Face detection is the process of locating the position where faces are present in an image.Currently, software-based face detection is hard to support real-time video streams in an embedded processor. By hardware and software co-analysis, we design a hardware imple-mentation of face detection which satisfies embedded constraints. The experimental results of our hardware design show that the video streams provided by RGB camera can be exe-cuted in real-time and the cost of hardware resource is also relatively low. We also analyze our detection rate and false positive number under different threshold settings and lighting conditions.