Multi-sensor integrated food volume measurement based on smartphone

Abstract

Accurate dietary records are important for chronic disease management. Current food management applications rely on users to manually estimate and enter the volume of food, not only the operation is complex but also the accuracy is not high. In this context, this paper designs and implements a food volume measurement algorithm based on smart phones that integrates image and sound sensor data. In this paper, the algorithm of shape matching and color matching is designed to detect and locate the container and food according to the task of food detection. Two basic models are proposed according to the characteristics of Chinese cuisine. In addition, this paper also designs a mobile phone-based acoustic echo ranging algorithm with high accuracy for calculating the actual volume of food. Experiments show that the accuracy of the proposed algorithm is 10% higher than that of the manual recording method. Compared with the image analysis method using the reference object, the accuracy is improved by 5%.

Publication
The 14th Annual Conference of Pervasive Computing of China (PCC) 2018