Published Paper Details:

Rapid urban growth, deforestation, and natural disasters are driving substantial land cover changes worldwide, posing significant challenges to sustainable development and environmental stewardship. Conventional change detection methods, such as image differencing and manual classification, often require intensive effort and deliver inconsistent accuracy under varying environmental conditions. GeoAlert is introduced as an AI-powered change monitoring system that automates the multi-temporal detection of land cover change using satellite imagery from high-resolution sensors, including Landsat-8 and Sentinel-2. The system preprocesses raw satellite data, aligns images, and applies advanced deep learning models, notably U-Net and Siamese CNN architectures, for per-pixel land cover classification and systematic change mapping. Changes identified are visualized via thematic maps and shared on a cloud-based dashboard, providing near-real-time alerts to stakeholders. In a comprehensive case study focusing on Hyderabad, India, GeoAlert achieved 89% classification accuracy, notably outperforming traditional image differencing approaches that scored 74%. These results highlight GeoAlert's potential as a scalable, robust decision-support tool for urban planners, environmental managers, and policy-makers in rapidly changing landscapes.