• Mr. Rony Preetam Prakash
• Ayesha Sabahath
• Mahek Sultana
• Md. Zubair
• Md. Shoeb Musharraf

Abstract : Vitamin deficiencies affect millions worldwide, especially in rural and low-resource regions where blood tests are costly and inaccessible. This paper presents VDP (Vitamin Deficiency Predictor), a mobile and web-based non-invasive screening system that uses smartphone-captured images of eyes, tongue, nails, and skin to detect deficiencies of Vitamin A, B12, C, D, and Iron using Convolutional Neural Networks (CNNs). The system employs transfer learning with MobileNetV2 backbone, multi-modal input fusion, and on-device inference via TensorFlow Lite. Tested on a dataset of 812 Indian-origin images, VDP achieved 81.3% overall accuracy with real-time prediction (<2 seconds) on mid-range smartphones. The system also provides personalized dietary recommendations and explainable results using Grad-CAM++. .

Vitamin Deficiency, Deep Learning, Convolutional Neural Networks, Transfer Learning, MobileNetV2, Non-invasive Diagnosis, mHealth, TensorFlow Lite, Image Processing, Nutritional Screening.

"VITAMIN DEFICIENCY PREDICTION SYSTEM (VDP)", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.13, Issue 2, page no.c622-c627, February-2026, Available :http://www.jetir.org/papers/JETIR2602294.pdf

"VITAMIN DEFICIENCY PREDICTION SYSTEM (VDP)", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.13, Issue 2, page no. ppc622-c627, February-2026, Available at : http://www.jetir.org/papers/JETIR2602294.pdf