, blendshape scores (coefficients representing facial expression) to infer detailed facial surfaces in real-time, and transformation matrices to perform the transformations required for effects rendering.

The code sample described in these instructions is available on GitHub. For more information about the capabilities, models, and configuration options of this task, see the Overview.

Code example

The MediaPipe Tasks example code is a simple implementation of a Face Landmarker app for Android. The example uses the camera on a physical Android device to detect faces in a continuous video stream. The app can also detect faces in images and videos from the device gallery.

You can use the app as a starting point for your own Android app, or refer to it when modifying an existing app. The Face Landmarker example code is hosted on GitHub.

Download the code

The following instructions show you how to create a local copy of the example code using the git command line tool.

To download the example code:

1. Clone the git repository using the following command:

   git clone https://github.com/google-ai-edge/mediapipe-samples
   

2. Optionally, configure your git instance to use sparse checkout, so you have only the files for the Face Landmarker example app:

   cd mediapipe
   git sparse-checkout init --cone
   git sparse-checkout set examples/face_landmarker/android
   

After creating a local version of the example code, you can import the project into Android Studio and run the app. For instructions, see the Setup Guide for Android.

Key components

The following files contain the crucial code for this face landmarking example application:

• FaceLandmarkerHelper.kt - Initializes the face landmarker and handles the model and delegate selection.
• CameraFragment.kt - Handles the device camera and processes the image and video input data.
• GalleryFragment.kt - Interacts with OverlayView to display the output image or video.
• OverlayView.kt - Implements the display with a face mesh for detected faces.

Setup

This section describes key steps for setting up your development environment and code projects specifically to use Face Landmarker. For general information on setting up your development environment for using MediaPipe tasks, including platform version requirements, see the Setup guide for Android.

Dependencies

The Face Landmarker task uses the com.google.mediapipe:tasks-vision library. Add this dependency to the build.gradle file of your Android app:

dependencies {
    implementation 'com.google.mediapipe:tasks-vision:latest.release'
}

Model

The MediaPipe Face Landmarker task requires a trained model bundle that is compatible with this task. For more information on available trained models for Face Landmarker, see the task overview Models section.

Select and download the model, and store it within your project directory:

<dev-project-root>/src/main/assets

Specify the path of the model within the ModelAssetPath parameter. In the example code, the model is defined in the FaceLandmarkerHelper.kt file:

baseOptionsBuilder.setModelAssetPath(MP_FACE_LANDMARKER_TASK)

Create the task

The MediaPipe Face Landmarker task uses the createFromOptions() function to set up the task. The createFromOptions() function accepts values for the configuration options. For more information on configuration options, see Configuration options.

The Face Landmarker supports the following input data types: still images, video files, and live video streams. You need to specify the running mode corresponding to your input data type when creating the task. Choose the tab for your input data type to see how to create the task and run inference.

val baseOptionsBuilder = BaseOptions.builder().setModelAssetPath(MP_FACE_LANDMARKER_TASK)
val baseOptions = baseOptionBuilder.build()

val optionsBuilder = 
    FaceLandmarker.FaceLandmarkerOptions.builder()
        .setBaseOptions(baseOptionsBuilder.build())
        .setMinFaceDetectionConfidence(minFaceDetectionConfidence)
        .setMinTrackingConfidence(minFaceTrackingConfidence)
        .setMinFacePresenceConfidence(minFacePresenceConfidence)
        .setNumFaces(maxNumFaces)
        .setRunningMode(RunningMode.IMAGE)

val options = optionsBuilder.build()
FaceLandmarker = FaceLandmarker.createFromOptions(context, options)
    

val baseOptionsBuilder = BaseOptions.builder().setModelAssetPath(MP_FACE_LANDMARKER_TASK)
val baseOptions = baseOptionBuilder.build()

val optionsBuilder = 
    FaceLandmarker.FaceLandmarkerOptions.builder()
        .setBaseOptions(baseOptionsBuilder.build())
        .setMinFaceDetectionConfidence(minFaceDetectionConfidence)
        .setMinTrackingConfidence(minFaceTrackingConfidence)
        .setMinFacePresenceConfidence(minFacePresenceConfidence)
        .setNumFaces(maxNumFaces)
        .setRunningMode(RunningMode.VIDEO)

val options = optionsBuilder.build()
FaceLandmarker = FaceLandmarker.createFromOptions(context, options)
    

val baseOptionsBuilder = BaseOptions.builder().setModelAssetPath(MP_FACE_LANDMARKER_TASK)
val baseOptions = baseOptionBuilder.build()

val optionsBuilder = 
    FaceLandmarker.FaceLandmarkerOptions.builder()
        .setBaseOptions(baseOptionsBuilder.build())
        .setMinFaceDetectionConfidence(minFaceDetectionConfidence)
        .setMinTrackingConfidence(minFaceTrackingConfidence)
        .setMinFacePresenceConfidence(minFacePresenceConfidence)
        .setNumFaces(maxNumFaces)
        .setResultListener(this::returnLivestreamResult)
        .setErrorListener(this::returnLivestreamError)
        .setRunningMode(RunningMode.LIVE_STREAM)

val options = optionsBuilder.build()
FaceLandmarker = FaceLandmarker.createFromOptions(context, options)
    

The Face Landmarker example code implementation allows the user to switch between processing modes. The approach makes the task creation code more complicated and may not be appropriate for your use case. You can see this code in the setupFaceLandmarker() function in the FaceLandmarkerHelper.kt file.

