BP-TTA: Balanced and Prototype-Guided Test-Time Adaptation in Dynamic Scenarios

2026-06-30

AI Summary

This work addresses the performance degradation of existing test-time adaptation methods under dynamic testing scenarios, where continuous domain shift and class imbalance are intricately coupled. To tackle this dual challenge, we propose a novel approach that integrates batch-balanced sampling with a prototype-guided mechanism. Our method constructs class-balanced adaptation batches on the fly by replaying cached high-confidence historical samples and dynamically maintaining evolving class prototypes. Model updates are constrained through prototype similarity, enhancing the reliability of pseudo-labels and stabilizing online learning. As the first method to jointly address these intertwined challenges, our approach significantly outperforms state-of-the-art alternatives under dynamic test-stream settings, demonstrating superior robustness and adaptability.

Abstract

Test-Time Adaptation (TTA) enables models trained on a source domain to adapt online to unlabeled test data under distribution shifts. While recent TTA methods have moved beyond static settings and begun to consider continual domain shifts, they primarily address distribution drift and fail to account for class imbalance in dynamic scenarios. In real-world test-time streams, class imbalance and continual domain shifts often occur at the same time and interact with each other. In this paper, we propose a novel Balanced and Prototype-Guided Test-Time Adaptation (BP-TTA) method, which combines batch-balanced sampling with prototype-guided adaptation to handle the class imbalance and continual domain shift problems. BP-TTA constructs balanced adaptation batches by integrating current samples with high-confidence historical instances, effectively mitigating bias toward dominant classes and stabilizing online updates. Meanwhile, BP-TTA maintains evolving class prototypes during inference and leverages prototype similarity as a constraint for model adaptation, thereby improving the reliability of pseudo-labels and enhancing the stability of online updates under persistent domain shifts. Extensive experiments demonstrate that BP-TTA consistently outperforms state-of-the-art TTA methods in dynamic test-time streaming settings.

Problem

• Test-Time Adaptation
• Class Imbalance
• Continual Domain Shift
• Dynamic Scenarios

Innovation

• Test-Time Adaptation
• Class Imbalance
• Continual Domain Shift
• Prototype-Guided Learning
• Balanced Sampling