vvvv ships with a range of video tracking nodes, all implementated as FreeFrame plugins.

These nodes have pretty good helppatches and speak for themselves.

The Image Pack provides plenty of nodes which offer computer vision and image manipulation algorithms defined in the OpenCV library.

These are the options vvvv offers for tracking faces. Look at their helppatches for details:

• via Kinect: FaceData (Kinect Microsoft). Kinect1 tracks a maximum of 2 faces. Not sure about the limits of Kinect2.
• via Freeframe:
• DetectObject (Freeframe DShow9) and a corresponding haarcascade file. Tracks multiple faces.
• FaceTracker (Freeframe DShow9)
• via CV.Image: DetectObject (CV.Image) and a corresponding haarcascade file. Tracks multiple faces.

Check the external Tracking Tools that can communicate with vvvv.