There are mobile applications that help your record your intake and nutrition. But, there is nothing that helps you detect them real-time saving time and improving precision. Japanese scientists have developed a mobile application that works on AI or artificial intelligence and helps you measure your intake.
The developers presented their finding at the 16th international conference on machine vision applications in Tokyo. The researchers claim that this is a plausible tool that can help diabetics or pre-diabetics record food intake, exercise, insulin intake, and more.
80% Accuracy in Detecting Food
To achieve their goal, researchers combined YOLO – a detection strategy along with CNN (convolutional neural network). Further, researchers trained the device using a database of food images. With time, the device reached almost 80% accuracy in detecting food items on the plate and deciphering its nutritional quotient.
Besides the mobile application’s accuracy, the researchers also boast about its low runtime and less memory it requires for storage on the phone. Further, they also believe that this app is important as the diabetic and obese population is growing not only in the US but across the world. Also, using this app, people can adopt a better lifestyle and streamline their habits to achieve health goals.
While the scientists are hopeful about their product, they do not deny the room for improvement. They say that aspects like multi-task learning, volume estimation, better generalization, and robustness require improvement.