MUMBAI, India, Aug. 31 -- Intellectual Property India has published a patent application (202641101012 A) filed by Easwari Engineering College; and Srm Institute Of Science And Technology, Ramapuram Campus on August 21, 2026, for Offline Multi-Agent Framework For Context-Aware Multilingual Signboard Understanding And Intelligent Assistance.
Inventors include Aparna S; Mirjana R; and Dr. G. Jeyaram.
The application for the patent was published on August 28, 2026, under issue no. 35/2026.
Abstract: Offline Multi-Agent Framework for Context-Aware Multilingual Signboard Understanding and Intelligent Assistance is designed to enable intelligent interpretation of multilingual signboards, particularly in environments with limited or no internet connectivity. The framework captures signboard images and performs multilingual text detection, OCR, language identification, translation and/or transliteration, followed by semantic interpretation of the recognized content. Multiple specialized agents work collaboratively to identify entities, points of interest, directions, warnings, public services, and other contextual information represented by the signboard, transforming conventional text recognition into meaningful understanding of the signboard and its purpose. The key novelty lies in integrating multi-agent semantic understanding with offline knowledge retrieval to provide context-aware assistance without dependence on cloud services or continuous internet connectivity. The system retrieves relevant information from a locally stored knowledge base and generates contextual explanations, translations/transliterations, and voice-based assistance in the user's selected language. By unifying multilingual scene-text recognition, semantic and contextual understanding, offline knowledge retrieval, multi-agent collaboration, and intelligent voice assistance, the framework goes beyond existing systems that primarily perform OCR, translation, or text-to-speech, providing an accessible and intelligent solution for tourists, migrants, regional-language users, and visually impaired users. Total Number of Words: 186 words
Disclaimer: Curated by HT Syndication.
