0

Is it possible when I detect an object with the haar cascade classifier, to render a 3d object over the detected area?

Lets say I detect a face and I want to render super mario in the area of the face. I know I have to use OpenGL ES for rendering objects.

Should it be something with SurfaceTexture? An advice would be useful too! Thank you in advance!

Jeru Luke
  • 20,118
  • 13
  • 80
  • 87
dephinera
  • 3,703
  • 11
  • 41
  • 75

1 Answers1

1

Here is the code for the work I did: MotionDetectionAR If the face detection gives you the size of the face then you will be able to scale a 3D object in the exact location. However, I would be worried about the orientation aspects - if say the person is not looking directly at the camera then will you be able to detect that angle (both the left/right and up/down angles)? The 3D Object would need to be oriented in the same way and would not look right if its not taken into account.

You might like to look at this face superposition technique - I thought this was impressive; I don't think its Android but may give you some ideas.

timegalore
  • 731
  • 7
  • 9