Complex Food Recognition using Hyper-Spectral Imagery

Document Type

Conference Proceeding

Publication Date

1-6-2020

Publication Title

2020 10th Annual Computing and Communication Workshop and Conference (CCWC)

Publisher

Institute of Electronics and Electrical Engineers

Publisher Location

Las Vegas, NV

First page number:

662

Last page number:

667

Abstract

Developing a rapid and low-cost approach for automated food recognition is a necessity for applications such as dietary assessment to determine the caloric and nutritional food intake for short-term rehab and elderly care centers are very critical for the health care system. The main step in dietary assessment is not only to identify (classify) different components of the food but also to identify food cooked by different processes. Many researchers have focused their efforts on developing image-based learning techniques of food classification. Hyperspectral Imagery (HSI) allows for examining the spectral response of foods over a larger frequency spectrum. This paper presents the use of hyperspectral images for the detection of different foods in a meal. The hyperspectral data of food have been collected under a controlled illumination environment and food classified using SVM (Support Vector Machine) and Logistic Regression classifiers. We compare RGB and hyperspectral data classification on a set of common foods and present results.

Keywords

Food Classification; Hyperspectral Imaging; Logistic Regression; RGB; Support Vector Machine (SVM)

Disciplines

Computer Sciences

Language

English

UNLV article access

Share

COinS