Liveness Using Mlkit Android
Published:
Integrate with CameraX by adding ImageAnalysis.Analyzer when build ImageAnalysis
val cameraExecutors = Executors.newSingleThreadExecutor()
ImageAnalysis.Builder()
.setTargetRotation(Surface.ROTATION_0)
.build()
.also { imageAnalysis ->
imageAnalysis.setAnalyzer(
cameraExecutors,
getCameraXAnalyzeManager(
frameHandler,
coroutineScope
)
)
}
get the image from this overriden function
override fun analyze(image: ImageProxy) {
imageProxy.image?.let { it ->
val image: InputImage = InputImage.fromMediaImage(
it,
imageProxy.imageInfo.rotationDegrees
)
}
detector.process(image)
...
}
Read this
Then you learned what attributes that you can get
1. tracking Id
2. headEulerAngleX
3. headEulerAngleY
4. headEulerAngleZ
5. boundingBox
6. smileProbability
7. rightEyeOpenProbability
8. leftEyeOpenProbability
it is from
package com.google.mlkit.vision.face;
Read the details here
It produces the value right away after capturing.
Read this
val realTimeOpts = FaceDetectorOptions.Builder()
.setContourMode(FaceDetectorOptions.CONTOUR_MODE_ALL)
.build()
val detector = FaceDetection.getClient(realTimeOpts)
val result = detector.process(image)
.addOnSuccessListener { faces ->
// Task completed successfully
// ...
}
.addOnFailureListener { e ->
// Task failed with an exception
// ...
}
faces is List makes sense, to handle multiple faces during capturing
so you can handle from there.