Input:
Image Bitmap (any size)
Output:
Bitmap with biggest people face on transparent background or NULL if there is no people found on input image.
-Output image should be same size as original.
-Face should be located at same position as in original bitmap.
Android API usage example:
Segmentation faceSegmentation = new FaceSegmentation(context);
Bitmap face = faceSegmentation.findFace(imageBitmap);
Requirements:
-compileSdkVersion 28
-output face with smooth transparent edges
-final compiled size < 30MB
-created as library (usable as dependency )
-java (not kotlin)
-Output image should be same size as original
-local model on device
-Use TensorFlow Lite
Example project: