How to use vision.foreground detector to extract human silhouette from webcam's video input

I'm familiar with using a video as an input. Now I need to give webcam as input. I want to initialize it with a static background and then to track the silhouette when a human enters into the frame.