Comparison of convolutional neural network models for food image classification

• Published in: 2017 IEEE International Conference on INnovations in Intelligent SysTems and Applications (INISTA) pp. 349 - 353
• Main Authors: Ozsert Yigit, Gozde, Ozyildirim, Buse Melis
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01-07-2017
• Subjects: Convolution; Convolutional Neural Network; Deep learning; Diabetes; Food Classification; Graphics processing units; Kernel; Neural networks; Nutrition Categorization; Obesity; Training
• DOI: 10.1109/INISTA.2017.8001184

According to some estimates of World Health Organization (WHO), in 2014, more than 1.9 billion adults aged 18 years and older were overweight. Overall, about 13% of the world's adult population (11% of men and 15% of women) were obese. 39% of adults aged 18 years and over (38% of men and 40% of women) were overweight. The worldwide prevalence of obesity more than doubled between 1980 and 2014. The purpose of this study is to design a convolutional neural network model and provide a food dataset collection to distinguish the nutrition groups which people take in daily life. For this aim, both two pretrained models Alexnet and Caffenet were finetuned and a similar structure was trained with dataset. Food images were generated from Food-11, FooDD, Food100 datasets and web archives. According to the test results, finetuned models provided better results than trained structure as expected. However, trained model can be improved by using more training examples and can be used as specific structure for classification of nutrition groups.