Remote Sensing Change Detection

Remote sensing change detection compares images of the same place across time to locate changes such as new buildings, disaster damage, forest loss, or shifting land cover. Methods range from paired convolutional networks and Transformers to state-space models, foundation-model features, and optical–SAR fusion; newer systems also classify change types or describe them in language. Research explores supervised learning alongside unsupervised, weakly supervised, and open-vocabulary approaches that reduce reliance on pixel-level labels. A central challenge is distinguishing real change from misregistration, seasonal or sensor differences, while ensuring models transfer across locations and remain efficient in practical monitoring settings.