Predicting Recipes from Photos
We all know that there are few
things on social media on which users love more than flooding their feeds with
the photos of food and dishes. Still most of the people use these images for
much more than a quick scroll on their cell phones. But the team of researchers
believes that analyzing these kinds of photos could be very helpful in learning
recipes as well as better understanding the eating habits of people.
This research has been published in many news
papers by Top Private Engineering College in Jaipur in which the researchers have
trained an artificial intelligence system which is called pic2recipe in order
to look at the photo of food as well as be able in order to predict the
ingredients and suggest similar recipes. According to the researchers, in the
computer vision, the food is the most neglected thing as the researchers do not
have the large scale datasets which is needed in order to make predictions. But
it seems that these useless photos on the social media can be very helpful in
providing valuable insight into health as well as dietary preferences.
The web has spurred a great growth of research
in the area of classifying the food data but much smaller data sets have been
used by the majority of it that generally leads to big gaps in labeling the
foods. A few years ago, the food 101 has been created by the team of
researchers as well as has been sued in order to develop an algorithm which
could recognize the images of food with maximum percentage accuracy as about 80
percent accuracy has been improved by the future iterations which also suggest
that the size of the dataset may be a limiting factor.
According to the Best EngineeringColleges Rajasthan 2016, even the larger datasets have generally been
somewhat limited in generalizing across populations. The team of this project
aims to build off of this work but it has been dramatically expand in scope.
The websites like all these recipes as well as food have been combed by the
team of researchers in order to develop a database over 1 million recipes which
have been annotated with the information about the ingredients in a broad range
of dishes.
Comments
Post a Comment