Vitamin Deficiency Detection System Using Efficientnet-B4

Abstract A significant global health concern, vitamin deficiencies can result in major medical complications like fatigue, neurological problems, weakened immunity, and vision problems. Conventional diagnostic techniques depend on clinical evaluation and blood tests, which are costly, invasive, and difficult to obtain in remote locations. The EfficientNet-B4 architecture, which is integrated into a web-based platform, is used in this paper to present a deep learning-based vitamin deficiency detection system. By examining pictures of physical characteristics like skin, eyes, lips, tongue, and nails, the system finds deficiencies in vitamins A, B-Complex, C, D, E, and K. The model attained a maximum validation accuracy of 82.05% after being trained on Kaggle datasets. The system has an AI assistant to respond to user inquiries about vitamin deficiencies, a comprehensive web interface, and database integration for user management. For early detection and preventive healthcare, this method offers a scalable, accessible, and non-invasive solution. Keywords: AI Assistant, Web Application, Deep Learning, EfficientNet-B4, Vitamin Deficiency Detection, Image Classification, and Healthcare AI