How to Show Facial Landmark Points in Android

asked 2018-12-19 00:11:16 -0600

Hello everyone, i am android developer today working on the research of facial recognition. I have done some experiment to show the facial landmark points over the face using Dlib. Adding some calculation on the program. For more information, i build the program over python and able to located the facial landmark in real-time. The result shown below. image description i'have been looking the answer by Shujaat Ali, he is able to show landmark points but do not mention the steps how to do.

My question is, how to implement OpenCV in android to show facial landmark points in real-time video? It is able or not? i have been looking the answer showed by dtdzung , but that was for images , not in real-time videos. That was very complex because using Dlib and i think can not be able to show the landmark in real-time. I hope to be able to get detailed steps for implementation on Android. Thanks

edit retag flag offensive close merge delete


what have you tried so far ? (we can't see it)

it is for sure possible to do so.

berak gravatar imageberak ( 2018-12-19 01:23:34 -0600 )edit

Most of the process has been explain by Thimira Amaratunga . I follow the steps and successful perform with python using webcam and laptop. I want to implements those process using OpenCV. I read some article that openCV is able to perform facial landmark over android. But i do not know how to do and there is not relevant article, steps or tutorial inside this to perform the process in android native development.

ChoirulHuda gravatar imageChoirulHuda ( 2018-12-20 09:10:42 -0600 )edit

there is a java sample here (using opencv's own Landmarks pipeline, not dlib, and the FacemarkLBF version is quite good !)

berak gravatar imageberak ( 2018-12-20 09:14:44 -0600 )edit

Based on your link, is it uses for single image? i appreciate, thanks. But what i mean is for real-time video inside android. For example, if we run the apps and point out to somebody face, automatically detect facial landmark. Thanks

ChoirulHuda gravatar imageChoirulHuda ( 2018-12-20 09:45:23 -0600 )edit

the example is using a single image only, but it's fast enough to be used with a webcam.

if at all, the face detection is usually the more expensive operation (and dlib's is terrible)

(the problem now seems more, that you expect to find something readymade for your situation, which probably does not exist, you still have to understand the algorithm, and write your own app)

berak gravatar imageberak ( 2018-12-20 10:31:43 -0600 )edit