Configuration options

This task has the following configuration options for Android apps:

Prepare data

Face Landmarker works with images, video files, and live video streams. The task handles the data input preprocessing, including resizing, rotation and value normalization.

The following code demonstrates how to hand off data for processing. These samples include details on how to handle data from images, video files, and live video streams.

import com.google.mediapipe.framework.image.BitmapImageBuilder
import com.google.mediapipe.framework.image.MPImage

// Convert the input Bitmap object to an MPImage object to run inference
val mpImage = BitmapImageBuilder(image).build()
    

import com.google.mediapipe.framework.image.BitmapImageBuilder
import com.google.mediapipe.framework.image.MPImage

val argb8888Frame =
    if (frame.config == Bitmap.Config.ARGB_8888) frame
    else frame.copy(Bitmap.Config.ARGB_8888, false)

// Convert the input Bitmap object to an MPImage object to run inference
val mpImage = BitmapImageBuilder(argb8888Frame).build()
    

import com.google.mediapipe.framework.image.BitmapImageBuilder
import com.google.mediapipe.framework.image.MPImage

// Convert the input Bitmap object to an MPImage object to run inference
val mpImage = BitmapImageBuilder(rotatedBitmap).build()
    

In the Face Landmarker example code, the data preparation is handled in the FaceLandmarkerHelper.kt file.

Run the task

Depending on the type of data your are working with, use the FaceLandmarker.detect...() method that is specific to that data type. Use detect() for individual images, detectForVideo() for frames in video files, and detectAsync() for video streams. When you are performing detections on a video stream, make sure you run the detections on a separate thread to avoid blocking the user interface thread.

The following code samples show simple examples of how to run Face Landmarker in these different data modes:

val result = FaceLandmarker.detect(mpImage)
    

val timestampMs = i * inferenceIntervalMs

FaceLandmarker.detectForVideo(mpImage, timestampMs)
    .let { detectionResult ->
        resultList.add(detectionResult)
    }
    

val mpImage = BitmapImageBuilder(rotatedBitmap).build()
val frameTime = SystemClock.uptimeMillis()

FaceLandmarker.detectAsync(mpImage, frameTime)
    

Note the following:

• When running in the video mode or the live stream mode, you must provide the timestamp of the input frame to the Face Landmarker task.
• When running in the image or the video mode, the Face Landmarker task blocks the current thread until it finishes processing the input image or frame. To avoid blocking the user interface, execute the processing in a background thread.
• When running in the live stream mode, the Face Landmarker task returns immediately and doesn’t block the current thread. It will invoke the result listener with the detection result every time it finishes processing an input frame.

In the Face Landmarker example code, the detect, detectForVideo, and detectAsync functions are defined in the FaceLandmarkerHelper.kt file.

Handle and display results

The Face Landmarker returns a FaceLandmarkerResult object for each detection run. The result object contains a face mesh for each detected face, with coordinates for each face landmark. Optionally, the result object can also contain blendshapes, which denote facial expressions, and a facial transformation matrices to apply face effects on the detected landmarks.

The following shows an example of the output data from this task:

FaceLandmarkerResult:
  face_landmarks:
    NormalizedLandmark #0:
      x: 0.5971359014511108
      y: 0.485361784696579
      z: -0.038440968841314316
    NormalizedLandmark #1:
      x: 0.3302789330482483
      y: 0.29289937019348145
      z: -0.09489090740680695
    ... (478 landmarks for each face)
  face_blendshapes:
    browDownLeft: 0.8296722769737244
    browDownRight: 0.8096957206726074
    browInnerUp: 0.00035583582939580083
    browOuterUpLeft: 0.00035752105759456754
    ... (52 blendshapes for each face)
  facial_transformation_matrixes:
    [9.99158978e-01, -1.23036895e-02, 3.91213447e-02, -3.70770246e-01]
    [1.66496094e-02,  9.93480563e-01, -1.12779640e-01, 2.27719707e+01]
    ...

The following image shows a visualization of the task output:

The Face Landmarker example code demonstrates how to display the results returned from the task, see the OverlayView class for more details.