MUMBAI, India, Feb. 13 -- Intellectual Property India has published a patent application (202641007427 A) filed by R. M. K. Engineering College; Sri Sairam College Of Engineering,; Panimalar Engineering College; Karpagam College Of Engineering; Kathir College Of Engineering; and M. Kumarasamy College Of Engineering, Chennai, Tamil Nadu, on Jan. 26, for 'ai and machine learning based human action recognition system using mobile sensor fusion.'
Inventor(s) include Mr M. Vengateshwaran; Ms N. Valarmathi; Ms R. Pramila; Ms L. Vidhya; Ms R. Mythreagi; and Ms S. Rajaambika.
The application for the patent was published on Feb. 13, under issue no. 07/2026.
According to the abstract released by the Intellectual Property India: "The present invention relates to an AI and Machine Learning based Human Action Recognition System utilizing mobile sensor fusion on intelligent devices. The system leverages multiple inertial sensors embedded in smartphones to accurately identify and classify human activities through a client-server architecture supported by network connectivity. To address inaccuracies caused by arbitrary device orientation and sensor axis misalignment, a robust multi-sensor fusion and attitude correction algorithm is employed to ensure reliable data acquisition and continuity. Meaningful features are extracted from multimodal sensor streams and classified using advanced machine learning techniques, including multi-kernel support vector machines, to enhance recognition accuracy and adaptability. The system continuously monitors sensor orientation, assigns and updates a dominant axis, and detects periodic human motions by analyzing acceleration patterns. Captured sensor data are transformed into structured sequences corresponding to predefined activities, enabling efficient real-time and offline recognition. The proposed system demonstrates high generality, robustness, and effectiveness, making it suitable for healthcare monitoring, activity tracking, and intelligent human-computer interaction applications."
Disclaimer: Curated by HT Syndication.
