Classifying Glycemic Index from Food Images with Convolutional Neural Networks

• Jailson Januário Universidade do Estado do Amazonas
• Elloá Guedes Universidade do Estado do Amazonas
• Fabio de Silva Universidade do Estado do Amazonas

DOI: https://doi.org/10.5753/eniac.2019.9309

Abstract

The identification of food glycemic index is fundamental for dietary planning and glycemic control, especially in patients with diabetes. In order to determine the glycemic index from food images, this paper considers the use of a convolutional neural network architecture to address two classification tasks, one with only fruits and the other with foods in general. Upon evaluating the proposed model, we verified an accuracy of 98.22% for the first task and of 84.30% for the second task, results that are competitive with the state of the art and that motivate the development of computational solutions that promote the improvement of quality of life through healthy eating.

Keywords: Convolutional Neural Networks, Deep Learning, Food Recognition, Image Recognition

References

Atkinson, F. S., Foster-Powell, K., and Brand-Miller, J. C. (2008). International tables of glycemic index and glycemic load values: 2008. Diabetes Care, 31(12):2281–2283.

Brownlee, J. (2016). Deep Learning With Python. Machine Learning Mastery, Estados Unidos.

Buduma, N. (2017). Fundamentals of Deep Learning. O’Reilly Media, Estados Unidos.

Chollet, F. (2018). Deep Learning with Python. Manning Shelter Island.

Freitas, C. N. C., Cordeiro, F. R., and da Silva, A. J. (2018). Food recognition system for nutrition monitoring. In Anais do XV Encontro Nacional de Inteligência Artificial e Computacional, pages 186–197, Porto Alegre, RS, Brasil. SBC.

Greenwood, D. C., Threapleton, D. E., Evans, C. E., Cleghorn, C. L., Nykjaer, C., Woodhead, C., and Burley, V. J. (2013). Glycemic index, glycemic load, carbohydrates, and type 2 diabetes. 36(12):4166–4171.

Imran, S. A., Agarwal, G., Bajaj, H. S., and Ross, S. (2018). Targets for glycemic control. Canadian Journal of Diabetes, 42:S42 – S46. Diabetes Canada 2018 Clinical Practice Guidelines for the Prevention and Management of Diabetes in Canada.

Khan, S., Rahmani, H., Shah, S. A. A., and Bennamoun, M. (2018). A Guide to Convolutional Neural Networks for Computer Vision. Morgan & Claypool, Austrália, 1 edition.

Marsh, K., Barclay, A., Colagiuri, S., and Brand-Miller, J. (2011). Glycemic index and glycemic load of carbohydrates in the diabetes diet. Current Diabetes Reports, 11(2):120–127.

Mezgec, S. and Seljak, B. K. (2017). Nutrinet: A deep learning food and drink image recognition system for dietary assessment. Nutrients, 9(657):1–19.

MMSPG (2017). Multimedia Signal Processing Group (MMSPG), Ecole Polytechnique Fédérale de Lausanne. Lausanne, Suı́ça. Disponı́vel em https://mmspg.epfl. ch/downloads/food-image-datasets/. Acesso em 27 de agosto de 2019.

Muresan, H. and Oltean, M. (2018). Fruit recognition from images using deep learning. Acta Univ. Sapientiae, Informatica, 10(1):26–42.

Philippi, S. T., Latterza, A. R., Cruz, A. T. R., and Ribeiro, L. C. (1999). Pirâmide alimentar adaptada: Guia para escolha dos alimentos. Revista de Nutrição Campinas, 12(1):65–80.

SBD (2017). Diretrizes da Sociedade Brasileira de Diabetes 2017-2018. Sociedade Brasileira de Diabetes. Clannad Editora Cientı́fica.

Silva, F. M. and de Mello, V. D. F. (2006). Índice glicêmico no manejo do diabetes melito. Revista HCPA, 26(2):73–81.

Simonyan, K. and Zisserman, A. (2015). Very deep convolution networks for large-scale image recognition.