Computer visions differn from human vision in several ways. The human vision does not need pre-programed information about visual contexs to understand it directly. However computer vision is unable to do so without pre-programmed assumption about the context, or else it fails to understand. Human vision is also better at image level understanding instead of understanding or counting pixels, while computer vision is the exact opposite.
Some techniques to help computers see and track would be background subtraction: to help isolate moving object; brightness thresholding: to detect change in picel brightness to infer change in the whole image; high contrast: so that the subject would stand out; simple backgrounds: to make background aubtraction and brightness thresholding easier.
Computer vision’s special capacity make it useful for bringing in a new perspective in interactive art. A computer vision will get differnt information or emphasis from the same image as a human, and the comparison or cooperation between the two would be a great lead for interaction. The computer vison is also objective and can pick up details, giving it the potential to learn its interactive audience better than a human would.