LG-CD: Enhancing Language-Guided Change Detection through SAM2 Adaptation

2025-09-26

AI Summary

Existing remote sensing change detection methods predominantly rely on unimodal visual features, neglecting textual semantic guidance, which limits their accuracy and robustness. To address this, we propose LG-CD, a language-guided change detection framework. LG-CD is the first to incorporate SAM2 as a multi-scale visual backbone for this task and introduces two novel components: a Text Fusion Attention Module (TFAM) and a cross-attention-based Visual–Semantic Fusion Decoder (V-SFD), enabling effective cross-modal alignment and fine-grained change localization. Furthermore, multi-layer adapters facilitate efficient parameter-efficient fine-tuning. Extensive experiments demonstrate that LG-CD achieves state-of-the-art performance on LEVIR-CD, WHU-CD, and SYSU-CD benchmarks, significantly improving both detection accuracy and generalization capability. This work establishes a new paradigm for generic multimodal remote sensing change detection.

Technology Category

• Computer Vision: Language and Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Remote Sensing Change Detection (RSCD) typically identifies changes in land cover or surface conditions by analyzing multi-temporal images. Currently, most deep learning-based methods primarily focus on learning unimodal visual information, while neglecting the rich semantic information provided by multimodal data such as text. To address this limitation, we propose a novel Language-Guided Change Detection model (LG-CD). This model leverages natural language prompts to direct the network's attention to regions of interest, significantly improving the accuracy and robustness of change detection. Specifically, LG-CD utilizes a visual foundational model (SAM2) as a feature extractor to capture multi-scale pyramid features from high-resolution to low-resolution across bi-temporal remote sensing images. Subsequently, multi-layer adapters are employed to fine-tune the model for downstream tasks, ensuring its effectiveness in remote sensing change detection. Additionally, we design a Text Fusion Attention Module (TFAM) to align visual and textual information, enabling the model to focus on target change regions using text prompts. Finally, a Vision-Semantic Fusion Decoder (V-SFD) is implemented, which deeply integrates visual and semantic information through a cross-attention mechanism to produce highly accurate change detection masks. Our experiments on three datasets (LEVIR-CD, WHU-CD, and SYSU-CD) demonstrate that LG-CD consistently outperforms state-of-the-art change detection methods. Furthermore, our approach provides new insights into achieving generalized change detection by leveraging multimodal information.

Problem

• Enhancing change detection accuracy using language guidance and visual models
• Integrating text prompts to focus on specific change regions in imagery
• Aligning multimodal data through vision-semantic fusion for robust detection

Innovation

• Uses SAM2 as visual feature extractor
• Employs adapters for fine-tuning downstream tasks
• Integrates vision-text fusion via cross-attention mechanism