Deep Learning – Based Change Detection For Identifying Illegal Encroachments On Water Bodies

The fast pace of urbanization has resulted in extensive exploitation of natural resources, especially water resources, leading to environmental pollution, flooding hazards, and reduced biological diversity.The traditional monitoring processes that involve field visits, observations, and surveys require lots of efforts, can lead to errors, and are not suitable for wide-range monitoring.In order to overcome the limitations, the current project utilizes satellite imaging and machine learning algorithms to enable automatic identification of any unapproved constructions in water bodies.The presented algorithm uses the lightweight semantic segmentation U-Net neural network model to process satellite images captured by Google Earth Pro from 2010 to 2024.After undergoing the preprocessing stage consisting of image resizing, normalization, and enhancement, the data is used to train the model that detects any newly added construction objects within a specified water body.The approach offers a practical solution for detecting and monitoring any new structures developed in or around water bodies.