MUMBAI, India, June 30 -- Intellectual Property India has published a patent application (202641074986 A) filed by Sr University on June 17, 2026, for Lightweight Ann-Based Human Activity Recognition System Optimized For Iot Edge Devices.
Inventors include Nagavelli Yogender Nath; and Dr. J. Bhavana.
The application for the patent was published on June 26, 2026, under issue no. 26/2026.
Abstract: LIGHTWEIGHT ANN-BASED HUMAN ACTIVITY RECOGNITION SYSTEM OPTIMIZED FOR IOT EDGE DEVICES The present invention discloses a lightweight artificial neural network (ANN)-based human activity recognition (HAR) system optimized for deployment on IoT edge devices such as microcontrollers, wearables, and embedded boards. The system comprises a multi-sensor data acquisition module, a feature-adaptive scaling layer for normalization, compression, and noise reduction of heterogeneous sensor inputs, an optimized ANN architecture with reduced parameters, and an edge inference engine enabling real-time, on-device classification of human activities without reliance on cloud computing. By eliminating cloud dependency, the invention reduces latency, enhances privacy, and achieves energy-efficient continuous operation on battery-powered devices. The proposed ANN model achieves 99% accuracy with fewer parameters compared to conventional CNN/RNN models, while maintaining robustness against noisy sensor data. This invention is applicable to healthcare monitoring, smart homes, industrial safety, fitness tracking, and elder-care systems, offering a cost-effective, privacy-preserving, and sustainable solution for intelligent IoT-enabled environments.
Disclaimer: Curated by HT Syndication.
