MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641077866 A) filed by Srm Institute Of Science And Technology, Ramapuram Campus; and Easwari Engineering College on June 24, 2026, for Unique Sign Language And Expression Based Communication Using Video Ai And Machine Learning.
Inventors include Dr. S. Sajini; Dr. S. Umarani; Dr. Rama Chaithanya Tanguturi; Dr. Sakthi Ganesh M; Dr. G. Deena; Dr. Sandhya; Mr. Birahadeeswaran; and Mr. Pravith Sriram.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: Abstract: Reading, writing, and speaking are essential to human communication. A person who has lost the ability to speak due to stroke, paralysis, or neurological disease must find alternate methods to communicate. Current devices and software have made converting thought into speech more feasible, but most require the user to speak specific utterances or train to a predefined set of symbols and interaction routines. For someone who has recently lost speech, acquiring new communication routines in a natural mode of expression can be arduous, and often impossible. The task becomes complicated further if there is concurrent hearing or vision impairment. This work aims to provide a communication system that learns from the user instead of forcing the user to learn the system. During normal operation, the system observes facial expressions, micro gestures, residual voice cues, and optional proprioceptive input in response to questions or provides of information to characterize a person’s natural communication behavior rather than his or her imposed communication routine. In order to provide a stable system, learning occurs only upon explicit correction from the user or caregiver. The correction factor is performed within a Hierarchically, Gated Correction, Modulated Personalized Intent Learning (HGCM, PIL) paradigm. Confidence measures and introspective uncertainty are calculated to determine when learning occurs, ensuring that only relevant person, specific Memory structures are affected. As the system adapts, a personalized interaction routine emerges that captures a person’s behavior. By avoiding undesired learning effects, we expect our approach to provide a robust and reliable alternative method for everyday practical communication.
Disclaimer: Curated by HT Syndication.
