Low Saturation Confidence Distribution-based Test-Time Adaptation for Cross-Domain Remote Sensing Image Classification

2024-08-29

International Journal of Applied Earth Observation and Geoinformation

AI Summary

Cross-domain remote sensing image classification faces the challenge of inaccessible source and target training data. Method: This paper proposes the first test-time adaptation (TTA) framework specifically designed for remote sensing imagery, enabling model optimization solely from a single target test sample during inference—without requiring source data, target training sets, or pre-collection procedures. It introduces a novel low-saturation confidence distribution modeling mechanism, jointly optimizing weak-confidence softmax entropy loss, balanced-class softmax entropy loss, and low-saturation distribution loss, with dynamic weighting via soft log-likelihood ratios. Results: The method achieves significant accuracy improvements across multiple cross-domain remote sensing benchmarks, demonstrating rapid single-sample/step adaptation while ensuring real-time efficiency, robustness, and class-balanced prediction.

Technology Category

• Machine Learning: Transfer, Domain Adaptation, Multi-Task Learning
• Computer Vision: Learning & Optimization for CV
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Abstract

Unsupervised Domain Adaptation (UDA) has emerged as a powerful technique for addressing the distribution shift across various Remote Sensing (RS) applications. However, most UDA approaches require access to source data, which may be infeasible due to data privacy or transmission constraints. Source-free Domain Adaptation addresses the absence of source data but usually demands a large amount of target domain data beforehand, hindering rapid adaptation and restricting their applicability in broader scenarios. In practical cross-domain RS image classification, achieving a balance between adaptation speed and accuracy is crucial. Therefore, we propose Low Saturation Confidence Distribution Test-Time Adaptation (LSCD-TTA), marketing the first attempt to explore Test-Time Adaptation for cross-domain RS image classification without requiring source or target training data. LSCD-TTA adapts a source-trained model on the fly using only the target test data encountered during inference, enabling immediate and efficient adaptation while maintaining high accuracy. Specifically, LSCD-TTA incorporates three optimization strategies tailored to the distribution characteristics of RS images. Firstly, weak-confidence softmax-entropy loss emphasizes categories that are more difficult to classify to address unbalanced class distribution. Secondly, balanced-categories softmax-entropy loss softens and balances the predicted probabilities to tackle the category diversity. Finally, low saturation distribution loss utilizes soft log-likelihood ratios to reduce the impact of low-confidence samples in the later stages of adaptation. By effectively combining these losses, LSCD-TTA enables rapid and accurate adaptation to the target domain for RS image classification.

Problem

• Adapts RS image classification without source data
• Balances speed and accuracy in domain adaptation
• Handles unbalanced class and category diversity

Innovation

• Test-Time Adaptation without source or target data
• Weak-confidence softmax-entropy loss for unbalanced classes
• Low saturation distribution loss for high-confidence adaptation