FOOD DENSITY ESTIMATION USING FUZZY INFERENCE
Chengliu Li · D-Scholarship@Pitt (University of Pittsburgh) · 2012
This work presents a new fuzzy logic approach to food density estimation in supporting research in diet and nutrition. It has been a historical problem to measure people’s daily food intake in real life. Recent advances in electronic devices have provided novel tools for recording volumetric information of food, while the current food databases often list nutrients and calories in terms of gram weights instead of volumes. Thus, a density value, which connects the volume to weight, is required to use the existing databases when the volumetric information is unavailable. In this work, we approach the density estimation problem using fuzzy inference which “guesses” the food density by collecting and organizing relevant human knowledge about a food. French fries are taken as an example of this new approach. A fuzzy Inference System (FIS) is constructed to estimate the bulk density of French fries under different cooking conditions. Our experimental results show that our FIS system built upon human knowledge about the frying time and temperature can accurately estimate the density of French fries under controlled conditions.