Застосування глибокого навчання для дієтологічного контролю продуктів харчування

Abstract

This paper presents an intelligent dietary monitoring system using deep learning for automated food classification and calorie estimation. By leveraging a DenseNet121 and ResNet-50 ensemble, the study ensures high feature extraction efficiency and robust recognition on the Food-101 dataset. The research proposes a methodology for nutritional assessment based on reference-object volume estimation. The approach aims to automate dietary tracking, reducing manual logging errors in mobile health applications.