Maybe that “flood” of food photos on Instagram start serve some purpose: Google developed an artificial technology to calculate the calories present in pictures of food.
Im2Calories call, the technology was announced during Rework Deep Learning Summit, and can hit up to 80% the number of calories present in a photographed plate.
Kevin P. Murphy, Google researcher, explains that the technology analyzes the pixels in depth and apply ‘sophisticated algorithms to deep learning system “to assess the size and shape of the food, and come up with an approximate value of calories it contains.
A demonstration made during the conference recognized two eggs, two pancakes, three strips of bacon and even the condiments present a photo. And it goes without even they are high resolution images for an approximate amount of calories: the quality of Instagram pictures is enough. A patent technology behind the Im2Calories has already been registered
Murphy believes that this technology can be useful to build statistics at the population level:.. Which the most popular dishes in a region of a city, for example
It also says that technology can make it simpler registration of calories in a journal, and prevent anyone lie about the amount of calories consumed in a meal.
Murphy explains that technology learn from their mistakes, differentiating one fried egg of a poached egg over time. The algorithms, however, are still being researched and there is no definite date for launching an application with the technology to the public
.
Comments
All comments are subject to moderation. DD reserves the right to delete comments that do not comply with the rules of use. The published comments are the sole responsibility of their authors.
No comments:
Post a Comment