NUTRITION ESTIMATION OF LEFTOVER USING IMPROVED FOOD IMAGE SEGMENTATION AND CONTOUR BASED CALCULATION ALGORITHM

Abstract

In pandemic conditions, awareness of keeping a healthy balance is necessary. One is considering food consumption and understanding its nutrition content to avert food waste. We have been developing a prototype to estimate the nutrition of leftover food, and the main problem lies in image segmentation. Therefore, we propose the Improved Food Image Segmentation (IFIS) and Contour Based Calculation (CBC) to measure the area of the segmented image instead of pixel-wise. First, the tray box image is acquired and broken down into compartments using an automated cropping algorithm. The first step of this proposed method is tray box image acquisition and dividing the compartment using an automatic cropping algorithm. Then each compartment is treated using IFIS, calculates the result of IFIS by CBC, measures the estimated leftover by Automatic Food Leftover Estimation (AFLE), and then predicts the nutritional content. The evaluation is applied by comparing the actual measurement from the Comstock method and leftover estimation by the proposed algorithm. The result shows that Root Square Means Error (RMSE) reaches 0.48 compared to the actual weighing scale and 96.67% accuracy compared to the Comstock method. Based on the results, the proposed algorithm is sufficient to be applied.