AdaSemiCD: An Adaptive Semi-Supervised Change Detection Method Based on Pseudo-Label Evaluation

2024-11-12

IEEE Transactions on Geoscience and Remote Sensing

AI Summary

Remote sensing change detection suffers from scarce labeled data and severe class imbalance, leading to unreliable pseudo-labels. To address this, we propose an adaptive semi-supervised learning framework. Our method introduces three key innovations: (1) a novel information-entropy-based pseudo-label quality metric that jointly incorporates class rebalancing and confusion-region enhancement; (2) AdaFusion, a dynamic mechanism that replaces uncertain regions with reliable pseudo-label-guided feature fusion; and (3) AdaEMA, an exponential moving average strategy that updates the teacher model exclusively with high-confidence samples to improve training stability. By integrating pseudo-label quality assessment, adaptive feature fusion, and consistency regularization, our framework achieves state-of-the-art performance on LEVIR-CD, WHU-CD, and CDD benchmarks—demonstrating both superior accuracy and strong cross-dataset generalization capability.

Technology Category

• Machine Learning: Semi-Supervised Learning
• Computer Vision: Adversarial Attacks & Robustness
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

Change Detection (CD) is an essential field in remote sensing, with a primary focus on identifying areas of change in bi-temporal image pairs captured at varying intervals of the same region by a satellite. The data annotation process for the CD task is both time-consuming and labor-intensive. To make better use of the scarce labeled data and abundant unlabeled data, we present an adaptive dynamic semi-supervised learning method, AdaSemiCD, to improve the use of pseudo-labels and optimize the training process. Initially, due to the extreme class imbalance inherent in CD, the model is more inclined to focus on the background class, and it is easy to confuse the boundary of the target object. Considering these two points, we develop a measurable evaluation metric for pseudo-labels that enhances the representation of information entropy by class rebalancing and amplification of confusing areas to give a larger weight to prospects change objects. Subsequently, to enhance the reliability of sample-wise pseudo-labels, we introduce the AdaFusion module, which is capable of dynamically identifying the most uncertain region and substituting it with more trustworthy content. Lastly, to ensure better training stability, we introduce the AdaEMA module, which updates the teacher model using only batches of trusted samples. Experimental results from LEVIR-CD, WHU-CD, and CDD datasets validate the efficacy and universality of our proposed adaptive training framework.

Problem

• Improves change detection in remote sensing using semi-supervised learning.
• Addresses class imbalance and boundary confusion in change detection tasks.
• Enhances pseudo-label reliability and training stability with adaptive modules.

Innovation

• Adaptive dynamic semi-supervised learning method
• AdaFusion module for uncertain region substitution
• AdaEMA module for stable teacher model updates