SmartFood AI: A Multimodal Framework for Accessible Nutritional Analysis

This work presents an advanced food detection system leveraging computer vision and AI to enhance food recognition and nutritional analysis, aligning with global health goals set by organizations like World Health Organization (WHO) and the Food and Agriculture Organization (FAO). These organizations emphasize the critical role of nutrition in public health and advocate for accessible, accurate dietary information to combat malnutrition and promote balanced diets. Powered by a deep learning based framework, the proposed system delivers high-precision food identification through a bounding box regressor for localization and a knowledge graph for nutritional insights. Its zero-shot learning capability enables recognition of previously unseen foods without additional training, addressing the limitations of traditional systems by providing accurate identification alongside contextual nutritional data. This adaptability allows the system to evolve with changing dietary trends and meet diverse user needs, supporting applications in healthcare, personal nutrition monitoring, and food. Moreover; It is beneficial for visually impaired individuals, enhancing autonomy by enabling them to recognize foods and access essential nutritional information independently, thereby promoting healthier living. The system demonstrates robust performance across varied conditions, making it effective for both personal use and public health applications. By providing critical dietary insights and adapting to new foods, this AI-powered solution represents a scalable and impactful tool in computer vision and dietary monitoring.It promotes global health initiatives and personal well-being by providing a useful strategy for raising awareness of nutrition and encouraging better lifestyles.