Exploring Generalizable Remote Sensing Change Detection via Low-Rank Exchange Adaptation of Vision Foundation Model

Change detection, as an important and widely applied technique in the field of remote sensing, aims to analyze changes in surface areas over time and has broad applications in areas such as environmental monitoring, urban development, and land use analysis. Due to the complexity of multi-source data in remote sensing images, such as sensor modality differences, noise, registration errors, and complex terrain, accurate change detection faces numerous challenges. In recent years, deep learning, especially the development of foundation models, has provided more powerful solutions for feature extraction and data fusion, effectively addressing these complexities. This paper systematically reviews the latest advancements in the field of change detection, with a focus on the application of foundation models in remote sensing tasks. First, we introduce the basic knowledge of change detection tasks, including task definitions, Transformer models, self-attention mechanisms, multimodal learning, foundation models, and vision-language models. Next, we provide a detailed classification of existing methods based on data modalities (single-modal and multi-modal) and network structures (encoder, decoder, encoder-decoder). This review provides readers with a comprehensive understanding of the field and summarizes the advantages and limitations of various methods. Finally, we summarize the performance of models on several key benchmark datasets, including newly proposed large-scale benchmarks, and compare them with other change detection models, offering an in-depth analysis of the role and limitations of foundation models in change detection. Based on this, we propose future research directions for remote sensing change detection. This paper provides a systematic review of the application of foundation models in the field of remote sensing change detection and offers insights for further exploration in this area.