Проєктування мобільного android-застосунку для розпізнавання та пояснення складу харчових продуктів з фокусом на українського користувача

Abstract

This paper presents the concept of a mobile Android application for recognizing and explaining food product ingredients with a focus on Ukrainian users. The study outlines the structure of a custom dataset of food label images, the selection of a multimodal large language model (MLLM) for ingredient recognizing and explaining, and the core functional and technological decisions behind the application. ChatGPT and Gemini are considered as the primary model candidates, while React Native and FastAPI are chosen as the technological stack.