Determination of Calorie Content in Different Type of Foods using Image Processing

Abstract

This paper discusses a system that estimates the weight and calorie content of a food specifically chicken and fish based single image by using a fixed-placed camera in a hardware system. Convolutional Neural Network (CNN) was the integrated algorithm to recognize food on an image. Graph Cut image segmentation was used to analyze to determine the regions of the food in the image. Volume estimation based its measurement on the area of the segmented food image and the height of the food measured by fixed-placed ultrasonic sensor. The system was tested by each part of the chicken and fish for 10 trials each as well as if it is fried or grilled which resulted to a food detection accuracy of 91.82%, a mean accuracy of 88.18% for the calorie estimation.