This is an example of precise face-tracking in an image using the javascript library clmtrackr. To try it out, simply click start.To try it out with your own image, choose a file above by clicking "choose file". If the tracking has problems, try selecting the face in the image manually by clicking "manually select face", and click and hold to drag a square around the face in the image.The image is from the FG-net Talking Face project
[Edit, June 2016: To the best of my knowledge, this work with Arturo Castro is the first example of realistic "face swapping" in realtime. If you're interested in experimenting with the technique, I recommend Audun's implementation here auduno.github.io/clmtrackr/examples/facesubstitution.html ]
clmtrackr is a javascript library for fitting facial models to faces in videos or images. It currently is an implementation of constrained local models fitted by regularized landmark mean-shift, as described in Jason M. Saragih's paper. clmtrackr tracks a face and outputs the coordinate positions of the face model as an array, following the numbering of the model below: