MUMBAI, India, Aug. 10 -- Intellectual Property India has published a patent application (202641091716 A) filed by Koneru Lakshmaiah Education Foundation on July 29, 2026, for System And Method For Pre-Ignition Wildfire Intelligence Using Multi-Modal Eco-Climatic Representation Learning.
Inventors include Koneru Lakshmaiah Education Foundation; K. Praveen Kumar; and Radhika Rani Chintala.
The application for the patent was published on July 31, 2026, under issue no. 31/2026.
Abstract: A computer-implemented pre-ignition wildfire intelligence system and method to evaluate the susceptibility of wildfires before they are ignited is disclosed. It is a multi-source environmental data acquisition module, designed to receive heterogeneous eco-climatic observations from one or more remote sensing data sources, such as multi-spectral satellite images, thermal observations, topographical information, vegetation indices, land surface temperature, and past fire histories. A pre-processing module is designed to pre-process the acquired environmental observations using radiometric correction, spatial registration, temporal alignment, geo-referencing, normalization, resampling and removal of invalid data to produce standardized multi-modal eco-climatic data. A multi-modal environmental feature extraction module is designed to extract multi-modal environmental features for both cross-modal feature fusion to generate a Unified Eco-Climatic Representation (UER) and to optimise the generated representation using contrastive self-supervised learning for spatio-temporal forecasting and for estimation of the susceptibility to wildfires before ignition. The AI analytics module also generates wildfire susceptibility scores, wildfire probability maps, geospatial wildfire risk maps, and early-warning notifications for disaster management and decision-support applications. The disclosed invention provides an integrated end-to-end framework for wildfire intelligence that combines multi-source environmental data acquisition, AI-based representation learning, contrastive self-supervised optimization, and spatio-temporal forecasting within a single computational architecture to enhance pre-ignition wildfire susceptibility assessment accuracy, robustness, and reliability.
Disclaimer: Curated by HT Syndication.
