Auto food tracking could be big business but nobody’s cracked it yet. We’ve seen all sorts of ideas to replace manual entering and photo recognition from a smart necklace that counts calories through sound to the controversial calorie counter HealBe GoBe.
Throwing another idea in the ring is Japanese network NTT with its ‘Dietary Content Recognition Technology’ being developed for smartwatch-style wearables.
It is designed to estimate what the wearer is eating by analysing the movement’s of their arm while eating. All the tech needs to do this is data from accelerometer and gyroscope inside a wearable. The idea is that our arms move differently and at a different frequency depending on what type of food we are consuming.
The tech has been tested on a few meals including bread, sushi, curry, pasta and a bowl of beef. It was able to correctly recognise each food though we don’t have details on accuracy levels yet – that’s the next focus of NTT’s market research. It then automatically logs these foods in a database.
One potential snag is that the nutrition and number of calories found in one portion varies widely between different curry or pasta dishes, for example.
At the NTT R&D Forum 2017 where this was presented, the developers said the technology isn’t ready yet for consumer use. Still, we’re sure the likes of Apple, Fitbit and Under Armour are searching for the next step in auto food tracking. Estimating meals from how we eat might not be the complete answer but we can see it working together with biometric or audio tracking.
Source: Japan Today