TECH LIFE

NEW WATCH ON WEIGHT

Diet analysis on the plate is his goal.

JEFF GELLES @JeffGelles

Tracking your calories. Watching your salt intake. Eating less sugar and more vegetables. Monitoring macronutrients such as protein, fats, and fiber.

All are among things health-savvy Americans try to do nowadays, sometimes the old-fashioned way — say, with notebooks and calorie charts

— and sometimes online or with mobile apps. But as anyone who tries food tracking can tell you, it’s tough to keep up.

Philadelphia’s Anthony Ortiz says he’s invented a better way: SmartPlate, a high-tech, patent-pending tool he unveiled this week at the Collision, a competitive technology expo in Las Vegas, and on Kickstarter, the crowdfunding website.

How ambitious is the product Ortiz aims to sell for $199 by mid-2016? He calls it “the world’s first WiFi- and Bluetooth-enabled connected-kitchen device that will instantly track and analyze everything you eat.” That’s a mouthful of words, but Ortiz is backing it up with serious technology.

The prototype he began pitching publicly Monday — to Kickstarter early birds for $99 and to potential investors at Collision — looks like a basic, hard-plastic, compartmentalized plate. But it’s not dishwasher- or microwave-

safe, and for good reason: Beneath its surface are cameras, load sensors, an LCD controller, and computer chips.

And that’s just the remote device. The real technological chops of Ortiz’s SmartPlate will be in the cloud, where a sophisticated algorithm will put all those tools to use.

Place, say, an apple, a pile of green beans, and a mound of chicken-and-rice pilaf into the three compartments, and the plate will go to work. Images of each type of food will be analyzed and matched against a U.S. Department of Agriculture database of 8,000 foods. The load sensors will weigh each portion. Calorie and nutrient values will then be sent to SmartPlate’s iOS or Android apps, and to platforms of other popular health-and-fitness products, Ortiz said in a recent interview in his co-working space at the University City Science Center.

Ortiz, 38, a serial entrepreneur who also runs Fitly, a fresh-food delivery service, plans to link his invention with tools such as the Fitbit activity tracker, the Withings Wireless Scale, and MyFitnessPal, an online food-and-activity diary he says “millions of people already use.”

How well will all of this work? That remains to be seen — along with its potential to aid people managing weight, or illnesses like diabetes, or aiming for particular fitness goals.

“Technology like this could really help, because it provides real-time feedback on how we’re eating,” says Jennifer Orlet Fisher, a Temple University public-health professor and obesity researcher not connected with Ortiz or Fitly. Fisher says most people “are really horrible at estimating portion sizes.”

Still, Fisher sees “lots of devils in the details. Whether it can accurately identify what you’re eating and any variations — for instance, how much butter or salt you’re adding — is hard to imagine.”

Ortiz says that’s where the US-DA’s estimates prove invaluable. And he says a cook also can easily tell the system about added ingredients — even by scanning the bar code on a stick of butter. SmartPlate also will have access to 300,000 commercial product codes and about 100,000 restaurant foods.

Ortiz acknowledges his algorithm is a work in progress, though he says it’s “up to 99 percent accurate” for some foods. “It is sure that’s an apple. It is sure that’s broccoli,” he says. “Mixed foods create complexity.”

Ortiz says the plate’s camera can “survey the landscape” of a dish that includes a visible mix of rice, carrots, and chicken, and might inform the user that “there’s a 70 percent chance this is rice pilaf.” If the user confirms, it then turns to the USDA data.

His own confidence was greatly boosted last year when he connected with Dah-Jye Lee, a Brigham Young University expert in object recognition and machine learning, who became a consultant on the project.

Lee, codeveloper of an algorithm used for food-quality control, says one key is providing as many different perspectives on a food as possible, so the machine can learn how it looks from any direction.

“It’s kind of like how humans learn. We teach a child, this is a dog, this is a cat,” he says. “After they learn, if they look at the backside of a dog, they still know it’s a dog.”

Ortiz says the power of machine learning is key to the project. “It’s going to come down to how well we train our data,” he says. “It’s like training new eyes to see for the first time.”

Ortiz has a personal motive for his business focus on healthy eating and fitness. His father has heart disease, and other relatives suffer with diabetes and obesity.

“I was concerned for my little nephews. They were eating the same things that my brother and his wife were eating,” Ortiz says. “What I realized is that behavior change is really tough.”

With Fitly as a healthy food source and SmartPlate as an “accountability tool,” Ortiz is aiming to make that just a little easier to accomplish. jgelles@phillynews.com

215-854-2776 @jeffgelles www.philly.com/consumer