Software development company throws its hat in the ring to try to passively track nutrition

Following DataArt’s announcement of DataArt ORANGE, MedCity News speaks with Eugene Goland about the role of automated food recognition and computer vision in nutrition tracking and healthy lifestyle.

Of all the biometric data that health wearables and mobile apps can track, from miles moved to hours slept to heart beats per minute, no one has yet been able to nail down a way to accurately and passively monitor food intake.

Now there’s a new player in the [health tracking] game and it’s software development firm DataArt, which is launching a series of R&D initiatives in an effort to automate nutrition tracking.

The firm’s technology takes a bit of a different approach. It scans pictures of food taken by the user and uses photo recognition to determine what it is. It also builds a context around the photo using geographic location and time of day to determine what the food in the picture might be. For example, if a person is in a restaurant in the evening and snaps a picture of food, DataArt may be able to automatically look up that restaurant’s dinner menu, said company President Eugene Goland.

The system is in prototype form and can currently identify about 100 foods, Goland told me. Once more foods are added and the team is sure the technology is a accurate as it can be, the hope will be to integrate with other tracking apps or devices within the next 12 months.

“We’re trying to position ourselves as a hub for food recognition,” he said.

This approach seems a little simpler than some of the others under development and also resembles efforts by teams at University of Washington and Purdue University and Purdue University.

Whoever can successfully bring a passive calorie tracker to market will get a piece of the $20 billion mobile health market expected by 2018.

DataArt is calling its project Orange after one of the easiest foods to identify. The company has offices in New York and throughout Europe.”

View original article or download PDF