Is user-input required?
	Our visual system is proof that automatic methods are possible • classical image segmentation methods are automatic Argument for user-directed methods? • only user knows desired scale/object of interest