Semantic-Driven Hierarchical Fusion With Adaptive Reasoning for Weak-Information Target Change Detection in Remote Sensing Images

The semantic change detection (SCD) plays a critical role in the intelligent information extraction from remote sensing (RS) images. However, its performance in detecting weak-information targets, such as small-sized ground objects, is still constrained by the semantic representation ambiguity, information loss, and complex background interference. This article proposes a semantic guidance inference network (SGI-Net) that establishes a collaborative optimization mechanism through the semantic-driven feature coupling, adaptive reasoning, and attention guidance to overcome these challenges. Specifically, we design a hierarchical feature coupling module (HFCM). This module enhances the geometric detail representation for weak-information targets through the cross-layer feature aggregation and dual-path decoupling. It generates discriminative semantic priors, establishing geometric-semantic correlations across multilevel features. Furthermore, a multidimensional adaptive inference module (MAIM) is developed to achieve the adaptive feature aggregation of local–global patterns via nonoverlapping patch partitioning, preserving spatial continuity of weak-information targets through the multiscale feature fusion. In addition, the change-guided attention (CGA) module dynamically adjusts the channelwise and spatial attention weights based on the prior knowledge, amplifying cross-hierarchical responses to weak change signals. Experimental results on the SECOND and SJH_SCD datasets demonstrate that SGI-Net achieves values of 73.86% and 77.43%, respectively, surpassing state-of-the-art (SOTA) methods by 0.73%–1.37% in mIoU and 1.26%–1.77% in separated Kappa coefficient (SeK). The visualization analysis further confirms that SGI-Net effectively enhances the feature responses of weak-information targets in complex object scenes, improving the discrimination accuracy of fine-grained change boundaries. These results provide the reliable technical support for high-precision RS change detection (CD).