Search papers, labs, and topics across Lattice.
This paper introduces SCDistill, a framework designed to enhance semantic robustness in change detection for remote sensing images by employing a semantic-invariant self-distillation strategy. By leveraging perturbed yet semantically consistent data and a diffusion-based perturbation simulation pipeline, the model effectively learns to differentiate between semantic changes and non-semantic variations, significantly reducing false alarms. Extensive experiments reveal that SCDistill achieves state-of-the-art performance across various benchmarks and demonstrates strong generalization capabilities in related tasks such as binary change detection and change captioning.
False alarms in change detection can be drastically reduced by learning to distinguish semantic changes from non-semantic variations through innovative self-distillation techniques.
Change detection aims to identify semantic changes between remote sensing images. However, features from models are easily disturbed by non-semantic variations, such as illumination, shadows, and atmospheric changes, leading to false alarms and limited generalization in real-world scenarios. In this paper, we propose \textbf{SCDistill}, a framework for learning semantic-robust change detection via semantic-invariant self-distillation. First, to strengthen semantic consistency, we introduce a semantic-invariant self-distillation strategy that learns semantic robustness from perturbed yet semantically consistent data, empowering the change detector to extract disturbance-resistant features and achieve more reliable and accurate semantic change identification. Second, to expand paired data with non-semantic variations, we design a diffusion-based perturbation simulation pipeline that synthesizes complex environmental changes, enabling the model to explicitly learn to distinguish semantic changes from appearance-level fluctuations and reduce false alarms caused by non-semantic disturbances. These components promote robustness from data and representation perspectives, leading to synergistic performance gains. Extensive experiments demonstrate that SCDistill achieves state-of-the-art performance on multiple semantic change detection benchmarks and exhibits strong generalization to binary change detection and change captioning tasks. Code is accessible at https://github.com/elecreak/SCDistill.
