Head Pose Estimation

Based on 2D
--Based on Appearance
--Based on Facial Features
Based on depth information
--Based on DCCE
--Based on random regression forest
--Based on stereo cameras

Pros
-Accuracy is high
-Not use GPU
-Work  in some difficult situations
-Using the depth image
-Speed is ok
Cons
-Ideas is not very creative
-Need large database to train

相关PPT下载详见 “视觉计算研究论坛”「SIGVC BBS」：http://www.sigvc.org/bbs/thread-30-1-2.html