AI wants to count your calories bite by bite
FoodTracker: An artificial intelligence-driven mobile app for food testing
McGill University in Canada is conducting a study called “FoodTracker” that uses artificial intelligence (AI) and smartphone cameras to record and analyze diners ‘food intake.
Researchers are using AI to record diners through mobile phone cameras to analyze food intake, bite by bite.
The algorithm measures the amount of food on the spoon from the plate to the mouth and aims to improve calorie and nutrition tracking beyond traditional food diaries and apps. Although the algorithm currently focuses on portion size, it is expected to identify food types within a few months. This research has broader application potential, especially in the growing market for diet and nutrition applications.
The method currently has an error rate of 22%, but researchers are working to improve accuracy and adapt it to different tableware. Funded by the National Research Council of Canada, the project focuses on addressing malnutrition in the elderly population. Prototypes are expected to be tested within next year.
This app can identify the ingredients of food in real time and provide corresponding nutritional information. Researchers have developed a model that efficiently identifies food by combining deep convolutional neural networks (CNN) and YOLO detection strategies. Not only will this app help users better understand their diet, it may also be widely used in the future to improve people’s eating habits and health (Tech Xplore) (HillNotes).
“The laboratory focuses on health-related applications on embedded systems,” Zelijko Zilic, one of the researchers who conducted the study, told TechXplore. “Our goal is to introduce automation into food diaries so that people or patients who care about their daily diet can continuously track meal items and nutritional components throughout their daily lives. To achieve this, we have been providing apps for the iPhone (DiaBeatMove and CarbAndMove) to help diabetics and prediabetics manage exercise, nutrition, insulin and health-related aspects of their lives.”
Rising obesity rates in the United States and other countries around the world and issues related to malnutrition have encouraged many researchers to develop mobile apps or online platforms to promote healthier lifestyle choices. In recent research, Zilic and his colleagues developed a smartphone application that quickly and effectively identifies what foods users are eating in real time and provides nutritional content for each ingredient in the meal.
FoodTracker is a mobile application developed by researchers that is very easy to use. When a user points a smartphone camera at the plate containing his/her meal, the app quickly identifies its different ingredients.
First, Zilic, Sun and their colleagues developed a model that combines a deep convolutional neural network (CNN) with YOLO, a state-of-the-art detection strategy. They trained the model using an extensive database of food images and found that it achieved an average accuracy of nearly 80% when detecting food based on images.
In the future, mobile apps such as FoodTracker could broaden people’s nutritional knowledge, support them in understanding the food they consume every day, and may even help them improve their eating habits. Zilic, Sun and their colleagues are now planning to integrate nutrition-related guidance provided by the app with other modules to encourage healthier lifestyles.
The research is supported by Canada’s National Artificial Intelligence Strategy, which aims to promote the application of AI technology in areas such as health
If you want to learn more, you can click on the link below the video.
Thank you for watching this video. If you like it, please subscribe and like it. thank
Thesis:https://arxiv.org/pdf/1909.05994
Report:https://techxplore.com/news/2019-09-foodtracker-ai-powered-food-mobile-application.html
Original text:https://www.wsj.com/tech/ai/ai-count-calories-weight-loss-6acc7019
Oil tubing:
