Supervised Change Detection

Supervised change detection aims to identify pixel-level differences between pairs of images (e.g., satellite imagery over time) to track changes in the environment. Recent research heavily emphasizes reducing the reliance on extensive manual labeling by exploring semi-supervised and weakly-supervised approaches, employing techniques like consistency regularization, self-supervised contrastive learning, and visual-language model guidance within convolutional neural networks and transformers. These advancements are crucial for handling the large datasets common in remote sensing and other applications where fully labeled data is scarce and expensive to obtain, enabling more efficient and scalable change detection analyses.