shared/com.mediasfu.sdk.background/AndroidVirtualBackgroundProcessor

AndroidVirtualBackgroundProcessor

class AndroidVirtualBackgroundProcessor(context: Context) : VirtualBackgroundProcessor

Android implementation of VirtualBackgroundProcessor using ML Kit Selfie Segmentation.

This captures frames from WebRTC video, runs ML Kit segmentation to detect the person, composites the person with a virtual background, and outputs to a virtual video source.

Constructors

constructor(context: Context)

Types

Companion

object Companion

Properties

currentBackground

open override val currentBackground: VirtualBackground?

The currently applied virtual background

isProcessing

open override val isProcessing: Boolean

Whether the processor is currently active

Functions

release

open override fun release()

Release all resources. Call when done with the processor.

setFrameCallback

open override fun setFrameCallback(onProcessedFrame: (ProcessedFrame) -> Unit?)

Update the frame callback for preview while processing continues. This allows registering a preview callback when the processor is already running.

startProcessing

open suspend override fun startProcessing(inputStream: MediaStream, background: VirtualBackground, onProcessedFrame: (ProcessedFrame) -> Unit?): MediaStream?

Start processing the input video stream with the given background.

startProcessingWithDevice

open suspend override fun startProcessingWithDevice(inputStream: MediaStream, background: VirtualBackground, device: WebRtcDevice, onProcessedFrame: (ProcessedFrame) -> Unit?): MediaStream?

Start processing with a WebRtcDevice to create a virtual output stream. This creates a new video source that receives processed frames and can be used with mediasoup to send to remote participants.

stopProcessing

open suspend override fun stopProcessing(): MediaStream?

Stop processing and release resources.

updateBackground

open suspend override fun updateBackground(background: VirtualBackground)

Update the virtual background while processing continues.