SelfieSegmentation
This project provides an example implementation of the ML Kit Selfie Segmentation API in an Android app.
Installation
Just clone the repo, open the project in Android Studio and run the project. For better performance run it on a physical device instead of AVD.
Features
The user can choose a front and a background image from his library. Then the image is processed and three modes can be selected.
Default mode
Displays the chosen front image.
Mask mode
Shows the front image and highlights the detected background of the segmentation mask as a green overlay.
Custom background mode
Shows the front image and uses the detected segmentation mask to overlay the background image.
Limitations
Since processing of a large image can take quite some time a selected image is resized to the current screen width. Furthermore, a background image is stretched to the same size as the current front image to simplify the overlay creation.