MUMBAI, India, Sept. 22 -- Intellectual Property India has published a patent application (202641109804 A) filed by Srm Institute Of Science And Technology, Ramapuram Campus; and Easwari Engineering College on September 13, 2026, for Multi-Scale Spatio-Temporal Deep Learning Model For Remote Sensing Change Detection.
Inventors include V. Gowri; T. Avinash; V. Tej Manendra; B. Nikhil Srinivas; and Dr. P. C. Sarah Prithvika.
The application for the patent was published on September 18, 2026, under issue no. 38/2026.
Abstract: The present invention describes a deep learning based multi-scale spatio-temporal framework for accurate change detection in remote sensing imagery. The system analyzes bi-temporal satellite images captured over the same geographic region at different time intervals to identify structural and environmental changes. The proposed model integrates convolutional neural networks and attention-based feature learning to extract hierarchical spatial and temporal features from high-resolution satellite images. The framework introduces a multi-scale feature interaction mechanism that dynamically captures objects of varying sizes while suppressing pseudo- changes caused by illumination variations, seasonal changes, and atmospheric conditions. The architecture consists of a shared feature extraction backbone, multi-scale attention modules, differential feature learning components, and a structural decoder that generates a binary change detection map. The proposed system improves detection accuracy while reducing computational complexity and false-positive detections. The invention is applicable to smart city monitoring, disaster damage assessment, environmental monitoring, urban development analysis, and land-use change detection using satellite imagery.
Disclaimer: Curated by HT Syndication.
