Rethinking Low-Confidence Pseudo Labels: Influence-Aware Semi-Supervised Fine-Tuning for Hyperspectral Change Detection

Hyperspectral image change detection (HSI-CD) suffers from severe annotation scarcity and complex change patterns, which fundamentally limit the effectiveness of directly fine-tuning pre-trained foundation models. Although semi-supervised learning provides a promising direction, existing approaches mainly rely on confidence-based pseudo-label selection, leading to limited data diversity or severe error propagation. In this paper, we propose Influence-Aware Semi-supervised Fine-tuning (IA-SFT), a novel framework that evaluates the influence of pseudo-labels on model decision behavior to identify truly valuable supervision signals. Instead of confidence-based selection, IA-SFT evaluates each low-confidence pseudo-label by measuring its impact on labeled data, enabling reliable filtering of high-value pseudo-labels with minimal noise. To further adapt foundation models to HSI-CD, we design an Adaptive Fusion Change Decoder (AFCD) that jointly models global semantic consistency and local change details. Extensive experiments on three benchmark datasets demonstrate that IA-SFT consistently improves pseudo-label quality and detection performance, achieving superior accuracy compared to state-of-the-art methods. Additional analyses validate the transferability of IA-SFT when integrated into different frameworks in a plug-and-play manner. Code will be released.

Accurate tracking of Earth’s surface dynamics via hyperspectral imagery is severely hindered by the high costs of manual data annotation. While existing semi-supervised models attempt to learn from unlabeled data, they typically rely on confidence thresholds, which either limit data diversity or introduce error accumulation. We developed IAF-CDNet, a novel framework that shifts the focus from how confident a prediction is to how much influence it has. By measuring how low-confidence labels alter the model's core decision boundaries, it selectively uncovers high-value signals while filtering out harmful noise, all paired with a specialized decoder designed to capture intricate change details. Tested across multiple real-world datasets, our approach dramatically improves change detection accuracy and outperforms state-of-the-art methods. This plug-and-play strategy provides a highly accurate, cost-effective tool for critical environmental monitoring.