Automated Household Food Management and Recipe Recommendation System Based on Visual Recognition and LLM Knowledge Base

Abstract

With continuous economic development and intensifying competition, individuals increasingly face the conflict between the pursuit of a better life and the demand for efficiency. This conflict is particularly pronounced in the realm of dietary habits. A healthy and well-balanced diet often requires significant time investment in planning and decision-making, yet people frequently lack sufficient time for such considerations. As a result, the need for an efficient system to assist with meal planning and food management has become more apparent. To address this challenge and help individuals achieve a balance between healthy eating and lifestyle efficiency, we have developed an easy-to-use, natural language-based automated household food management system leveraging visual recognition technology, large language models (LLM), and knowledge base technologies. The system automates household food management tasks, including inventory input, stock display, expiration monitoring, and food output. Additionally, it customizes personalized recipes based on factors such as the current time, number of family members, taste preferences, special dietary needs, and available ingredients. According to user surveys, over 70% of respondents recognized the necessity of the system, and its recipe design received an average rating of 3.75 out of 5, indicating that the majority of users found the system’s recipe recommendations acceptable.