MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641109301 A) filed by Amrita Vishwa Vidyapeetham on September 11, 2026, for Rice Yield Prediction System.
Inventors include Khanna, Manju; and Gantala Reddy, Shalini.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: Embodiments of the present disclosure relate to a system (100) and a method (700) of rice yield prediction. The system (100) performs rice-yield prediction by receiving meteorological time-series data, soil-nutrient time-series data, and historical yield data through a data interface (106). A processor (102-2) of the system (100) preprocesses the received data to obtain standardized data representations and supplies the standardized representations to a temporal feature extraction unit (110) that extracts local temporal patterns and long-range temporal dependencies using one-dimensional convolutional and transformer processing. A multi-head cross-attention fusion unit (112) determines interdependencies among the extracted meteorological, soil-nutrient, and historical-yield features and generates a unified feature representation. The processor (102-2) processes the unified feature representation using an adaptively optimized residual spatio-temporal prediction network to generate a numerical rice-yield prediction. An input/output interface (130) communicates the generated prediction to an external device or agricultural management platform.
Disclaimer: Curated by HT Syndication.
