Face detection and facial features localization are important parts of face analysis techniques.It's task is to search facial features in an image.Although human can finish this work without any effort

it's still a very difficult work for computer.In the past years

this technique has developed vastly and has been successfully applied in the fields of face recognition

pose recognition

expression recognition and face cartoon

etc.In this paper

we use a multi-stage facial features detection algorithm proposed by David Cristinacce and Tim Coote to implement a real-time facial features localization system.Meanwhile

we also improve the origin facial features localization algorithm to run much faster with only a little loss in accuracy.