To that end, we introduce GarDA (Generative Appearance Replay for continual Domain Adaptation), a generative-replay based approach that can adapt a segmentation model sequentially to new domains with unlabeled
Then, a new paradigm of domain adaptation method-continual unsupervised domain adaptation is proposed to solve the issue of converting data between the source and target domains. Simultaneously, this paper integrates the detector and domain adaptation method to improve the model's detection accuracy ...
Domain Adaptation Unsupervised domain adaptation (UDA) [44,46] aims to improve the target model performance in the presence of a domain shift between the labeled source domain and unla- beled target domain. During training, UDA methods often align the feature distributions between the two domains...
Self- and unsupervised incremental learning 有一种方法,可以执行明确的任务分类并采用高斯混合模型拟合学习的表示。 Meta-learning 在解决相关任务时积累的信息用来学习新任务。 这种方法可以学习那些能减少未来梯度干扰并基于此改善传递的参数。 作者最后得到的一些结论: 比较无样本方法,LwF获得最佳结果。其他正则化方...
A review of single-source deep unsupervised visual domain adaptation. IEEE Trans. Neural Netw. Learn. Syst. https://10.1109/TNNLS.2020.3028503 (2020). Guan, H. & Liu, M. Domain adaptation for medical image analysis: a survey. Preprint at https://arxiv.org/abs/2102.09508 (2021). Zhuang,...
Thandiackal et al. [58] introduced a novel approach to unsupervised domain adaptation (UDA) for CL across multiple unlabeled target domains, without storing data from previously seen domains. The method uses generative feature-driven image replay combined with a dual-purpose discriminator, which aids...
论文关键词:Coarse-to-fine learning,Unsupervised domain adaptation,Semantic segmentation,Continual learning,Deep learning论文评审过程:Received 26 January 2022, Accepted 2 March 2022, Available online 8 March 2022, Version of Record 25 March 2022.
Li. Mutual mean-teaching: Pseudo label refinery for unsupervised domain adaptation on person re-identification. In International Conference on Learning Representations, 2020. 2 [14] Y. Ge, F. Zhu, D. Chen, R. Zhao, and H. Li. Self-paced con- trastive learning ...
In our work, we introduce a new, data-constrained DA paradigm where unlabeled target samples are received in batches and adaptation is performed continually. We propose a novel source-free method for continual unsupervised domain adaptation that utilizes a buffer for selective replay of previously ...
Paper tables with annotated results for Color Prompting for Data-Free Continual Unsupervised Domain Adaptive Person Re-Identification