International Journal of Scientific Research and Engineering Development

The ai edge nutrition scanner is an intelligent web-based application designed to automate food recognition and nutritional estimation from images. The system addresses the limitations of manual nutrition tracking, including the need to search food items, consult nutrition charts, and calculate portion-dependent values. The proposed solution accepts an uploaded image or live camera input and uses computer vision and deep learning to identify the food category. The report describes a transfer learning approach based on mobilenetv2 with the fruits-360 dataset, while the implemented application also documents yolo world and openclip components for food recognition. After recognition, the backend retrieves nutrition values from a structured nutrition.csv dataset and scales calories, protein, carbohydrates, fat, fibre, and sugar according to the user-entered weight. A documented banana test at 80 g achieved a 98% confidence score with calculated nutritional values, demonstrating the feasibility of the proposed workflow. The modular architecture is intended to support future multi-food detection, tensorflow lite deployment, mobile applications, portion estimation, barcode recognition, and personalized dietary recommendations.

Bhavya D, "AI Edge Nutrition Scanner" International Journal of Scientific Research and Engineering Development, V9(4): Page(1274-1278) July-August 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.