期刊
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR)
卷 -, 期 -, 页码 7191-7201出版社
IEEE COMPUTER SOC
DOI: 10.1109/CVPR52688.2022.00706
关键词
-
资金
- Swiss National Science Foundation [PP00P2 176878]
- Toyota Motor Europe
Test-time domain adaptation aims to adapt a source pretrained model to a continually changing target domain. We propose a continual test-time adaptation approach (CoTTA) to address distribution shift and catastrophic forgetting by reducing error accumulation and preserving source knowledge. CoTTA is easy to implement and outperforms existing methods in experiments.
Test-time domain adaptation aims to adapt a source pretrained model to a target domain without using any source data. Existing works mainly consider the case where the target domain is static. However, real-world machine perception systems are running in non-stationary and continually changing environments where the target domain distribution can change over time. Existing methods, which are mostly based on self-training and entropy regularization, can suffer from these non-stationary environments. Due to the distribution shift over time in the target domain, pseudo-labels become unreliable. The noisy pseudolabels can further lead to error accumulation and catastrophic forgetting. To tackle these issues, we propose a continual test-time adaptation approach (CoTTA) which comprises two parts. Firstly, we propose to reduce the error accumulation by using weight-averaged and augmentation-averaged predictions which are often more accurate. On the other hand, to avoid catastrophic forgetting, we propose to stochastically restore a small part of the neurons to the source pre-trained weights during each iteration to help preserve source knowledge in the long-term. The proposed method enables the long-term adaptation for all parameters in the network. CoTTA is easy to implement and can be readily incorporated in off-the-shelf pre-trained models. We demonstrate the effectiveness of our approach on four classification tasks and a segmentation task for continual testtime adaptation, on which we outperform existing methods. Our code is available at https://qin.ee/cotta.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据