Incorporating Embedded Vision into Systems

Consider a machine that not only recognizes you but can also tailor its work to suit your preferences. It also remembers any recommendations you make and uses it the next time you employ the machine. Now think of a toaster with these capabilities. When you turn the toaster on, it will greet you by name, ask you how you prefer your toast and bake it appropriately. After the bread pops out, the toaster will ask you how it’s been done. If you give suggestions, the toaster remembers them for the next time.

This is called embedded vision and some people say that it is still far off. The fact is, it may be closer than you think. Some companies have already developed chipsets that can make embedded vision a reality. One of these chipsets is capable of human body detection, face detection, hand detection, face recognition, age estimation, gender estimation, gaze estimation, blink detection, mood/expression estimation among other things. The interesting thing is, all of these features are combined in a small 60mm x 40 mm printed circuit board.

There are so many applications of this technology. For example, consider that a flat screen display is embedded with this feature. Before it runs any ads, the display will scan the person in front of it, analyze his features and deliver tailor advertisements, most likely to interest the person. Another application is a vending machine that shows food and beverage recommendations, according to the gender or age of the customer.