Keeping a journal of what and when you eat is one of the standard ways of tracking your diet. That said, it’s a rather inexact method, which is why scientists are creating an eyeglasses-based system that may do the job more accurately.
Known as FitByte, the experimental technology is being developed at Pittsburgh’s Carnegie Mellon University. It consists of multiple sensors that can be mounted on existing third-party glasses.
For starters, there’s an infrared proximity sensor at one side of the front of the glasses that recognizes the distinctive hand-to-mouth motions associated with eating and drinking. These movements trigger a camera on the other side that captures images of the food or beverage items, for the wearer’s records.
Although the user currently has to manually identify what types of foods and drinks were photographed, plans call for that task to eventually be handled by an artificially intelligent computer vision system.
There are additionally a total of six inertial measurement units (which combine an accelerometer and a gyroscope), located in the glasses’ ear hooks and nose bridge. These are used to identify the jaw motions associated with chewing, along with the throat vibrations that accompany swallowing.
All of the data is processed offline (onboard the glasses), providing a record of what the user ate and drank, in what amounts, and at what times. Non-invasive sensors that track blood glucose levels and other physiological factors may be added down the line, along with connectivity with a dedicated smartphone app.
“Our team can take sensor data and find behavior patterns,” says Asst. Prof. Mayank Goel. “In what situations do people consume the most? Are they binge eating? Do they eat more when they’re alone or with other people? We are also working with clinicians and practitioners on the problems they’d like to address.”
We’re told that a commercial product could be available in around three years.