This paper introduces ReservoirTTA, a novel plug-in framework designed for prolonged test-time adaptation (TTA) in scenarios where the test domain continuously shifts over time, including cases where domains recur or evolve gradually. At its core, ReservoirTTA maintains a reservoir of domain-specialized models -- an adaptive test-time model ensemble -- that both detects new domains via online clustering over style features of incoming samples and routes each sample to the appropriate specialized model, and thereby enables domain-specific adaptation. This multi-model strategy overcomes key limitations of single model adaptation, such as catastrophic forgetting, inter-domain interference, and error accumulation, ensuring robust and stable performance on sustained non-stationary test distributions. Our theoretical analysis reveals key components that bound parameter variance and prevent model collapse, while our plug-in TTA module mitigates catastrophic forgetting of previously encountered domains. Extensive experiments on the classification corruption benchmarks, including ImageNet-C and CIFAR-10/100-C, as well as the CityscapesACDC semantic segmentation task, covering recurring and continuously evolving domain shifts, demonstrate that ReservoirTTA significantly improves adaptation accuracy and maintains stable performance across prolonged, recurring shifts, outperforming state-of-the-art methods.
View on arXiv@article{vray2025_2505.14511,
title={ ReservoirTTA: Prolonged Test-time Adaptation for Evolving and Recurring Domains },
author={ Guillaume Vray and Devavrat Tomar and Xufeng Gao and Jean-Philippe Thiran and Evan Shelhamer and Behzad Bozorgtabar },
journal={arXiv preprint arXiv:2505.14511},
year={ 2025 }
}