MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202541112959 A) filed by International Institute Of Information Technology, Hyderabad on November 18, 2025, for Modular Training-Free Food Scanning System For Automated Zero-Shot Nutrition Estimation.
Inventors include C. V. Jawahar; Aditya Arun; Yash Arora; and Varun Bhargavan.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: The present invention discloses a modular, training-free food scanning system (100) for automated zero-shot nutrition estimation of multi-dish thalis. The system (100) is an AI-driven modular platform designed to automatically determine the per-dish and overall nutritional composition of a meal from a single captured image acquired via an image capture unit (102) optimized for full-plate coverage. The system architecture integrates multiple functional modules configured to perform visual segmentation, dish classification, weight estimation, and nutrient computation sequentially. It employs zero-shot segmentation to isolate individual food regions, embedding-based prototype matching for identifying dishes, and a regression-based model for estimating portion weights. Nutrient values are computed using predefined nutrition lookup tables corresponding to each identified dish. The modular, training-free design enables rapid adaptation to new or unseen dishes, ensures real-time operation, and supports scalable deployment across diverse cuisines and dining environments without retraining or manual calibration. FIG. 1
Disclaimer: Curated by HT Syndication.
