[转载]Kinect for Windows V2是否提供人脸识别功能？

Does Kinect for Windows SDK offer facial recognition capabilities?

Since version 1.5, the Kinect for Windows SDK and toolkit have supported real-time face tracking. Face tracking is not intended for facial recognition or identification. Face tracking detects and tracks the positions and orientations of faces in real time and provides animated three-dimensional (3D) meshes that are positioned in 3D over the Kinect depth sensor data. Face tracking can animate, in real time, eyebrow positions and mouth shape, as well as provide a high-resolution 3D mesh of the face than can be animated. With the SDK 2.0, Kinect can also track whether eyes are open or closed, the position of the nose, facial expressions, and the direction the face is pointing. The resolution of the tracked face has been increased by 20 times with a 1,360-point mesh, so that faces look more realistic. Conventional face-recognition algorithms can be used with the Kinect for Windows sensor’s RGB stream; we continue to investigate how the Kinect for Windows sensor’s extra capabilities can be used.