Search papers, labs, and topics across Lattice.
This paper introduces BP-TTA, a novel method for Test-Time Adaptation (TTA) that addresses both class imbalance and continual domain shifts in dynamic scenarios. By integrating batch-balanced sampling with prototype-guided adaptation, BP-TTA constructs balanced adaptation batches and maintains evolving class prototypes, which enhances the reliability of pseudo-labels and stabilizes online updates. Experimental results show that BP-TTA significantly outperforms existing TTA methods, indicating its effectiveness in real-world applications where both class imbalance and distribution shifts occur simultaneously.
Class imbalance and continual domain shifts can be tackled simultaneously with BP-TTA, leading to more reliable model adaptations in dynamic environments.
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.
