Forget facial recognition, now machines can identify pastries

It's a good thing we got the whole facial recognition thing figured out, because as it turns out there was another high tech problem looming. The sticky problem of pastry identification.

Grocery stores have pretty much mastered the whole bar code thing, so Japanese researchers upped the retail stakes with a scanner that uses a camera to distinguish items — something we reported on in March when it was in the demo phase.

Meanwhile, back in the real world the pastry problem was taking shape. Scientists working on scanner recognition realized some items — namely breads and pastries — look a lot alike. There could be hundreds of variations in bakeries, plus with the addition of unique toppings or custom ingredients it could stymie scanners and staffers alike.

In cooperation with the University of Hyogo, the Brain Corporation has developed an updated camera scanner system that should eliminate the guesswork.

The new system also relies on cameras, but to identify items properly they are positioned to better identify a product. While items are placed on a scanner, they are also being identified via camera from above.

A staffer will know if the item has been identified correctly if a green outline surrounds its scanned image. If it is yellow there is a doubt, and the staff can choose from a list of probable items. By gradually collecting data on entries that have had to be entered by hand, the system will learn to better identify items and hopefully be scannable in the future.

Retailers are quick to see the benefit of having more reliable scanners for pastries and other challenging items. They can help staff — notably part timers — with checkout duties without them having to remember items. This helps keeps the lines moving and customers happy.

While the new system was developed for bakeries and is in active trials at a shop in Tokyo, the developers of the system point out that it can help identify other goods as well.

That's good because frankly we weren't quite sure how the first system was going to recognize the difference between a regular avocado and a Hass avocado, but this system is likely to wrestle that problem under control in no time.

Here's hoping we see it in the States sometime soon!

Via DigInfo

For the latest tech stories, follow DVICE on Twitter
at @dvice or find us on Facebook