Google has been building an automated food dairy which could help to estimate calorie counts from the images taken of our meal according to a report in Popular Science. Named Im2Calories, the project is said to utilise technology from DeepMind, which is an AI start-up Google acquired in 2014. Google has filed a patent for Im2Calories, however there is no mention yet on when the service is likely to be made available.
Meanwhile, several other companies are making similar attempts on the projects like Im2Calories which are being supported by equally questionable science. Im2Calorie is said to measure the size of the food, in absolute terms as well as relative to what is on the plate and could also identify the condiments. According to the report, it does not even matter if the picture tends to be high resolution or not. Presently, the technology is still imprecise in estimating calorie counts.
According to Kevin Murphy, a researcher at Google acknowledges as much to Popular Science stating that `it’s okay fine, maybe we get the calories off by 20%. It does not matter’. The focus for Google is in minimizing those inaccuracies over a period of time, claims Murphy. There does not seem to be any doubt in Google’s ability in collecting and processing data from large section of users though the company would still have to solve a big problem for the function of Im2Calories – calorie counting could be unreliable.
Counting Calories – Look at Kinds of Calories Eaten
A main study published in The New England Journal of Medicine in 2011 observed that the quality of food is a main contributing factor than quantity. Dr Dariush Mozaffarian, lead author of the study informed the New York Times - `Conventional wisdom – to eat everything in moderation eat fewer calories and avoid fatty foods, isn’t the best approach.
Just counting calories won’t matter much unless you look at the kinds of calories you’re eating’. However if calories seems to a good measure for health, food labels, - the basis of Google’s date, has also considered to be inaccurate. In 2010, there has been much uncertainty with regards to calorie counting where David Kirchoff, the CEO then of the Weight Watchers acknowledged that `calorie-counting has become unhelpful’ for those intending to lose weight. Since then the company is said to have abandoned its calorie first approach.
Im2Calories – A Potential Tool
Im2Calories, which is the potential tool, had been announced at the Rework Deep Learning Summit, recently. Murphy had unveiled a project that utilises `sophisticated deep learning algorithms to analyse a still image of food, estimating how many calories could be on the plate. In one instance, the system viewed an image and counted `two eggs, two pancakes and three strips of bacon.
Although the food stuff does not seem to be a universal unit of measurement, the system could scale the size of each piece of foot with regards to the plate together with the condiments. Murphy has no intention of shaming users with this new system but wants to have the `process of keeping a food diary and awareness of foods and this seems to be much easier than feeding information manually to a food app which could include portions of food items, types of food and much more.
It is assumed that the latest tool Im2Calories could be popular especially in the US since obesity tends to be a crisis. However, even if Im2Calories does not seem to be accurate, Murphy is of the opinion that the new system will tend to have an impact on the users.
