Distribution-Aware Continual Test-Time Adaptation for Semantic Segmentation

TL;DR

This paper introduces a distribution-aware tuning method for continual test-time adaptation in semantic segmentation, effectively addressing error accumulation and catastrophic forgetting by selectively updating small parameter groups based on data distribution shifts.

Contribution

The proposed DAT method adaptively updates domain-specific and task-relevant parameters, improving efficiency and robustness in long-term continual test-time adaptation for semantic segmentation.

Findings

Outperforms previous CTTA methods on benchmark datasets.

Reduces error accumulation caused by noisy pseudo-labels.

Mitigates catastrophic forgetting through selective parameter updates.

Abstract

Since autonomous driving systems usually face dynamic and ever-changing environments, continual test-time adaptation (CTTA) has been proposed as a strategy for transferring deployed models to continually changing target domains. However, the pursuit of long-term adaptation often introduces catastrophic forgetting and error accumulation problems, which impede the practical implementation of CTTA in the real world. Recently, existing CTTA methods mainly focus on utilizing a majority of parameters to fit target domain knowledge through self-training. Unfortunately, these approaches often amplify the challenge of error accumulation due to noisy pseudo-labels, and pose practical limitations stemming from the heavy computational costs associated with entire model updates. In this paper, we propose a distribution-aware tuning (DAT) method to make the semantic segmentation CTTA efficient and practical in real-world applications. DAT adaptively selects and updates two small groups of trainable parameters based on data distribution during the continual adaptation process, including domain-specific parameters (DSP) and task-relevant parameters (TRP). Specifically, DSP exhibits sensitivity to outputs with substantial distribution shifts, effectively mitigating the problem of error accumulation. In contrast, TRP are allocated to positions that are responsive to outputs with minor distribution shifts, which are fine-tuned to avoid the catastrophic forgetting problem. In addition, since CTTA is a temporal task, we introduce the Parameter Accumulation Update (PAU) strategy to collect the updated DSP and TRP in target domain sequences. We conduct extensive experiments on two widely-used semantic segmentation CTTA benchmarks, achieving promising performance compared to previous state-of-the-art methods.